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 1.118 by jsr166, Sat Jan 28 04:34:54 2012 UTC vs.
Revision 1.174 by jsr166, Sun Jan 13 21:56:12 2013 UTC

#	Line 11 | Line 11 | import java.util.Arrays;
11		import java.util.Collection;
12		import java.util.Collections;
13		import java.util.List;
14	–	import java.util.Random;
14		import java.util.concurrent.AbstractExecutorService;
15		import java.util.concurrent.Callable;
16		import java.util.concurrent.ExecutorService;
#	Line 19 | Line 18 | import java.util.concurrent.Future;
18		import java.util.concurrent.RejectedExecutionException;
19		import java.util.concurrent.RunnableFuture;
20		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;
25	–	import java.util.concurrent.locks.Condition;
21
22		/**
23		* An {@link ExecutorService} for running {@link ForkJoinTask}s.
#	Line 42 | Line 37 | import java.util.concurrent.locks.Condit
37		* ForkJoinPool}s may also be appropriate for use with event-style
38		* tasks that are never joined.
39		*
40	<	* <p>A {@code ForkJoinPool} is constructed with a given target
41	<	* parallelism level; by default, equal to the number of available
42	<	* processors. The pool attempts to maintain enough active (or
43	<	* available) threads by dynamically adding, suspending, or resuming
44	<	* internal worker threads, even if some tasks are stalled waiting to
45	<	* join others. However, no such adjustments are guaranteed in the
46	<	* face of blocked IO or other unmanaged synchronization. The nested
47	<	* {@link ManagedBlocker} interface enables extension of the kinds of
40	>	* <p>A static {@link #commonPool()} is available and appropriate for
41	>	* most applications. The common pool is used by any ForkJoinTask that
42	>	* is not explicitly submitted to a specified pool. Using the common
43	>	* pool normally reduces resource usage (its threads are slowly
44	>	* reclaimed during periods of non-use, and reinstated upon subsequent
45	>	* use).
46	>	*
47	>	* <p>For applications that require separate or custom pools, a {@code
48	>	* ForkJoinPool} may be constructed with a given target parallelism
49	>	* level; by default, equal to the number of available processors. The
50	>	* pool attempts to maintain enough active (or available) threads by
51	>	* dynamically adding, suspending, or resuming internal worker
52	>	* threads, even if some tasks are stalled waiting to join
53	>	* others. However, no such adjustments are guaranteed in the face of
54	>	* blocked I/O or other unmanaged synchronization. The nested {@link
55	>	* ManagedBlocker} interface enables extension of the kinds of
56		* synchronization accommodated.
57		*
58		* <p>In addition to execution and lifecycle control methods, this
#	Line 59 | Line 62 | import java.util.concurrent.locks.Condit
62		* {@link #toString} returns indications of pool state in a
63		* convenient form for informal monitoring.
64		*
65	<	* <p> As is the case with other ExecutorServices, there are three
65	>	* <p>As is the case with other ExecutorServices, there are three
66		* main task execution methods summarized in the following table.
67		* These are designed to be used primarily by clients not already
68		* engaged in fork/join computations in the current pool.  The main
#	Line 94 | Line 97 | import java.util.concurrent.locks.Condit
97		*  </tr>
98		* </table>
99		*
100	<	* <p><b>Sample Usage.</b> Normally a single {@code ForkJoinPool} is
101	<	* used for all parallel task execution in a program or subsystem.
102	<	* Otherwise, use would not usually outweigh the construction and
103	<	* bookkeeping overhead of creating a large set of threads. For
104	<	* example, a common pool could be used for the {@code SortTasks}
105	<	* illustrated in {@link RecursiveAction}. Because {@code
106	<	* ForkJoinPool} uses threads in {@linkplain java.lang.Thread#isDaemon
107	<	* daemon} mode, there is typically no need to explicitly {@link
108	<	* #shutdown} such a pool upon program exit.
109	<	*
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>
100	>	* <p>The common pool is by default constructed with default
101	>	* parameters, but these may be controlled by setting three {@link
102	>	* System#getProperty system properties} with prefix {@code
103	>	* java.util.concurrent.ForkJoinPool.common}: {@code parallelism} --
104	>	* an integer greater than zero, {@code threadFactory} -- the class
105	>	* name of a {@link ForkJoinWorkerThreadFactory}, and {@code
106	>	* exceptionHandler} -- the class name of a {@link
107	>	* java.lang.Thread.UncaughtExceptionHandler
108	>	* Thread.UncaughtExceptionHandler}. Upon any error in establishing
109	>	* these settings, default parameters are used.
110		*
111		* <p><b>Implementation notes</b>: This implementation restricts the
112		* maximum number of running threads to 32767. Attempts to create
#	Line 177 | Line 174 | public class ForkJoinPool extends Abstra
174		* If an attempted steal fails, a thief always chooses a different
175		* random victim target to try next. So, in order for one thief to
176		* progress, it suffices for any in-progress poll or new push on
177	<	* any empty queue to complete.
177	>	* any empty queue to complete. (This is why we normally use
178	>	* method pollAt and its variants that try once at the apparent
179	>	* base index, else consider alternative actions, rather than
180	>	* method poll.)
181		*
182		* This approach also enables support of a user mode in which local
183		* task processing is in FIFO, not LIFO order, simply by using
#	Line 194 | Line 194 | public class ForkJoinPool extends Abstra
194		* WorkQueues are also used in a similar way for tasks submitted
195		* to the pool. We cannot mix these tasks in the same queues used
196		* for work-stealing (this would contaminate lifo/fifo
197	<	* processing). Instead, we loosely associate submission queues
197	>	* processing). Instead, we randomly associate submission queues
198		* with submitting threads, using a form of hashing.  The
199		* ThreadLocal Submitter class contains a value initially used as
200		* a hash code for choosing existing queues, but may be randomly
201		* repositioned upon contention with other submitters.  In
202	<	* essence, submitters act like workers except that they never
203	<	* take tasks, and they are multiplexed on to a finite number of
204	<	* shared work queues. However, classes are set up so that future
205	<	* extensions could allow submitters to optionally help perform
206	<	* tasks as well. Pool submissions from internal workers are also
207	<	* allowed, but use randomized rather than thread-hashed queue
208	<	* indices to avoid imbalance.  Insertion of tasks in shared mode
209	<	* requires a lock (mainly to protect in the case of resizing) but
210	<	* we use only a simple spinlock (using bits in field runState),
211	<	* because submitters encountering a busy queue try or create
212	<	* others so never block.
202	>	* essence, submitters act like workers except that they are
203	>	* restricted to executing local tasks that they submitted (or in
204	>	* the case of CountedCompleters, others with the same root task).
205	>	* However, because most shared/external queue operations are more
206	>	* expensive than internal, and because, at steady state, external
207	>	* submitters will compete for CPU with workers, ForkJoinTask.join
208	>	* and related methods disable them from repeatedly helping to
209	>	* process tasks if all workers are active.  Insertion of tasks in
210	>	* shared mode requires a lock (mainly to protect in the case of
211	>	* resizing) but we use only a simple spinlock (using bits in
212	>	* field qlock), because submitters encountering a busy queue move
213	>	* on to try or create other queues -- they block only when
214	>	* creating and registering new queues.
215		*
216		* Management
217		* ==========
#	Line 231 | Line 233 | public class ForkJoinPool extends Abstra
233		* and their negations (used for thresholding) to fit into 16bit
234		* fields.
235		*
236	<	* Field "runState" contains 32 bits needed to register and
237	<	* deregister WorkQueues, as well as to enable shutdown. It is
238	<	* only modified under a lock (normally briefly held, but
239	<	* occasionally protecting allocations and resizings) but even
240	<	* when locked remains available to check consistency.
236	>	* Field "plock" is a form of sequence lock with a saturating
237	>	* shutdown bit (similarly for per-queue "qlocks"), mainly
238	>	* protecting updates to the workQueues array, as well as to
239	>	* enable shutdown.  When used as a lock, it is normally only very
240	>	* briefly held, so is nearly always available after at most a
241	>	* brief spin, but we use a monitor-based backup strategy to
242	>	* block when needed.
243		*
244		* Recording WorkQueues.  WorkQueues are recorded in the
245	<	* "workQueues" array that is created upon pool construction and
246	<	* expanded if necessary.  Updates to the array while recording
247	<	* new workers and unrecording terminated ones are protected from
248	<	* each other by a lock but the array is otherwise concurrently
249	<	* readable, and accessed directly.  To simplify index-based
250	<	* operations, the array size is always a power of two, and all
251	<	* readers must tolerate null slots. Shared (submission) queues
252	<	* are at even indices, worker queues at odd indices. Grouping
253	<	* them together in this way simplifies and speeds up task
254	<	* scanning. To avoid flailing during start-up, the array is
255	<	* presized to hold twice #parallelism workers (which is unlikely
256	<	* to need further resizing during execution). But to avoid
257	<	* dealing with so many null slots, variable runState includes a
258	<	* 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
245	>	* "workQueues" array that is created upon first use and expanded
246	>	* if necessary.  Updates to the array while recording new workers
247	>	* and unrecording terminated ones are protected from each other
248	>	* by a lock but the array is otherwise concurrently readable, and
249	>	* accessed directly.  To simplify index-based operations, the
250	>	* array size is always a power of two, and all readers must
251	>	* tolerate null slots. Worker queues are at odd indices. Shared
252	>	* (submission) queues are at even indices, up to a maximum of 64
253	>	* slots, to limit growth even if array needs to expand to add
254	>	* more workers. Grouping them together in this way simplifies and
255	>	* speeds up task scanning.
256	>	*
257	>	* All worker thread creation is on-demand, triggered by task
258	>	* submissions, replacement of terminated workers, and/or
259		* compensation for blocked workers. However, all other support
260		* code is set up to work with other policies.  To ensure that we
261		* do not hold on to worker references that would prevent GC, ALL
#	Line 266 | Line 268 | public class ForkJoinPool extends Abstra
268		* both index-check and null-check the IDs. All such accesses
269		* ignore bad IDs by returning out early from what they are doing,
270		* since this can only be associated with termination, in which
271	<	* case it is OK to give up.
272	<	*
273	<	* All uses of the workQueues array check that it is non-null
274	<	* (even if previously non-null). This allows nulling during
275	<	* termination, which is currently not necessary, but remains an
276	<	* option for resource-revocation-based shutdown schemes. It also
275	<	* helps reduce JIT issuance of uncommon-trap code, which tends to
271	>	* case it is OK to give up.  All uses of the workQueues array
272	>	* also check that it is non-null (even if previously
273	>	* non-null). This allows nulling during termination, which is
274	>	* currently not necessary, but remains an option for
275	>	* resource-revocation-based shutdown schemes. It also helps
276	>	* reduce JIT issuance of uncommon-trap code, which tends to
277		* unnecessarily complicate control flow in some methods.
278		*
279		* Event Queuing. Unlike HPC work-stealing frameworks, we cannot
#	Line 312 | Line 313 | public class ForkJoinPool extends Abstra
313		*
314		* Signalling.  We create or wake up workers only when there
315		* appears to be at least one task they might be able to find and
316	<	* execute.  When a submission is added or another worker adds a
317	<	* task to a queue that previously had fewer than two tasks, they
318	<	* signal waiting workers (or trigger creation of new ones if
319	<	* fewer than the given parallelism level -- see signalWork).
320	<	* These primary signals are buttressed by signals during rescans;
321	<	* together these cover the signals needed in cases when more
322	<	* tasks are pushed but untaken, and improve performance compared
323	<	* to having one thread wake up all workers.
316	>	* execute. However, many other threads may notice the same task
317	>	* and each signal to wake up a thread that might take it. So in
318	>	* general, pools will be over-signalled.  When a submission is
319	>	* added or another worker adds a task to a queue that has fewer
320	>	* than two tasks, they signal waiting workers (or trigger
321	>	* creation of new ones if fewer than the given parallelism level
322	>	* -- signalWork), and may leave a hint to the unparked worker to
323	>	* help signal others upon wakeup).  These primary signals are
324	>	* buttressed by others (see method helpSignal) whenever other
325	>	* threads scan for work or do not have a task to process.  On
326	>	* most platforms, signalling (unpark) overhead time is noticeably
327	>	* long, and the time between signalling a thread and it actually
328	>	* making progress can be very noticeably long, so it is worth
329	>	* offloading these delays from critical paths as much as
330	>	* possible.
331		*
332		* Trimming workers. To release resources after periods of lack of
333		* use, a worker starting to wait when the pool is quiescent will
334	<	* time out and terminate if the pool has remained quiescent for
335	<	* SHRINK_RATE nanosecs. This will slowly propagate, eventually
336	<	* terminating all workers after long periods of non-use.
334	>	* time out and terminate if the pool has remained quiescent for a
335	>	* given period -- a short period if there are more threads than
336	>	* parallelism, longer as the number of threads decreases. This
337	>	* will slowly propagate, eventually terminating all workers after
338	>	* periods of non-use.
339		*
340		* Shutdown and Termination. A call to shutdownNow atomically sets
341	<	* a runState bit and then (non-atomically) sets each worker's
342	<	* runState status, cancels all unprocessed tasks, and wakes up
341	>	* a plock bit and then (non-atomically) sets each worker's
342	>	* qlock status, cancels all unprocessed tasks, and wakes up
343		* all waiting workers.  Detecting whether termination should
344		* commence after a non-abrupt shutdown() call requires more work
345		* and bookkeeping. We need consensus about quiescence (i.e., that
#	Line 357 | Line 367 | public class ForkJoinPool extends Abstra
367		*      method tryCompensate() may create or re-activate a spare
368		*      thread to compensate for blocked joiners until they unblock.
369		*
370	<	* A third form (implemented in tryRemoveAndExec and
371	<	* tryPollForAndExec) amounts to helping a hypothetical
372	<	* compensator: If we can readily tell that a possible action of a
373	<	* compensator is to steal and execute the task being joined, the
374	<	* joining thread can do so directly, without the need for a
375	<	* compensation thread (although at the expense of larger run-time
376	<	* stacks, but the tradeoff is typically worthwhile).
370	>	* A third form (implemented in tryRemoveAndExec) amounts to
371	>	* helping a hypothetical compensator: If we can readily tell that
372	>	* a possible action of a compensator is to steal and execute the
373	>	* task being joined, the joining thread can do so directly,
374	>	* without the need for a compensation thread (although at the
375	>	* expense of larger run-time stacks, but the tradeoff is
376	>	* typically worthwhile).
377		*
378		* The ManagedBlocker extension API can't use helping so relies
379		* only on compensation in method awaitBlocker.
#	Line 383 | Line 393 | public class ForkJoinPool extends Abstra
393		* (http://portal.acm.org/citation.cfm?id=155354). It differs in
394		* that: (1) We only maintain dependency links across workers upon
395		* steals, rather than use per-task bookkeeping.  This sometimes
396	<	* requires a linear scan of workers array to locate stealers, but
397	<	* often doesn't because stealers leave hints (that may become
398	<	* stale/wrong) of where to locate them.  A stealHint is only a
399	<	* hint because a worker might have had multiple steals and the
400	<	* hint records only one of them (usually the most current).
401	<	* Hinting isolates cost to when it is needed, rather than adding
402	<	* to per-task overhead.  (2) It is "shallow", ignoring nesting
403	<	* and potentially cyclic mutual steals.  (3) It is intentionally
396	>	* requires a linear scan of workQueues array to locate stealers,
397	>	* but often doesn't because stealers leave hints (that may become
398	>	* stale/wrong) of where to locate them.  It is only a hint
399	>	* because a worker might have had multiple steals and the hint
400	>	* records only one of them (usually the most current).  Hinting
401	>	* isolates cost to when it is needed, rather than adding to
402	>	* per-task overhead.  (2) It is "shallow", ignoring nesting and
403	>	* potentially cyclic mutual steals.  (3) It is intentionally
404		* racy: field currentJoin is updated only while actively joining,
405		* which means that we miss links in the chain during long-lived
406		* tasks, GC stalls etc (which is OK since blocking in such cases
407		* is usually a good idea).  (4) We bound the number of attempts
408	<	* to find work (see MAX_HELP_DEPTH) and fall back to suspending
409	<	* the worker and if necessary replacing it with another.
408	>	* to find work (see MAX_HELP) and fall back to suspending the
409	>	* worker and if necessary replacing it with another.
410	>	*
411	>	* Helping actions for CountedCompleters are much simpler: Method
412	>	* helpComplete can take and execute any task with the same root
413	>	* as the task being waited on. However, this still entails some
414	>	* traversal of completer chains, so is less efficient than using
415	>	* CountedCompleters without explicit joins.
416		*
417		* It is impossible to keep exactly the target parallelism number
418		* of threads running at any given time.  Determining the
419		* existence of conservatively safe helping targets, the
420		* availability of already-created spares, and the apparent need
421		* to create new spares are all racy, so we rely on multiple
422	<	* retries of each.  Currently, in keeping with on-demand
423	<	* signalling policy, we compensate only if blocking would leave
424	<	* less than one active (non-waiting, non-blocked) worker.
425	<	* Additionally, to avoid some false alarms due to GC, lagging
426	<	* counters, system activity, etc, compensated blocking for joins
427	<	* is only attempted after rechecks stabilize in
428	<	* ForkJoinTask.awaitJoin. (Retries are interspersed with
429	<	* Thread.yield, for good citizenship.)
430	<	*
431	<	* Style notes: There is a lot of representation-level coupling
432	<	* among classes ForkJoinPool, ForkJoinWorkerThread, and
433	<	* ForkJoinTask.  The fields of WorkQueue maintain data structures
434	<	* managed by ForkJoinPool, so are directly accessed.  There is
435	<	* little point trying to reduce this, since any associated future
436	<	* changes in representations will need to be accompanied by
437	<	* algorithmic changes anyway. All together, these low-level
438	<	* implementation choices produce as much as a factor of 4
439	<	* performance improvement compared to naive implementations, and
440	<	* enable the processing of billions of tasks per second, at the
441	<	* expense of some ugliness.
442	<	*
443	<	* Methods signalWork() and scan() are the main bottlenecks, so are
444	<	* especially heavily micro-optimized/mangled.  There are lots of
445	<	* inline assignments (of form "while ((local = field) != 0)")
446	<	* which are usually the simplest way to ensure the required read
447	<	* orderings (which are sometimes critical). This leads to a
448	<	* "C"-like style of listing declarations of these locals at the
449	<	* heads of methods or blocks.  There are several occurrences of
450	<	* the unusual "do {} while (!cas...)"  which is the simplest way
451	<	* to force an update of a CAS'ed variable. There are also other
452	<	* coding oddities that help some methods perform reasonably even
453	<	* when interpreted (not compiled).
422	>	* retries of each.  Compensation in the apparent absence of
423	>	* helping opportunities is challenging to control on JVMs, where
424	>	* GC and other activities can stall progress of tasks that in
425	>	* turn stall out many other dependent tasks, without us being
426	>	* able to determine whether they will ever require compensation.
427	>	* Even though work-stealing otherwise encounters little
428	>	* degradation in the presence of more threads than cores,
429	>	* aggressively adding new threads in such cases entails risk of
430	>	* unwanted positive feedback control loops in which more threads
431	>	* cause more dependent stalls (as well as delayed progress of
432	>	* unblocked threads to the point that we know they are available)
433	>	* leading to more situations requiring more threads, and so
434	>	* on. This aspect of control can be seen as an (analytically
435	>	* intractable) game with an opponent that may choose the worst
436	>	* (for us) active thread to stall at any time.  We take several
437	>	* precautions to bound losses (and thus bound gains), mainly in
438	>	* methods tryCompensate and awaitJoin.
439	>	*
440	>	* Common Pool
441	>	* ===========
442	>	*
443	>	* The static common Pool always exists after static
444	>	* initialization.  Since it (or any other created pool) need
445	>	* never be used, we minimize initial construction overhead and
446	>	* footprint to the setup of about a dozen fields, with no nested
447	>	* allocation. Most bootstrapping occurs within method
448	>	* fullExternalPush during the first submission to the pool.
449	>	*
450	>	* When external threads submit to the common pool, they can
451	>	* perform some subtask processing (see externalHelpJoin and
452	>	* related methods).  We do not need to record whether these
453	>	* submissions are to the common pool -- if not, externalHelpJoin
454	>	* returns quickly (at the most helping to signal some common pool
455	>	* workers). These submitters would otherwise be blocked waiting
456	>	* for completion, so the extra effort (with liberally sprinkled
457	>	* task status checks) in inapplicable cases amounts to an odd
458	>	* form of limited spin-wait before blocking in ForkJoinTask.join.
459	>	*
460	>	* Style notes
461	>	* ===========
462	>	*
463	>	* There is a lot of representation-level coupling among classes
464	>	* ForkJoinPool, ForkJoinWorkerThread, and ForkJoinTask.  The
465	>	* fields of WorkQueue maintain data structures managed by
466	>	* ForkJoinPool, so are directly accessed.  There is little point
467	>	* trying to reduce this, since any associated future changes in
468	>	* representations will need to be accompanied by algorithmic
469	>	* changes anyway. Several methods intrinsically sprawl because
470	>	* they must accumulate sets of consistent reads of volatiles held
471	>	* in local variables.  Methods signalWork() and scan() are the
472	>	* main bottlenecks, so are especially heavily
473	>	* micro-optimized/mangled.  There are lots of inline assignments
474	>	* (of form "while ((local = field) != 0)") which are usually the
475	>	* simplest way to ensure the required read orderings (which are
476	>	* sometimes critical). This leads to a "C"-like style of listing
477	>	* declarations of these locals at the heads of methods or blocks.
478	>	* There are several occurrences of the unusual "do {} while
479	>	* (!cas...)"  which is the simplest way to force an update of a
480	>	* CAS'ed variable. There are also other coding oddities (including
481	>	* several unnecessary-looking hoisted null checks) that help
482	>	* some methods perform reasonably even when interpreted (not
483	>	* compiled).
484		*
485		* The order of declarations in this file is:
486	<	* (1) statics
487	<	* (2) fields (along with constants used when unpacking some of
488	<	*     them), listed in an order that tends to reduce contention
489	<	*     among them a bit under most JVMs;
490	<	* (3) nested classes
491	<	* (4) internal control methods
492	<	* (5) callbacks and other support for ForkJoinTask methods
493	<	* (6) exported methods (plus a few little helpers)
494	<	* (7) static block initializing all statics in a minimally
495	<	*     dependent order.
486	>	* (1) Static utility functions
487	>	* (2) Nested (static) classes
488	>	* (3) Static fields
489	>	* (4) Fields, along with constants used when unpacking some of them
490	>	* (5) Internal control methods
491	>	* (6) Callbacks and other support for ForkJoinTask methods
492	>	* (7) Exported methods
493	>	* (8) Static block initializing statics in minimally dependent order
494	>	*/
495	>
496	>	// Static utilities
497	>
498	>	/**
499	>	* If there is a security manager, makes sure caller has
500	>	* permission to modify threads.
501		*/
502	+	private static void checkPermission() {
503	+	SecurityManager security = System.getSecurityManager();
504	+	if (security != null)
505	+	security.checkPermission(modifyThreadPermission);
506	+	}
507	+
508	+	// Nested classes
509
510		/**
511		* Factory for creating new {@link ForkJoinWorkerThread}s.
#	Line 469 | Line 527 | public class ForkJoinPool extends Abstra
527		* Default ForkJoinWorkerThreadFactory implementation; creates a
528		* new ForkJoinWorkerThread.
529		*/
530	<	static class DefaultForkJoinWorkerThreadFactory
530	>	static final class DefaultForkJoinWorkerThreadFactory
531		implements ForkJoinWorkerThreadFactory {
532	<	public ForkJoinWorkerThread newThread(ForkJoinPool pool) {
532	>	public final ForkJoinWorkerThread newThread(ForkJoinPool pool) {
533		return new ForkJoinWorkerThread(pool);
534		}
535		}
536
537		/**
538	<	* Creates a new ForkJoinWorkerThread. This factory is used unless
539	<	* overridden in ForkJoinPool constructors.
540	<	*/
541	<	public static final ForkJoinWorkerThreadFactory
542	<	defaultForkJoinWorkerThreadFactory;
543	<
544	<	/**
545	<	* Permission required for callers of methods that may start or
546	<	* kill threads.
547	<	*/
548	<	private static final RuntimePermission modifyThreadPermission;
549	<
550	<	/**
493	<	* If there is a security manager, makes sure caller has
494	<	* permission to modify threads.
538	>	* Per-thread records for threads that submit to pools. Currently
539	>	* holds only pseudo-random seed / index that is used to choose
540	>	* submission queues in method externalPush. In the future, this may
541	>	* also incorporate a means to implement different task rejection
542	>	* and resubmission policies.
543	>	*
544	>	* Seeds for submitters and workers/workQueues work in basically
545	>	* the same way but are initialized and updated using slightly
546	>	* different mechanics. Both are initialized using the same
547	>	* approach as in class ThreadLocal, where successive values are
548	>	* unlikely to collide with previous values. Seeds are then
549	>	* randomly modified upon collisions using xorshifts, which
550	>	* requires a non-zero seed.
551		*/
552	<	private static void checkPermission() {
553	<	SecurityManager security = System.getSecurityManager();
554	<	if (security != null)
499	<	security.checkPermission(modifyThreadPermission);
552	>	static final class Submitter {
553	>	int seed;
554	>	Submitter(int s) { seed = s; }
555		}
556
557		/**
558	<	* Generator for assigning sequence numbers as pool names.
559	<	*/
560	<	private static final AtomicInteger poolNumberGenerator;
561	<
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.
558	>	* Class for artificial tasks that are used to replace the target
559	>	* of local joins if they are removed from an interior queue slot
560	>	* in WorkQueue.tryRemoveAndExec. We don't need the proxy to
561	>	* actually do anything beyond having a unique identity.
562		*/
563	<
564	<	volatile long ctl;                       // main pool control
565	<	final int parallelism;                   // parallelism level
566	<	final int localMode;                     // per-worker scheduling mode
567	<	int nextPoolIndex;                       // hint used in registerWorker
568	<	volatile int runState;                   // shutdown status, seq, and mask
569	<	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
563	>	static final class EmptyTask extends ForkJoinTask<Void> {
564	>	private static final long serialVersionUID = -7721805057305804111L;
565	>	EmptyTask() { status = ForkJoinTask.NORMAL; } // force done
566	>	public final Void getRawResult() { return null; }
567	>	public final void setRawResult(Void x) {}
568	>	public final boolean exec() { return true; }
569	>	}
570
571		/**
572		* Queues supporting work-stealing as well as external task
#	Line 650 | Line 582 | public class ForkJoinPool extends Abstra
582		*
583		* Field "top" is the index (mod array.length) of the next queue
584		* slot to push to or pop from. It is written only by owner thread
585	<	* for push, or under lock for trySharedPush, and accessed by
586	<	* other threads only after reading (volatile) base.  Both top and
587	<	* base are allowed to wrap around on overflow, but (top - base)
588	<	* (or more commonly -(base - top) to force volatile read of base
589	<	* before top) still estimates size.
585	>	* for push, or under lock for external/shared push, and accessed
586	>	* by other threads only after reading (volatile) base.  Both top
587	>	* and base are allowed to wrap around on overflow, but (top -
588	>	* base) (or more commonly -(base - top) to force volatile read of
589	>	* base before top) still estimates size. The lock ("qlock") is
590	>	* forced to -1 on termination, causing all further lock attempts
591	>	* to fail. (Note: we don't need CAS for termination state because
592	>	* upon pool shutdown, all shared-queues will stop being used
593	>	* anyway.)  Nearly all lock bodies are set up so that exceptions
594	>	* within lock bodies are "impossible" (modulo JVM errors that
595	>	* would cause failure anyway.)
596		*
597		* The array slots are read and written using the emulation of
598		* volatiles/atomics provided by Unsafe. Insertions must in
599		* general use putOrderedObject as a form of releasing store to
600		* ensure that all writes to the task object are ordered before
601	<	* its publication in the queue. (Although we can avoid one case
602	<	* of this when locked in trySharedPush.) All removals entail a
603	<	* CAS to null.  The array is always a power of two. To ensure
604	<	* safety of Unsafe array operations, all accesses perform
667	<	* explicit null checks and implicit bounds checks via
668	<	* power-of-two masking.
601	>	* its publication in the queue.  All removals entail a CAS to
602	>	* null.  The array is always a power of two. To ensure safety of
603	>	* Unsafe array operations, all accesses perform explicit null
604	>	* checks and implicit bounds checks via power-of-two masking.
605		*
606		* In addition to basic queuing support, this class contains
607		* fields described elsewhere to control execution. It turns out
608	<	* to work better memory-layout-wise to include them in this
609	<	* class rather than a separate class.
608	>	* to work better memory-layout-wise to include them in this class
609	>	* rather than a separate class.
610		*
611		* Performance on most platforms is very sensitive to placement of
612		* instances of both WorkQueues and their arrays -- we absolutely
#	Line 684 | Line 620 | public class ForkJoinPool extends Abstra
620		* trades off slightly slower average field access for the sake of
621		* avoiding really bad worst-case access. (Until better JVM
622		* support is in place, this padding is dependent on transient
623	<	* properties of JVM field layout rules.)  We also take care in
624	<	* allocating and sizing and resizing the array. Non-shared queue
625	<	* arrays are initialized (via method growArray) by workers before
626	<	* use. Others are allocated on first use.
623	>	* properties of JVM field layout rules.) We also take care in
624	>	* allocating, sizing and resizing the array. Non-shared queue
625	>	* arrays are initialized by workers before use. Others are
626	>	* allocated on first use.
627		*/
628		static final class WorkQueue {
629		/**
630		* Capacity of work-stealing queue array upon initialization.
631	<	* Must be a power of two; at least 4, but set larger to
632	<	* reduce cacheline sharing among queues.
631	>	* Must be a power of two; at least 4, but should be larger to
632	>	* reduce or eliminate cacheline sharing among queues.
633	>	* Currently, it is much larger, as a partial workaround for
634	>	* the fact that JVMs often place arrays in locations that
635	>	* share GC bookkeeping (especially cardmarks) such that
636	>	* per-write accesses encounter serious memory contention.
637		*/
638	<	static final int INITIAL_QUEUE_CAPACITY = 1 << 8;
638	>	static final int INITIAL_QUEUE_CAPACITY = 1 << 13;
639
640		/**
641		* Maximum size for queue arrays. Must be a power of two less
#	Line 706 | Line 646 | public class ForkJoinPool extends Abstra
646		*/
647		static final int MAXIMUM_QUEUE_CAPACITY = 1 << 26; // 64M
648
649	<	volatile long totalSteals; // cumulative number of steals
649	>	// Heuristic padding to ameliorate unfortunate memory placements
650	>	volatile long pad00, pad01, pad02, pad03, pad04, pad05, pad06;
651	>
652		int seed;                  // for random scanning; initialize nonzero
653		volatile int eventCount;   // encoded inactivation count; < 0 if inactive
654		int nextWait;              // encoded record of next event waiter
655	<	int rescans;               // remaining scans until block
714	<	int nsteals;               // top-level task executions since last idle
715	<	final int mode;            // lifo, fifo, or shared
655	>	int hint;                  // steal or signal hint (index)
656		int poolIndex;             // index of this queue in pool (or 0)
657	<	int stealHint;             // index of most recent known stealer
658	<	volatile int runState;     // 1: locked, -1: terminate; else 0
657	>	final int mode;            // 0: lifo, > 0: fifo, < 0: shared
658	>	int nsteals;               // number of steals
659	>	volatile int qlock;        // 1: locked, -1: terminate; else 0
660		volatile int base;         // index of next slot for poll
661		int top;                   // index of next slot for push
662		ForkJoinTask<?>[] array;   // the elements (initially unallocated)
663	+	final ForkJoinPool pool;   // the containing pool (may be null)
664		final ForkJoinWorkerThread owner; // owning thread or null if shared
665		volatile Thread parker;    // == owner during call to park; else null
666	<	ForkJoinTask<?> currentJoin;  // task being joined in awaitJoin
666	>	volatile ForkJoinTask<?> currentJoin;  // task being joined in awaitJoin
667		ForkJoinTask<?> currentSteal; // current non-local task being executed
726	–	// Heuristic padding to ameliorate unfortunate memory placements
727	–	Object p00, p01, p02, p03, p04, p05, p06, p07, p08, p09, p0a;
668
669	<	WorkQueue(ForkJoinWorkerThread owner, int mode) {
669	>	volatile Object pad10, pad11, pad12, pad13, pad14, pad15, pad16, pad17;
670	>	volatile Object pad18, pad19, pad1a, pad1b, pad1c, pad1d;
671	>
672	>	WorkQueue(ForkJoinPool pool, ForkJoinWorkerThread owner, int mode,
673	>	int seed) {
674	>	this.pool = pool;
675		this.owner = owner;
676		this.mode = mode;
677	+	this.seed = seed;
678		// Place indices in the center of array (that is not yet allocated)
679		base = top = INITIAL_QUEUE_CAPACITY >>> 1;
680		}
681
682		/**
683	<	* Returns number of tasks in the queue.
683	>	* Returns the approximate number of tasks in the queue.
684		*/
685		final int queueSize() {
686	<	int n = base - top; // non-owner callers must read base first
687	<	return (n >= 0) ? 0 : -n;
686	>	int n = base - top;       // non-owner callers must read base first
687	>	return (n >= 0) ? 0 : -n; // ignore transient negative
688	>	}
689	>
690	>	/**
691	>	* Provides a more accurate estimate of whether this queue has
692	>	* any tasks than does queueSize, by checking whether a
693	>	* near-empty queue has at least one unclaimed task.
694	>	*/
695	>	final boolean isEmpty() {
696	>	ForkJoinTask<?>[] a; int m, s;
697	>	int n = base - (s = top);
698	>	return (n >= 0 ||
699	>	(n == -1 &&
700	>	((a = array) == null ||
701	>	(m = a.length - 1) < 0 ||
702	>	U.getObject
703	>	(a, (long)((m & (s - 1)) << ASHIFT) + ABASE) == null)));
704		}
705
706		/**
707	<	* Pushes a task. Call only by owner in unshared queues.
707	>	* Pushes a task. Call only by owner in unshared queues.  (The
708	>	* shared-queue version is embedded in method externalPush.)
709		*
710		* @param task the task. Caller must ensure non-null.
748	–	* @param p if non-null, pool to signal if necessary
711		* @throw RejectedExecutionException if array cannot be resized
712		*/
713	<	final void push(ForkJoinTask<?> task, ForkJoinPool p) {
714	<	ForkJoinTask<?>[] a;
713	>	final void push(ForkJoinTask<?> task) {
714	>	ForkJoinTask<?>[] a; ForkJoinPool p;
715		int s = top, m, n;
716		if ((a = array) != null) {    // ignore if queue removed
717	<	U.putOrderedObject
718	<	(a, (((m = a.length - 1) & s) << ASHIFT) + ABASE, task);
717	>	int j = (((m = a.length - 1) & s) << ASHIFT) + ABASE;
718	>	U.putOrderedObject(a, j, task);
719		if ((n = (top = s + 1) - base) <= 2) {
720	<	if (p != null)
721	<	p.signalWork();
720	>	if ((p = pool) != null)
721	>	p.signalWork(this);
722		}
723		else if (n >= m)
724	<	growArray(true);
724	>	growArray();
725	>	}
726	>	}
727	>
728	>	/**
729	>	* Initializes or doubles the capacity of array. Call either
730	>	* by owner or with lock held -- it is OK for base, but not
731	>	* top, to move while resizings are in progress.
732	>	*/
733	>	final ForkJoinTask<?>[] growArray() {
734	>	ForkJoinTask<?>[] oldA = array;
735	>	int size = oldA != null ? oldA.length << 1 : INITIAL_QUEUE_CAPACITY;
736	>	if (size > MAXIMUM_QUEUE_CAPACITY)
737	>	throw new RejectedExecutionException("Queue capacity exceeded");
738	>	int oldMask, t, b;
739	>	ForkJoinTask<?>[] a = array = new ForkJoinTask<?>[size];
740	>	if (oldA != null && (oldMask = oldA.length - 1) >= 0 &&
741	>	(t = top) - (b = base) > 0) {
742	>	int mask = size - 1;
743	>	do {
744	>	ForkJoinTask<?> x;
745	>	int oldj = ((b & oldMask) << ASHIFT) + ABASE;
746	>	int j    = ((b &    mask) << ASHIFT) + ABASE;
747	>	x = (ForkJoinTask<?>)U.getObjectVolatile(oldA, oldj);
748	>	if (x != null &&
749	>	U.compareAndSwapObject(oldA, oldj, x, null))
750	>	U.putObjectVolatile(a, j, x);
751	>	} while (++b != t);
752		}
753	+	return a;
754		}
755
756		/**
757	<	* Pushes a task if lock is free and array is either big
758	<	* enough or can be resized to be big enough.
769	<	*
770	<	* @param task the task. Caller must ensure non-null.
771	<	* @return true if submitted
757	>	* Takes next task, if one exists, in LIFO order.  Call only
758	>	* by owner in unshared queues.
759		*/
760	<	final boolean trySharedPush(ForkJoinTask<?> task) {
761	<	boolean submitted = false;
762	<	if (runState == 0 && U.compareAndSwapInt(this, RUNSTATE, 0, 1)) {
763	<	ForkJoinTask<?>[] a = array;
764	<	int s = top, n = s - base;
765	<	try {
766	<	if ((a != null && n < a.length - 1) ||
767	<	(a = growArray(false)) != null) { // must presize
768	<	int j = (((a.length - 1) & s) << ASHIFT) + ABASE;
769	<	U.putObject(a, (long)j, task);    // don't need "ordered"
783	<	top = s + 1;
784	<	submitted = true;
760	>	final ForkJoinTask<?> pop() {
761	>	ForkJoinTask<?>[] a; ForkJoinTask<?> t; int m;
762	>	if ((a = array) != null && (m = a.length - 1) >= 0) {
763	>	for (int s; (s = top - 1) - base >= 0;) {
764	>	long j = ((m & s) << ASHIFT) + ABASE;
765	>	if ((t = (ForkJoinTask<?>)U.getObject(a, j)) == null)
766	>	break;
767	>	if (U.compareAndSwapObject(a, j, t, null)) {
768	>	top = s;
769	>	return t;
770		}
786	–	} finally {
787	–	runState = 0;                         // unlock
771		}
772		}
773	<	return submitted;
773	>	return null;
774		}
775
776		/**
777	<	* Takes next task, if one exists, in FIFO order.
777	>	* Takes a task in FIFO order if b is base of queue and a task
778	>	* can be claimed without contention. Specialized versions
779	>	* appear in ForkJoinPool methods scan and tryHelpStealer.
780		*/
781	<	final ForkJoinTask<?> poll() {
782	<	ForkJoinTask<?>[] a; int b, i;
783	<	while ((b = base) - top < 0 && (a = array) != null &&
784	<	(i = (a.length - 1) & b) >= 0) {
785	<	int j = (i << ASHIFT) + ABASE;
786	<	ForkJoinTask<?> t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
802	<	if (t != null && base == b &&
781	>	final ForkJoinTask<?> pollAt(int b) {
782	>	ForkJoinTask<?> t; ForkJoinTask<?>[] a;
783	>	if ((a = array) != null) {
784	>	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
785	>	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
786	>	base == b &&
787		U.compareAndSwapObject(a, j, t, null)) {
788		base = b + 1;
789		return t;
#	Line 809 | Line 793 | public class ForkJoinPool extends Abstra
793		}
794
795		/**
796	<	* Takes next task, if one exists, in LIFO order.
813	<	* Call only by owner in unshared queues.
796	>	* Takes next task, if one exists, in FIFO order.
797		*/
798	<	final ForkJoinTask<?> pop() {
799	<	ForkJoinTask<?> t; int m;
800	<	ForkJoinTask<?>[] a = array;
801	<	if (a != null && (m = a.length - 1) >= 0) {
802	<	for (int s; (s = top - 1) - base >= 0;) {
803	<	int j = ((m & s) << ASHIFT) + ABASE;
804	<	if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) == null)
805	<	break;
806	<	if (U.compareAndSwapObject(a, j, t, null)) {
824	<	top = s;
798	>	final ForkJoinTask<?> poll() {
799	>	ForkJoinTask<?>[] a; int b; ForkJoinTask<?> t;
800	>	while ((b = base) - top < 0 && (a = array) != null) {
801	>	int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
802	>	t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
803	>	if (t != null) {
804	>	if (base == b &&
805	>	U.compareAndSwapObject(a, j, t, null)) {
806	>	base = b + 1;
807		return t;
808		}
809		}
810	+	else if (base == b) {
811	+	if (b + 1 == top)
812	+	break;
813	+	Thread.yield(); // wait for lagging update (very rare)
814	+	}
815		}
816		return null;
817		}
#	Line 849 | Line 836 | public class ForkJoinPool extends Abstra
836		}
837
838		/**
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	–	/**
839		* Pops the given task only if it is at the current top.
840	+	* (A shared version is available only via FJP.tryExternalUnpush)
841		*/
842		final boolean tryUnpush(ForkJoinTask<?> t) {
843		ForkJoinTask<?>[] a; int s;
#	Line 881 | Line 851 | public class ForkJoinPool extends Abstra
851		}
852
853		/**
854	<	* Polls the given task only if it is at the current base.
854	>	* Removes and cancels all known tasks, ignoring any exceptions.
855		*/
856	<	final boolean pollFor(ForkJoinTask<?> task) {
857	<	ForkJoinTask<?>[] a; int b, i;
858	<	if ((b = base) - top < 0 && (a = array) != null &&
859	<	(i = (a.length - 1) & b) >= 0) {
860	<	int j = (i << ASHIFT) + ABASE;
861	<	if (U.getObjectVolatile(a, j) == task && base == b &&
862	<	U.compareAndSwapObject(a, j, task, null)) {
863	<	base = b + 1;
864	<	return true;
856	>	final void cancelAll() {
857	>	ForkJoinTask.cancelIgnoringExceptions(currentJoin);
858	>	ForkJoinTask.cancelIgnoringExceptions(currentSteal);
859	>	for (ForkJoinTask<?> t; (t = poll()) != null; )
860	>	ForkJoinTask.cancelIgnoringExceptions(t);
861	>	}
862	>
863	>	/**
864	>	* Computes next value for random probes.  Scans don't require
865	>	* a very high quality generator, but also not a crummy one.
866	>	* Marsaglia xor-shift is cheap and works well enough.  Note:
867	>	* This is manually inlined in its usages in ForkJoinPool to
868	>	* avoid writes inside busy scan loops.
869	>	*/
870	>	final int nextSeed() {
871	>	int r = seed;
872	>	r ^= r << 13;
873	>	r ^= r >>> 17;
874	>	return seed = r ^= r << 5;
875	>	}
876	>
877	>	// Specialized execution methods
878	>
879	>	/**
880	>	* Pops and runs tasks until empty.
881	>	*/
882	>	private void popAndExecAll() {
883	>	// A bit faster than repeated pop calls
884	>	ForkJoinTask<?>[] a; int m, s; long j; ForkJoinTask<?> t;
885	>	while ((a = array) != null && (m = a.length - 1) >= 0 &&
886	>	(s = top - 1) - base >= 0 &&
887	>	(t = ((ForkJoinTask<?>)
888	>	U.getObject(a, j = ((m & s) << ASHIFT) + ABASE)))
889	>	!= null) {
890	>	if (U.compareAndSwapObject(a, j, t, null)) {
891	>	top = s;
892	>	t.doExec();
893		}
894		}
897	–	return false;
895		}
896
897		/**
898	<	* If present, removes from queue and executes the given task, or
899	<	* any other cancelled task. Returns (true) immediately on any CAS
898	>	* Polls and runs tasks until empty.
899	>	*/
900	>	private void pollAndExecAll() {
901	>	for (ForkJoinTask<?> t; (t = poll()) != null;)
902	>	t.doExec();
903	>	}
904	>
905	>	/**
906	>	* If present, removes from queue and executes the given task,
907	>	* or any other cancelled task. Returns (true) on any CAS
908		* or consistency check failure so caller can retry.
909		*
910	<	* @return false if no progress can be made
910	>	* @return false if no progress can be made, else true;
911		*/
912		final boolean tryRemoveAndExec(ForkJoinTask<?> task) {
913	<	boolean removed = false, empty = true, progress = true;
913	>	boolean stat = true, removed = false, empty = true;
914		ForkJoinTask<?>[] a; int m, s, b, n;
915		if ((a = array) != null && (m = a.length - 1) >= 0 &&
916		(n = (s = top) - (b = base)) > 0) {
#	Line 935 | Line 940 | public class ForkJoinPool extends Abstra
940		}
941		if (--n == 0) {
942		if (!empty && base == b)
943	<	progress = false;
943	>	stat = false;
944		break;
945		}
946		}
947		}
948		if (removed)
949		task.doExec();
950	<	return progress;
950	>	return stat;
951		}
952
953		/**
954	<	* Initializes or doubles the capacity of array. Call either
955	<	* by owner or with lock held -- it is OK for base, but not
951	<	* top, to move while resizings are in progress.
952	<	*
953	<	* @param rejectOnFailure if true, throw exception if capacity
954	<	* exceeded (relayed ultimately to user); else return null.
954	>	* Polls for and executes the given task or any other task in
955	>	* its CountedCompleter computation
956		*/
957	<	final ForkJoinTask<?>[] growArray(boolean rejectOnFailure) {
958	<	ForkJoinTask<?>[] oldA = array;
959	<	int size = oldA != null ? oldA.length << 1 : INITIAL_QUEUE_CAPACITY;
960	<	if (size <= MAXIMUM_QUEUE_CAPACITY) {
961	<	int oldMask, t, b;
962	<	ForkJoinTask<?>[] a = array = new ForkJoinTask<?>[size];
963	<	if (oldA != null && (oldMask = oldA.length - 1) >= 0 &&
964	<	(t = top) - (b = base) > 0) {
965	<	int mask = size - 1;
966	<	do {
967	<	ForkJoinTask<?> x;
968	<	int oldj = ((b & oldMask) << ASHIFT) + ABASE;
969	<	int j    = ((b &    mask) << ASHIFT) + ABASE;
970	<	x = (ForkJoinTask<?>)U.getObjectVolatile(oldA, oldj);
971	<	if (x != null &&
972	<	U.compareAndSwapObject(oldA, oldj, x, null))
973	<	U.putObjectVolatile(a, j, x);
974	<	} while (++b != t);
957	>	final boolean pollAndExecCC(ForkJoinTask<?> root) {
958	>	ForkJoinTask<?>[] a; int b; Object o;
959	>	outer: while ((b = base) - top < 0 && (a = array) != null) {
960	>	long j = (((a.length - 1) & b) << ASHIFT) + ABASE;
961	>	if ((o = U.getObject(a, j)) == null ||
962	>	!(o instanceof CountedCompleter))
963	>	break;
964	>	for (CountedCompleter<?> t = (CountedCompleter<?>)o, r = t;;) {
965	>	if (r == root) {
966	>	if (base == b &&
967	>	U.compareAndSwapObject(a, j, t, null)) {
968	>	base = b + 1;
969	>	t.doExec();
970	>	return true;
971	>	}
972	>	else
973	>	break; // restart
974	>	}
975	>	if ((r = r.completer) == null)
976	>	break outer; // not part of root computation
977		}
975	–	return a;
976	–	}
977	–	else if (!rejectOnFailure)
978	–	return null;
979	–	else
980	–	throw new RejectedExecutionException("Queue capacity exceeded");
981	–	}
982	–
983	–	/**
984	–	* Removes and cancels all known tasks, ignoring any exceptions.
985	–	*/
986	–	final void cancelAll() {
987	–	ForkJoinTask.cancelIgnoringExceptions(currentJoin);
988	–	ForkJoinTask.cancelIgnoringExceptions(currentSteal);
989	–	for (ForkJoinTask<?> t; (t = poll()) != null; )
990	–	ForkJoinTask.cancelIgnoringExceptions(t);
991	–	}
992	–
993	–	// Execution methods
994	–
995	–	/**
996	–	* Removes and runs tasks until empty, using local mode
997	–	* ordering.
998	–	*/
999	–	final void runLocalTasks() {
1000	–	if (base - top < 0) {
1001	–	for (ForkJoinTask<?> t; (t = nextLocalTask()) != null; )
1002	–	t.doExec();
978		}
979	+	return false;
980		}
981
982		/**
983		* Executes a top-level task and any local tasks remaining
984		* after execution.
1009	–	*
1010	–	* @return true unless terminating
985		*/
986	<	final boolean runTask(ForkJoinTask<?> t) {
1013	<	boolean alive = true;
986	>	final void runTask(ForkJoinTask<?> t) {
987		if (t != null) {
988	<	currentSteal = t;
1016	<	t.doExec();
1017	<	runLocalTasks();
1018	<	++nsteals;
988	>	(currentSteal = t).doExec();
989		currentSteal = null;
990	+	++nsteals;
991	+	if (base - top < 0) {       // process remaining local tasks
992	+	if (mode == 0)
993	+	popAndExecAll();
994	+	else
995	+	pollAndExecAll();
996	+	}
997		}
1021	–	else if (runState < 0)            // terminating
1022	–	alive = false;
1023	–	return alive;
998		}
999
1000		/**
#	Line 1029 | Line 1003 | public class ForkJoinPool extends Abstra
1003		final void runSubtask(ForkJoinTask<?> t) {
1004		if (t != null) {
1005		ForkJoinTask<?> ps = currentSteal;
1006	<	currentSteal = t;
1033	<	t.doExec();
1006	>	(currentSteal = t).doExec();
1007		currentSteal = ps;
1008		}
1009		}
1010
1011		/**
1012	<	* 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.
1012	>	* Returns true if owned and not known to be blocked.
1013		*/
1014	<	final int nextSeed() {
1015	<	int r = seed;
1016	<	r ^= r << 13;
1017	<	r ^= r >>> 17;
1018	<	r ^= r << 5;
1019	<	return seed = r;
1014	>	final boolean isApparentlyUnblocked() {
1015	>	Thread wt; Thread.State s;
1016	>	return (eventCount >= 0 &&
1017	>	(wt = owner) != null &&
1018	>	(s = wt.getState()) != Thread.State.BLOCKED &&
1019	>	s != Thread.State.WAITING &&
1020	>	s != Thread.State.TIMED_WAITING);
1021		}
1022
1023		// Unsafe mechanics
1024		private static final sun.misc.Unsafe U;
1025	<	private static final long RUNSTATE;
1025	>	private static final long QLOCK;
1026		private static final int ABASE;
1027		private static final int ASHIFT;
1028		static {
#	Line 1061 | Line 1031 | public class ForkJoinPool extends Abstra
1031		U = getUnsafe();
1032		Class<?> k = WorkQueue.class;
1033		Class<?> ak = ForkJoinTask[].class;
1034	<	RUNSTATE = U.objectFieldOffset
1035	<	(k.getDeclaredField("runState"));
1034	>	QLOCK = U.objectFieldOffset
1035	>	(k.getDeclaredField("qlock"));
1036		ABASE = U.arrayBaseOffset(ak);
1037		s = U.arrayIndexScale(ak);
1038		} catch (Exception e) {
#	Line 1074 | Line 1044 | public class ForkJoinPool extends Abstra
1044		}
1045		}
1046
1047	+	// static fields (initialized in static initializer below)
1048	+
1049		/**
1050	<	* Class for artificial tasks that are used to replace the target
1051	<	* of local joins if they are removed from an interior queue slot
1080	<	* in WorkQueue.tryRemoveAndExec. We don't need the proxy to
1081	<	* actually do anything beyond having a unique identity.
1050	>	* Creates a new ForkJoinWorkerThread. This factory is used unless
1051	>	* overridden in ForkJoinPool constructors.
1052		*/
1053	<	static final class EmptyTask extends ForkJoinTask<Void> {
1054	<	EmptyTask() { status = ForkJoinTask.NORMAL; } // force done
1085	<	public Void getRawResult() { return null; }
1086	<	public void setRawResult(Void x) {}
1087	<	public boolean exec() { return true; }
1088	<	}
1053	>	public static final ForkJoinWorkerThreadFactory
1054	>	defaultForkJoinWorkerThreadFactory;
1055
1056		/**
1057	<	* Per-thread records for (typically non-FJ) threads that submit
1058	<	* to pools. Cureently holds only psuedo-random seed / index that
1059	<	* is used to choose submission queues in method doSubmit. In the
1060	<	* future, this may incorporate a means to implement different
1061	<	* task rejection and resubmission policies.
1057	>	* Per-thread submission bookkeeping. Shared across all pools
1058	>	* to reduce ThreadLocal pollution and because random motion
1059	>	* to avoid contention in one pool is likely to hold for others.
1060	>	* Lazily initialized on first submission (but null-checked
1061	>	* in other contexts to avoid unnecessary initialization).
1062		*/
1063	<	static final class Submitter {
1098	<	int seed; // seed for random submission queue selection
1063	>	static final ThreadLocal<Submitter> submitters;
1064
1065	<	// Heuristic padding to ameliorate unfortunate memory placements
1066	<	int p0, p1, p2, p3, p4, p5, p6, p7, p8, p9, pa, pb, pc, pd, pe;
1065	>	/**
1066	>	* Permission required for callers of methods that may start or
1067	>	* kill threads.
1068	>	*/
1069	>	private static final RuntimePermission modifyThreadPermission;
1070
1071	<	Submitter() {
1072	<	// Use identityHashCode, forced negative, for seed
1073	<	seed = System.identityHashCode(Thread.currentThread()) | (1 << 31);
1074	<	}
1071	>	/**
1072	>	* Common (static) pool. Non-null for public use unless a static
1073	>	* construction exception, but internal usages null-check on use
1074	>	* to paranoically avoid potential initialization circularities
1075	>	* as well as to simplify generated code.
1076	>	*/
1077	>	static final ForkJoinPool common;
1078
1079	<	/**
1080	<	* Computes next value for random probes.  Like method
1081	<	* WorkQueue.nextSeed, this is manually inlined in several
1082	<	* usages to avoid writes inside busy loops.
1083	<	*/
1084	<	final int nextSeed() {
1085	<	int r = seed;
1086	<	r ^= r << 13;
1087	<	r ^= r >>> 17;
1117	<	return seed = r ^= r << 5;
1118	<	}
1119	<	}
1079	>	/**
1080	>	* Common pool parallelism. Must equal common.parallelism.
1081	>	*/
1082	>	static final int commonParallelism;
1083	>
1084	>	/**
1085	>	* Sequence number for creating workerNamePrefix.
1086	>	*/
1087	>	private static int poolNumberSequence;
1088
1089	<	/** ThreadLocal class for Submitters */
1090	<	static final class ThreadSubmitter extends ThreadLocal<Submitter> {
1091	<	public Submitter initialValue() { return new Submitter(); }
1089	>	/**
1090	>	* Returns the next sequence number. We don't expect this to
1091	>	* ever contend, so use simple builtin sync.
1092	>	*/
1093	>	private static final synchronized int nextPoolId() {
1094	>	return ++poolNumberSequence;
1095		}
1096
1097	+	// static constants
1098	+
1099		/**
1100	<	* Per-thread submission bookeeping. Shared across all pools
1101	<	* to reduce ThreadLocal pollution and because random motion
1102	<	* to avoid contention in one pool is likely to hold for others.
1100	>	* Initial timeout value (in nanoseconds) for the thread
1101	>	* triggering quiescence to park waiting for new work. On timeout,
1102	>	* the thread will instead try to shrink the number of
1103	>	* workers. The value should be large enough to avoid overly
1104	>	* aggressive shrinkage during most transient stalls (long GCs
1105	>	* etc).
1106		*/
1107	<	static final ThreadSubmitter submitters = new ThreadSubmitter();
1107	>	private static final long IDLE_TIMEOUT      = 2000L * 1000L * 1000L; // 2sec
1108
1109		/**
1110	<	* Top-level runloop for workers
1110	>	* Timeout value when there are more threads than parallelism level
1111		*/
1112	<	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);
1112	>	private static final long FAST_IDLE_TIMEOUT =  200L * 1000L * 1000L;
1113
1114	<	do {} while (w.runTask(scan(w)));
1115	<	}
1114	>	/**
1115	>	* Tolerance for idle timeouts, to cope with timer undershoots
1116	>	*/
1117	>	private static final long TIMEOUT_SLOP = 2000000L;
1118
1119	<	// Creating, registering and deregistering workers
1119	>	/**
1120	>	* The maximum stolen->joining link depth allowed in method
1121	>	* tryHelpStealer.  Must be a power of two.  Depths for legitimate
1122	>	* chains are unbounded, but we use a fixed constant to avoid
1123	>	* (otherwise unchecked) cycles and to bound staleness of
1124	>	* traversal parameters at the expense of sometimes blocking when
1125	>	* we could be helping.
1126	>	*/
1127	>	private static final int MAX_HELP = 64;
1128
1129		/**
1130	<	* Tries to create and start a worker
1130	>	* Increment for seed generators. See class ThreadLocal for
1131	>	* explanation.
1132		*/
1133	<	private void addWorker() {
1134	<	Throwable ex = null;
1135	<	ForkJoinWorkerThread w = null;
1136	<	try {
1137	<	if ((w = factory.newThread(this)) != null) {
1138	<	w.start();
1139	<	return;
1133	>	private static final int SEED_INCREMENT = 0x61c88647;
1134	>
1135	>	/**
1136	>	* Bits and masks for control variables
1137	>	*
1138	>	* Field ctl is a long packed with:
1139	>	* AC: Number of active running workers minus target parallelism (16 bits)
1140	>	* TC: Number of total workers minus target parallelism (16 bits)
1141	>	* ST: true if pool is terminating (1 bit)
1142	>	* EC: the wait count of top waiting thread (15 bits)
1143	>	* ID: poolIndex of top of Treiber stack of waiters (16 bits)
1144	>	*
1145	>	* When convenient, we can extract the upper 32 bits of counts and
1146	>	* the lower 32 bits of queue state, u = (int)(ctl >>> 32) and e =
1147	>	* (int)ctl.  The ec field is never accessed alone, but always
1148	>	* together with id and st. The offsets of counts by the target
1149	>	* parallelism and the positionings of fields makes it possible to
1150	>	* perform the most common checks via sign tests of fields: When
1151	>	* ac is negative, there are not enough active workers, when tc is
1152	>	* negative, there are not enough total workers, and when e is
1153	>	* negative, the pool is terminating.  To deal with these possibly
1154	>	* negative fields, we use casts in and out of "short" and/or
1155	>	* signed shifts to maintain signedness.
1156	>	*
1157	>	* When a thread is queued (inactivated), its eventCount field is
1158	>	* set negative, which is the only way to tell if a worker is
1159	>	* prevented from executing tasks, even though it must continue to
1160	>	* scan for them to avoid queuing races. Note however that
1161	>	* eventCount updates lag releases so usage requires care.
1162	>	*
1163	>	* Field plock is an int packed with:
1164	>	* SHUTDOWN: true if shutdown is enabled (1 bit)
1165	>	* SEQ:  a sequence lock, with PL_LOCK bit set if locked (30 bits)
1166	>	* SIGNAL: set when threads may be waiting on the lock (1 bit)
1167	>	*
1168	>	* The sequence number enables simple consistency checks:
1169	>	* Staleness of read-only operations on the workQueues array can
1170	>	* be checked by comparing plock before vs after the reads.
1171	>	*/
1172	>
1173	>	// bit positions/shifts for fields
1174	>	private static final int  AC_SHIFT   = 48;
1175	>	private static final int  TC_SHIFT   = 32;
1176	>	private static final int  ST_SHIFT   = 31;
1177	>	private static final int  EC_SHIFT   = 16;
1178	>
1179	>	// bounds
1180	>	private static final int  SMASK      = 0xffff;  // short bits
1181	>	private static final int  MAX_CAP    = 0x7fff;  // max #workers - 1
1182	>	private static final int  EVENMASK   = 0xfffe;  // even short bits
1183	>	private static final int  SQMASK     = 0x007e;  // max 64 (even) slots
1184	>	private static final int  SHORT_SIGN = 1 << 15;
1185	>	private static final int  INT_SIGN   = 1 << 31;
1186	>
1187	>	// masks
1188	>	private static final long STOP_BIT   = 0x0001L << ST_SHIFT;
1189	>	private static final long AC_MASK    = ((long)SMASK) << AC_SHIFT;
1190	>	private static final long TC_MASK    = ((long)SMASK) << TC_SHIFT;
1191	>
1192	>	// units for incrementing and decrementing
1193	>	private static final long TC_UNIT    = 1L << TC_SHIFT;
1194	>	private static final long AC_UNIT    = 1L << AC_SHIFT;
1195	>
1196	>	// masks and units for dealing with u = (int)(ctl >>> 32)
1197	>	private static final int  UAC_SHIFT  = AC_SHIFT - 32;
1198	>	private static final int  UTC_SHIFT  = TC_SHIFT - 32;
1199	>	private static final int  UAC_MASK   = SMASK << UAC_SHIFT;
1200	>	private static final int  UTC_MASK   = SMASK << UTC_SHIFT;
1201	>	private static final int  UAC_UNIT   = 1 << UAC_SHIFT;
1202	>	private static final int  UTC_UNIT   = 1 << UTC_SHIFT;
1203	>
1204	>	// masks and units for dealing with e = (int)ctl
1205	>	private static final int E_MASK      = 0x7fffffff; // no STOP_BIT
1206	>	private static final int E_SEQ       = 1 << EC_SHIFT;
1207	>
1208	>	// plock bits
1209	>	private static final int SHUTDOWN    = 1 << 31;
1210	>	private static final int PL_LOCK     = 2;
1211	>	private static final int PL_SIGNAL   = 1;
1212	>	private static final int PL_SPINS    = 1 << 8;
1213	>
1214	>	// access mode for WorkQueue
1215	>	static final int LIFO_QUEUE          =  0;
1216	>	static final int FIFO_QUEUE          =  1;
1217	>	static final int SHARED_QUEUE        = -1;
1218	>
1219	>	// bounds for #steps in scan loop -- must be power 2 minus 1
1220	>	private static final int MIN_SCAN    = 0x1ff;   // cover estimation slop
1221	>	private static final int MAX_SCAN    = 0x1ffff; // 4 * max workers
1222	>
1223	>	// Instance fields
1224	>
1225	>	/*
1226	>	* Field layout of this class tends to matter more than one would
1227	>	* like. Runtime layout order is only loosely related to
1228	>	* declaration order and may differ across JVMs, but the following
1229	>	* empirically works OK on current JVMs.
1230	>	*/
1231	>
1232	>	// Heuristic padding to ameliorate unfortunate memory placements
1233	>	volatile long pad00, pad01, pad02, pad03, pad04, pad05, pad06;
1234	>
1235	>	volatile long stealCount;                  // collects worker counts
1236	>	volatile long ctl;                         // main pool control
1237	>	volatile int plock;                        // shutdown status and seqLock
1238	>	volatile int indexSeed;                    // worker/submitter index seed
1239	>	final int config;                          // mode and parallelism level
1240	>	WorkQueue[] workQueues;                    // main registry
1241	>	final ForkJoinWorkerThreadFactory factory;
1242	>	final Thread.UncaughtExceptionHandler ueh; // per-worker UEH
1243	>	final String workerNamePrefix;             // to create worker name string
1244	>
1245	>	volatile Object pad10, pad11, pad12, pad13, pad14, pad15, pad16, pad17;
1246	>	volatile Object pad18, pad19, pad1a, pad1b;
1247	>
1248	>	/*
1249	>	* Acquires the plock lock to protect worker array and related
1250	>	* updates. This method is called only if an initial CAS on plock
1251	>	* fails. This acts as a spinlock for normal cases, but falls back
1252	>	* to builtin monitor to block when (rarely) needed. This would be
1253	>	* a terrible idea for a highly contended lock, but works fine as
1254	>	* a more conservative alternative to a pure spinlock.
1255	>	*/
1256	>	private int acquirePlock() {
1257	>	int spins = PL_SPINS, r = 0, ps, nps;
1258	>	for (;;) {
1259	>	if (((ps = plock) & PL_LOCK) == 0 &&
1260	>	U.compareAndSwapInt(this, PLOCK, ps, nps = ps + PL_LOCK))
1261	>	return nps;
1262	>	else if (r == 0) { // randomize spins if possible
1263	>	Thread t = Thread.currentThread(); WorkQueue w; Submitter z;
1264	>	if ((t instanceof ForkJoinWorkerThread) &&
1265	>	(w = ((ForkJoinWorkerThread)t).workQueue) != null)
1266	>	r = w.seed;
1267	>	else if ((z = submitters.get()) != null)
1268	>	r = z.seed;
1269	>	else
1270	>	r = 1;
1271	>	}
1272	>	else if (spins >= 0) {
1273	>	r ^= r << 1; r ^= r >>> 3; r ^= r << 10; // xorshift
1274	>	if (r >= 0)
1275	>	--spins;
1276	>	}
1277	>	else if (U.compareAndSwapInt(this, PLOCK, ps, ps | PL_SIGNAL)) {
1278	>	synchronized (this) {
1279	>	if ((plock & PL_SIGNAL) != 0) {
1280	>	try {
1281	>	wait();
1282	>	} catch (InterruptedException ie) {
1283	>	try {
1284	>	Thread.currentThread().interrupt();
1285	>	} catch (SecurityException ignore) {
1286	>	}
1287	>	}
1288	>	}
1289	>	else
1290	>	notifyAll();
1291	>	}
1292		}
1159	–	} catch (Throwable e) {
1160	–	ex = e;
1293		}
1162	–	deregisterWorker(w, ex);
1294		}
1295
1296		/**
1297	<	* Callback from ForkJoinWorkerThread constructor to assign a
1298	<	* public name. This must be separate from registerWorker because
1299	<	* it is called during the "super" constructor call in
1300	<	* ForkJoinWorkerThread.
1297	>	* Unlocks and signals any thread waiting for plock. Called only
1298	>	* when CAS of seq value for unlock fails.
1299	>	*/
1300	>	private void releasePlock(int ps) {
1301	>	plock = ps;
1302	>	synchronized (this) { notifyAll(); }
1303	>	}
1304	>
1305	>	/**
1306	>	* Tries to create and start one worker if fewer than target
1307	>	* parallelism level exist. Adjusts counts etc on failure.
1308		*/
1309	<	final String nextWorkerName() {
1310	<	return workerNamePrefix.concat
1311	<	(Integer.toString(nextWorkerNumber.addAndGet(1)));
1309	>	private void tryAddWorker() {
1310	>	long c; int u;
1311	>	while ((u = (int)((c = ctl) >>> 32)) < 0 &&
1312	>	(u & SHORT_SIGN) != 0 && (int)c == 0) {
1313	>	long nc = (long)(((u + UTC_UNIT) & UTC_MASK) |
1314	>	((u + UAC_UNIT) & UAC_MASK)) << 32;
1315	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1316	>	ForkJoinWorkerThreadFactory fac;
1317	>	Throwable ex = null;
1318	>	ForkJoinWorkerThread wt = null;
1319	>	try {
1320	>	if ((fac = factory) != null &&
1321	>	(wt = fac.newThread(this)) != null) {
1322	>	wt.start();
1323	>	break;
1324	>	}
1325	>	} catch (Throwable e) {
1326	>	ex = e;
1327	>	}
1328	>	deregisterWorker(wt, ex);
1329	>	break;
1330	>	}
1331	>	}
1332		}
1333
1334	+	//  Registering and deregistering workers
1335	+
1336		/**
1337	<	* Callback from ForkJoinWorkerThread constructor to establish and
1338	<	* record its WorkQueue.
1337	>	* Callback from ForkJoinWorkerThread to establish and record its
1338	>	* WorkQueue. To avoid scanning bias due to packing entries in
1339	>	* front of the workQueues array, we treat the array as a simple
1340	>	* power-of-two hash table using per-thread seed as hash,
1341	>	* expanding as needed.
1342		*
1343		* @param wt the worker thread
1344	+	* @return the worker's queue
1345		*/
1346	<	final void registerWorker(ForkJoinWorkerThread wt) {
1347	<	WorkQueue w = wt.workQueue;
1348	<	ReentrantLock lock = this.lock;
1349	<	lock.lock();
1346	>	final WorkQueue registerWorker(ForkJoinWorkerThread wt) {
1347	>	Thread.UncaughtExceptionHandler handler; WorkQueue[] ws; int s, ps;
1348	>	wt.setDaemon(true);
1349	>	if ((handler = ueh) != null)
1350	>	wt.setUncaughtExceptionHandler(handler);
1351	>	do {} while (!U.compareAndSwapInt(this, INDEXSEED, s = indexSeed,
1352	>	s += SEED_INCREMENT) ||
1353	>	s == 0); // skip 0
1354	>	WorkQueue w = new WorkQueue(this, wt, config >>> 16, s);
1355	>	if (((ps = plock) & PL_LOCK) != 0 ||
1356	>	!U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1357	>	ps = acquirePlock();
1358	>	int nps = (ps & SHUTDOWN) | ((ps + PL_LOCK) & ~SHUTDOWN);
1359		try {
1360	<	int k = nextPoolIndex;
1361	<	WorkQueue[] ws = workQueues;
1362	<	if (ws != null) {                       // ignore on shutdown
1363	<	int n = ws.length;
1364	<	if (k < 0 || (k & 1) == 0 || k >= n || ws[k] != null) {
1365	<	for (k = 1; k < n && ws[k] != null; k += 2)
1366	<	;                           // workers are at odd indices
1367	<	if (k >= n)                     // resize
1368	<	workQueues = ws = Arrays.copyOf(ws, n << 1);
1369	<	}
1370	<	w.poolIndex = k;
1371	<	w.eventCount = ~(k >>> 1) & SMASK;  // Set up wait count
1372	<	ws[k] = w;                          // record worker
1373	<	nextPoolIndex = k + 2;
1374	<	int rs = runState;
1375	<	int m = rs & SMASK;                 // recalculate runState mask
1203	<	if (k > m)
1204	<	m = (m << 1) + 1;
1205	<	runState = (rs & SHUTDOWN) | ((rs + RS_SEQ) & RS_SEQ_MASK) | m;
1360	>	if ((ws = workQueues) != null) {    // skip if shutting down
1361	>	int n = ws.length, m = n - 1;
1362	>	int r = (s << 1) | 1;           // use odd-numbered indices
1363	>	if (ws[r &= m] != null) {       // collision
1364	>	int probes = 0;             // step by approx half size
1365	>	int step = (n <= 4) ? 2 : ((n >>> 1) & EVENMASK) + 2;
1366	>	while (ws[r = (r + step) & m] != null) {
1367	>	if (++probes >= n) {
1368	>	workQueues = ws = Arrays.copyOf(ws, n <<= 1);
1369	>	m = n - 1;
1370	>	probes = 0;
1371	>	}
1372	>	}
1373	>	}
1374	>	w.eventCount = w.poolIndex = r; // volatile write orders
1375	>	ws[r] = w;
1376		}
1377		} finally {
1378	<	lock.unlock();
1378	>	if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
1379	>	releasePlock(nps);
1380		}
1381	+	wt.setName(workerNamePrefix.concat(Integer.toString(w.poolIndex)));
1382	+	return w;
1383		}
1384
1385		/**
1386	<	* Final callback from terminating worker, as well as failure to
1387	<	* construct or start a worker in addWorker.  Removes record of
1388	<	* worker from array, and adjusts counts. If pool is shutting
1389	<	* down, tries to complete termination.
1386	>	* Final callback from terminating worker, as well as upon failure
1387	>	* to construct or start a worker.  Removes record of worker from
1388	>	* array, and adjusts counts. If pool is shutting down, tries to
1389	>	* complete termination.
1390		*
1391	<	* @param wt the worker thread or null if addWorker failed
1391	>	* @param wt the worker thread or null if construction failed
1392		* @param ex the exception causing failure, or null if none
1393		*/
1394		final void deregisterWorker(ForkJoinWorkerThread wt, Throwable ex) {
1395		WorkQueue w = null;
1396		if (wt != null && (w = wt.workQueue) != null) {
1397	<	w.runState = -1;                // ensure runState is set
1398	<	stealCount.getAndAdd(w.totalSteals + w.nsteals);
1399	<	int idx = w.poolIndex;
1400	<	ReentrantLock lock = this.lock;
1401	<	lock.lock();
1402	<	try {                           // remove record from array
1397	>	int ps;
1398	>	w.qlock = -1;                // ensure set
1399	>	long ns = w.nsteals, sc;     // collect steal count
1400	>	do {} while (!U.compareAndSwapLong(this, STEALCOUNT,
1401	>	sc = stealCount, sc + ns));
1402	>	if (((ps = plock) & PL_LOCK) != 0 ||
1403	>	!U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1404	>	ps = acquirePlock();
1405	>	int nps = (ps & SHUTDOWN) | ((ps + PL_LOCK) & ~SHUTDOWN);
1406	>	try {
1407	>	int idx = w.poolIndex;
1408		WorkQueue[] ws = workQueues;
1409		if (ws != null && idx >= 0 && idx < ws.length && ws[idx] == w)
1410	<	ws[nextPoolIndex = idx] = null;
1410	>	ws[idx] = null;
1411		} finally {
1412	<	lock.unlock();
1412	>	if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
1413	>	releasePlock(nps);
1414		}
1415		}
1416
1417	<	long c;                             // adjust ctl counts
1417	>	long c;                          // adjust ctl counts
1418		do {} while (!U.compareAndSwapLong
1419		(this, CTL, c = ctl, (((c - AC_UNIT) & AC_MASK) |
1420		((c - TC_UNIT) & TC_MASK) |
1421		(c & ~(AC_MASK|TC_MASK)))));
1422
1423	<	if (!tryTerminate(false) && w != null) {
1424	<	w.cancelAll();                  // cancel remaining tasks
1425	<	if (w.array != null)            // suppress signal if never ran
1426	<	signalWork();               // wake up or create replacement
1423	>	if (!tryTerminate(false, false) && w != null && w.array != null) {
1424	>	w.cancelAll();               // cancel remaining tasks
1425	>	WorkQueue[] ws; WorkQueue v; Thread p; int u, i, e;
1426	>	while ((u = (int)((c = ctl) >>> 32)) < 0 && (e = (int)c) >= 0) {
1427	>	if (e > 0) {             // activate or create replacement
1428	>	if ((ws = workQueues) == null ||
1429	>	(i = e & SMASK) >= ws.length ||
1430	>	(v = ws[i]) == null)
1431	>	break;
1432	>	long nc = (((long)(v.nextWait & E_MASK)) |
1433	>	((long)(u + UAC_UNIT) << 32));
1434	>	if (v.eventCount != (e | INT_SIGN))
1435	>	break;
1436	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1437	>	v.eventCount = (e + E_SEQ) & E_MASK;
1438	>	if ((p = v.parker) != null)
1439	>	U.unpark(p);
1440	>	break;
1441	>	}
1442	>	}
1443	>	else {
1444	>	if ((short)u < 0)
1445	>	tryAddWorker();
1446	>	break;
1447	>	}
1448	>	}
1449		}
1450	+	if (ex == null)                     // help clean refs on way out
1451	+	ForkJoinTask.helpExpungeStaleExceptions();
1452	+	else                                // rethrow
1453	+	ForkJoinTask.rethrow(ex);
1454	+	}
1455
1456	<	if (ex != null)                     // rethrow
1457	<	U.throwException(ex);
1456	>	// Submissions
1457	>
1458	>	/**
1459	>	* Unless shutting down, adds the given task to a submission queue
1460	>	* at submitter's current queue index (modulo submission
1461	>	* range). Only the most common path is directly handled in this
1462	>	* method. All others are relayed to fullExternalPush.
1463	>	*
1464	>	* @param task the task. Caller must ensure non-null.
1465	>	*/
1466	>	final void externalPush(ForkJoinTask<?> task) {
1467	>	WorkQueue[] ws; WorkQueue q; Submitter z; int m; ForkJoinTask<?>[] a;
1468	>	if ((z = submitters.get()) != null && plock > 0 &&
1469	>	(ws = workQueues) != null && (m = (ws.length - 1)) >= 0 &&
1470	>	(q = ws[m & z.seed & SQMASK]) != null &&
1471	>	U.compareAndSwapInt(q, QLOCK, 0, 1)) { // lock
1472	>	int b = q.base, s = q.top, n, an;
1473	>	if ((a = q.array) != null && (an = a.length) > (n = s + 1 - b)) {
1474	>	int j = (((an - 1) & s) << ASHIFT) + ABASE;
1475	>	U.putOrderedObject(a, j, task);
1476	>	q.top = s + 1;                     // push on to deque
1477	>	q.qlock = 0;
1478	>	if (n <= 2)
1479	>	signalWork(q);
1480	>	return;
1481	>	}
1482	>	q.qlock = 0;
1483	>	}
1484	>	fullExternalPush(task);
1485		}
1486
1487		/**
1488	<	* Tries to add and register a new queue at the given index.
1489	<	*
1490	<	* @param idx the workQueues array index to register the queue
1491	<	* @return the queue, or null if could not add because could
1492	<	* not acquire lock or idx is unusable
1493	<	*/
1494	<	private WorkQueue tryAddSharedQueue(int idx) {
1495	<	WorkQueue q = null;
1496	<	ReentrantLock lock = this.lock;
1497	<	if (idx >= 0 && (idx & 1) == 0 && !lock.isLocked()) {
1498	<	// create queue outside of lock but only if apparently free
1499	<	WorkQueue nq = new WorkQueue(null, SHARED_QUEUE);
1500	<	if (lock.tryLock()) {
1501	<	try {
1502	<	WorkQueue[] ws = workQueues;
1503	<	if (ws != null && idx < ws.length) {
1504	<	if ((q = ws[idx]) == null) {
1505	<	int rs;         // update runState seq
1506	<	ws[idx] = q = nq;
1507	<	runState = (((rs = runState) & SHUTDOWN) |
1508	<	((rs + RS_SEQ) & ~SHUTDOWN));
1488	>	* Full version of externalPush. This method is called, among
1489	>	* other times, upon the first submission of the first task to the
1490	>	* pool, so must perform secondary initialization.  It also
1491	>	* detects first submission by an external thread by looking up
1492	>	* its ThreadLocal, and creates a new shared queue if the one at
1493	>	* index if empty or contended. The plock lock body must be
1494	>	* exception-free (so no try/finally) so we optimistically
1495	>	* allocate new queues outside the lock and throw them away if
1496	>	* (very rarely) not needed.
1497	>	*
1498	>	* Secondary initialization occurs when plock is zero, to create
1499	>	* workQueue array and set plock to a valid value.  This lock body
1500	>	* must also be exception-free. Because the plock seq value can
1501	>	* eventually wrap around zero, this method harmlessly fails to
1502	>	* reinitialize if workQueues exists, while still advancing plock.
1503	>	*/
1504	>	private void fullExternalPush(ForkJoinTask<?> task) {
1505	>	int r = 0; // random index seed
1506	>	for (Submitter z = submitters.get();;) {
1507	>	WorkQueue[] ws; WorkQueue q; int ps, m, k;
1508	>	if (z == null) {
1509	>	if (U.compareAndSwapInt(this, INDEXSEED, r = indexSeed,
1510	>	r += SEED_INCREMENT) && r != 0)
1511	>	submitters.set(z = new Submitter(r));
1512	>	}
1513	>	else if (r == 0) {                  // move to a different index
1514	>	r = z.seed;
1515	>	r ^= r << 13;                   // same xorshift as WorkQueues
1516	>	r ^= r >>> 17;
1517	>	z.seed = r ^ (r << 5);
1518	>	}
1519	>	else if ((ps = plock) < 0)
1520	>	throw new RejectedExecutionException();
1521	>	else if (ps == 0 || (ws = workQueues) == null ||
1522	>	(m = ws.length - 1) < 0) { // initialize workQueues
1523	>	int p = config & SMASK;         // find power of two table size
1524	>	int n = (p > 1) ? p - 1 : 1;    // ensure at least 2 slots
1525	>	n |= n >>> 1; n |= n >>> 2;  n |= n >>> 4;
1526	>	n |= n >>> 8; n |= n >>> 16; n = (n + 1) << 1;
1527	>	WorkQueue[] nws = ((ws = workQueues) == null || ws.length == 0 ?
1528	>	new WorkQueue[n] : null);
1529	>	if (((ps = plock) & PL_LOCK) != 0 ||
1530	>	!U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1531	>	ps = acquirePlock();
1532	>	if (((ws = workQueues) == null || ws.length == 0) && nws != null)
1533	>	workQueues = nws;
1534	>	int nps = (ps & SHUTDOWN) | ((ps + PL_LOCK) & ~SHUTDOWN);
1535	>	if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
1536	>	releasePlock(nps);
1537	>	}
1538	>	else if ((q = ws[k = r & m & SQMASK]) != null) {
1539	>	if (q.qlock == 0 && U.compareAndSwapInt(q, QLOCK, 0, 1)) {
1540	>	ForkJoinTask<?>[] a = q.array;
1541	>	int s = q.top;
1542	>	boolean submitted = false;
1543	>	try {                      // locked version of push
1544	>	if ((a != null && a.length > s + 1 - q.base) ||
1545	>	(a = q.growArray()) != null) {   // must presize
1546	>	int j = (((a.length - 1) & s) << ASHIFT) + ABASE;
1547	>	U.putOrderedObject(a, j, task);
1548	>	q.top = s + 1;
1549	>	submitted = true;
1550		}
1551	+	} finally {
1552	+	q.qlock = 0;  // unlock
1553	+	}
1554	+	if (submitted) {
1555	+	signalWork(q);
1556	+	return;
1557		}
1278	–	} finally {
1279	–	lock.unlock();
1558		}
1559	+	r = 0; // move on failure
1560	+	}
1561	+	else if (((ps = plock) & PL_LOCK) == 0) { // create new queue
1562	+	q = new WorkQueue(this, null, SHARED_QUEUE, r);
1563	+	if (((ps = plock) & PL_LOCK) != 0 ||
1564	+	!U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1565	+	ps = acquirePlock();
1566	+	if ((ws = workQueues) != null && k < ws.length && ws[k] == null)
1567	+	ws[k] = q;
1568	+	int nps = (ps & SHUTDOWN) | ((ps + PL_LOCK) & ~SHUTDOWN);
1569	+	if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
1570	+	releasePlock(nps);
1571		}
1572	+	else
1573	+	r = 0; // try elsewhere while lock held
1574		}
1283	–	return q;
1575		}
1576
1577		// Maintaining ctl counts
#	Line 1294 | Line 1585 | public class ForkJoinPool extends Abstra
1585		}
1586
1587		/**
1588	<	* Activates or creates a worker.
1588	>	* Tries to create or activate a worker if too few are active.
1589	>	*
1590	>	* @param q the (non-null) queue holding tasks to be signalled
1591		*/
1592	<	final void signalWork() {
1593	<	/*
1594	<	* The while condition is true if: (there is are too few total
1595	<	* workers OR there is at least one waiter) AND (there are too
1596	<	* few active workers OR the pool is terminating).  The value
1597	<	* of e distinguishes the remaining cases: zero (no waiters)
1598	<	* for create, negative if terminating (in which case do
1599	<	* nothing), else release a waiter. The secondary checks for
1600	<	* release (non-null array etc) can fail if the pool begins
1601	<	* terminating after the test, and don't impose any added cost
1602	<	* because JVMs must perform null and bounds checks anyway.
1603	<	*/
1604	<	long c; int e, u;
1605	<	while ((((e = (int)(c = ctl)) | (u = (int)(c >>> 32))) &
1606	<	(INT_SIGN|SHORT_SIGN)) == (INT_SIGN|SHORT_SIGN)) {
1607	<	WorkQueue[] ws = workQueues; int i; WorkQueue w; Thread p;
1608	<	if (e == 0) {                    // add a new worker
1609	<	if (U.compareAndSwapLong
1317	<	(this, CTL, c, (long)(((u + UTC_UNIT) & UTC_MASK) |
1318	<	((u + UAC_UNIT) & UAC_MASK)) << 32)) {
1319	<	addWorker();
1320	<	break;
1592	>	final void signalWork(WorkQueue q) {
1593	>	int hint = q.poolIndex;
1594	>	long c; int e, u, i, n; WorkQueue[] ws; WorkQueue w; Thread p;
1595	>	while ((u = (int)((c = ctl) >>> 32)) < 0) {
1596	>	if ((e = (int)c) > 0) {
1597	>	if ((ws = workQueues) != null && ws.length > (i = e & SMASK) &&
1598	>	(w = ws[i]) != null && w.eventCount == (e | INT_SIGN)) {
1599	>	long nc = (((long)(w.nextWait & E_MASK)) |
1600	>	((long)(u + UAC_UNIT) << 32));
1601	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1602	>	w.hint = hint;
1603	>	w.eventCount = (e + E_SEQ) & E_MASK;
1604	>	if ((p = w.parker) != null)
1605	>	U.unpark(p);
1606	>	break;
1607	>	}
1608	>	if (q.top - q.base <= 0)
1609	>	break;
1610		}
1611	<	}
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
1611	>	else
1612		break;
1334	–	}
1613		}
1614	<	else
1614	>	else {
1615	>	if ((short)u < 0)
1616	>	tryAddWorker();
1617		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
1372	–	}
1373	–	else if (tc + pc < MAX_ID) {
1374	–	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
1375	–	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1376	–	addWorker();
1377	–	return true;             // create replacement
1378	–	}
1618		}
1619		}
1381	–	return false;
1620		}
1621
1622	<	// Submissions
1622	>	// Scanning for tasks
1623
1624		/**
1625	<	* 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.
1625	>	* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
1626		*/
1627	<	private void doSubmit(ForkJoinTask<?> task) {
1628	<	if (task == null)
1629	<	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	<	}
1627	>	final void runWorker(WorkQueue w) {
1628	>	w.growArray(); // allocate queue
1629	>	do { w.runTask(scan(w)); } while (w.qlock >= 0);
1630		}
1631
1416	–
1417	–	// Scanning for tasks
1418	–
1632		/**
1633		* Scans for and, if found, returns one task, else possibly
1634		* inactivates the worker. This method operates on single reads of
1635	<	* volatile state and is designed to be re-invoked continuously in
1636	<	* part because it returns upon detecting inconsistencies,
1635	>	* volatile state and is designed to be re-invoked continuously,
1636	>	* in part because it returns upon detecting inconsistencies,
1637		* contention, or state changes that indicate possible success on
1638		* re-invocation.
1639		*
1640	<	* The scan searches for tasks across queues, randomly selecting
1641	<	* the first #queues probes, favoring steals 2:1 over submissions
1642	<	* (by exploiting even/odd indexing), and then performing a
1643	<	* circular sweep of all queues.  The scan terminates upon either
1644	<	* finding a non-empty queue, or completing a full sweep. If the
1645	<	* worker is not inactivated, it takes and returns a task from
1646	<	* this queue.  On failure to find a task, we take one of the
1647	<	* following actions, after which the caller will retry calling
1648	<	* this method unless terminated.
1649	<	*
1650	<	* * If not a complete sweep, try to release a waiting worker.  If
1651	<	* the scan terminated because the worker is inactivated, then the
1439	<	* released worker will often be the calling worker, and it can
1440	<	* succeed obtaining a task on the next call. Or maybe it is
1441	<	* another worker, but with same net effect. Releasing in other
1442	<	* cases as well ensures that we have enough workers running.
1443	<	*
1444	<	* * If the caller has run a task since 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.
1640	>	* The scan searches for tasks across queues (starting at a random
1641	>	* index, and relying on registerWorker to irregularly scatter
1642	>	* them within array to avoid bias), checking each at least twice.
1643	>	* The scan terminates upon either finding a non-empty queue, or
1644	>	* completing the sweep. If the worker is not inactivated, it
1645	>	* takes and returns a task from this queue. Otherwise, if not
1646	>	* activated, it signals workers (that may include itself) and
1647	>	* returns so caller can retry. Also returns for true if the
1648	>	* worker array may have changed during an empty scan.  On failure
1649	>	* to find a task, we take one of the following actions, after
1650	>	* which the caller will retry calling this method unless
1651	>	* terminated.
1652		*
1653		* * If pool is terminating, terminate the worker.
1654		*
1655		* * If not already enqueued, try to inactivate and enqueue the
1656	<	* worker on wait queue.
1656	>	* worker on wait queue. Or, if inactivating has caused the pool
1657	>	* to be quiescent, relay to idleAwaitWork to possibly shrink
1658	>	* pool.
1659	>	*
1660	>	* * If already enqueued and none of the above apply, possibly
1661	>	* park awaiting signal, else lingering to help scan and signal.
1662		*
1663	<	* * If already enqueued and none of the above apply, either park
1664	<	* 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.
1663	>	* * If a non-empty queue discovered or left as a hint,
1664	>	* help wake up other workers before return.
1665		*
1666		* @param w the worker (via its WorkQueue)
1667	<	* @return a task or null of none found
1667	>	* @return a task or null if none found
1668		*/
1669		private final ForkJoinTask<?> scan(WorkQueue w) {
1670	<	boolean swept = false;                 // true after full empty scan
1671	<	WorkQueue[] ws;                        // volatile read order matters
1672	<	int r = w.seed, ec = w.eventCount;     // ec is negative if inactive
1673	<	int rs = runState, m = rs & SMASK;
1674	<	if ((ws = workQueues) != null && ws.length > m) {
1675	<	ForkJoinTask<?> task = null;
1676	<	for (int k = 0, j = -2 - m; ; ++j) {
1677	<	WorkQueue q; int b;
1678	<	if (j < 0) {                   // random probes while j negative
1679	<	r ^= r << 13; r ^= r >>> 17; k = (r ^= r << 5) | (j & 1);
1680	<	}                              // worker (not submit) for odd j
1681	<	else                           // cyclic scan when j >= 0
1682	<	k += (m >>> 1) | 1;        // step by half to reduce bias
1683	<
1684	<	if ((q = ws[k & m]) != null && (b = q.base) - q.top < 0) {
1685	<	if (ec >= 0)
1686	<	task = q.pollAt(b);    // steal
1687	<	break;
1670	>	WorkQueue[] ws; int m;
1671	>	int ps = plock;                          // read plock before ws
1672	>	if (w != null && (ws = workQueues) != null && (m = ws.length - 1) >= 0) {
1673	>	int ec = w.eventCount;               // ec is negative if inactive
1674	>	int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1675	>	w.hint = -1;                         // update seed and clear hint
1676	>	int j = ((m + m + 1) | MIN_SCAN) & MAX_SCAN;
1677	>	do {
1678	>	WorkQueue q; ForkJoinTask<?>[] a; int b;
1679	>	if ((q = ws[(r + j) & m]) != null && (b = q.base) - q.top < 0 &&
1680	>	(a = q.array) != null) {     // probably nonempty
1681	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1682	>	ForkJoinTask<?> t = (ForkJoinTask<?>)
1683	>	U.getObjectVolatile(a, i);
1684	>	if (q.base == b && ec >= 0 && t != null &&
1685	>	U.compareAndSwapObject(a, i, t, null)) {
1686	>	if ((q.base = b + 1) - q.top < 0)
1687	>	signalWork(q);
1688	>	return t;                // taken
1689	>	}
1690	>	else if ((ec < 0 || j < m) && (int)(ctl >> AC_SHIFT) <= 0) {
1691	>	w.hint = (r + j) & m;    // help signal below
1692	>	break;                   // cannot take
1693	>	}
1694		}
1695	<	else if (j > m) {
1696	<	if (rs == runState)        // staleness check
1697	<	swept = true;
1698	<	break;
1695	>	} while (--j >= 0);
1696	>
1697	>	int h, e, ns; long c, sc; WorkQueue q;
1698	>	if ((ns = w.nsteals) != 0) {
1699	>	if (U.compareAndSwapLong(this, STEALCOUNT,
1700	>	sc = stealCount, sc + ns))
1701	>	w.nsteals = 0;               // collect steals and rescan
1702	>	}
1703	>	else if (plock != ps)                // consistency check
1704	>	;                                // skip
1705	>	else if ((e = (int)(c = ctl)) < 0)
1706	>	w.qlock = -1;                    // pool is terminating
1707	>	else {
1708	>	if ((h = w.hint) < 0) {
1709	>	if (ec >= 0) {               // try to enqueue/inactivate
1710	>	long nc = (((long)ec |
1711	>	((c - AC_UNIT) & (AC_MASK|TC_MASK))));
1712	>	w.nextWait = e;          // link and mark inactive
1713	>	w.eventCount = ec | INT_SIGN;
1714	>	if (ctl != c || !U.compareAndSwapLong(this, CTL, c, nc))
1715	>	w.eventCount = ec;   // unmark on CAS failure
1716	>	else if ((int)(c >> AC_SHIFT) == 1 - (config & SMASK))
1717	>	idleAwaitWork(w, nc, c);
1718	>	}
1719	>	else if (w.eventCount < 0 && ctl == c) {
1720	>	Thread wt = Thread.currentThread();
1721	>	Thread.interrupted();    // clear status
1722	>	U.putObject(wt, PARKBLOCKER, this);
1723	>	w.parker = wt;           // emulate LockSupport.park
1724	>	if (w.eventCount < 0)    // recheck
1725	>	U.park(false, 0L);   // block
1726	>	w.parker = null;
1727	>	U.putObject(wt, PARKBLOCKER, null);
1728	>	}
1729	>	}
1730	>	if ((h >= 0 || (h = w.hint) >= 0) &&
1731	>	(ws = workQueues) != null && h < ws.length &&
1732	>	(q = ws[h]) != null) {      // signal others before retry
1733	>	WorkQueue v; Thread p; int u, i, s;
1734	>	for (int n = (config & SMASK) - 1;;) {
1735	>	int idleCount = (w.eventCount < 0) ? 0 : -1;
1736	>	if (((s = idleCount - q.base + q.top) <= n &&
1737	>	(n = s) <= 0) ||
1738	>	(u = (int)((c = ctl) >>> 32)) >= 0 ||
1739	>	(e = (int)c) <= 0 || m < (i = e & SMASK) ||
1740	>	(v = ws[i]) == null)
1741	>	break;
1742	>	long nc = (((long)(v.nextWait & E_MASK)) |
1743	>	((long)(u + UAC_UNIT) << 32));
1744	>	if (v.eventCount != (e | INT_SIGN) ||
1745	>	!U.compareAndSwapLong(this, CTL, c, nc))
1746	>	break;
1747	>	v.hint = h;
1748	>	v.eventCount = (e + E_SEQ) & E_MASK;
1749	>	if ((p = v.parker) != null)
1750	>	U.unpark(p);
1751	>	if (--n <= 0)
1752	>	break;
1753	>	}
1754		}
1489	–	}
1490	–	w.seed = r;                        // save seed for next scan
1491	–	if (task != null)
1492	–	return task;
1493	–	}
1494	–
1495	–	// Decode ctl on empty scan
1496	–	long c = ctl; int e = (int)c, a = (int)(c >> AC_SHIFT), nr, ns;
1497	–	if (!swept) {                          // try to release a waiter
1498	–	WorkQueue v; Thread p;
1499	–	if (e > 0 && a < 0 && ws != null &&
1500	–	(v = ws[((~e << 1) | 1) & m]) != null &&
1501	–	v.eventCount == (e | INT_SIGN) && U.compareAndSwapLong
1502	–	(this, CTL, c, ((long)(v.nextWait & E_MASK) |
1503	–	((c + AC_UNIT) & (AC_MASK|TC_MASK))))) {
1504	–	v.eventCount = (e + E_SEQ) & E_MASK;
1505	–	if ((p = v.parker) != null)
1506	–	U.unpark(p);
1507	–	}
1508	–	}
1509	–	else if ((nr = w.rescans) > 0) {       // continue rescanning
1510	–	int ac = a + parallelism;
1511	–	if ((w.rescans = (ac < nr) ? ac : nr - 1) > 0 && w.seed < 0 &&
1512	–	w.eventCount == ec)
1513	–	Thread.yield();                // 1 bit randomness for yield call
1514	–	}
1515	–	else if (e < 0)                        // pool is terminating
1516	–	w.runState = -1;
1517	–	else if (ec >= 0) {                    // try to enqueue
1518	–	long nc = (long)ec | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
1519	–	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);
1755		}
1756		}
1757		return null;
1758		}
1759
1760		/**
1761	<	* If inactivating worker w has caused pool to become quiescent,
1762	<	* checks for pool termination, and, so long as this is not the
1763	<	* only worker, waits for event for up to SHRINK_RATE nanosecs.
1764	<	* On timeout, if ctl has not changed, terminates the worker,
1765	<	* which will in turn wake up another worker to possibly repeat
1766	<	* this process.
1761	>	* If inactivating worker w has caused the pool to become
1762	>	* quiescent, checks for pool termination, and, so long as this is
1763	>	* not the only worker, waits for event for up to a given
1764	>	* duration.  On timeout, if ctl has not changed, terminates the
1765	>	* worker, which will in turn wake up another worker to possibly
1766	>	* repeat this process.
1767		*
1768		* @param w the calling worker
1769	+	* @param currentCtl the ctl value triggering possible quiescence
1770	+	* @param prevCtl the ctl value to restore if thread is terminated
1771		*/
1772	<	private void idleAwaitWork(WorkQueue w) {
1773	<	long c; int nw, ec;
1774	<	if (!tryTerminate(false) &&
1775	<	(int)((c = ctl) >> AC_SHIFT) + parallelism == 0 &&
1776	<	(ec = w.eventCount) == ((int)c | INT_SIGN) &&
1777	<	(nw = w.nextWait) != 0) {
1778	<	long nc = ((long)(nw & E_MASK) | // ctl to restore on timeout
1779	<	((c + AC_UNIT) & AC_MASK) | (c & TC_MASK));
1780	<	ForkJoinTask.helpExpungeStaleExceptions(); // help clean
1566	<	ForkJoinWorkerThread wt = w.owner;
1567	<	while (ctl == c) {
1568	<	long startTime = System.nanoTime();
1772	>	private void idleAwaitWork(WorkQueue w, long currentCtl, long prevCtl) {
1773	>	if (w != null && w.eventCount < 0 &&
1774	>	!tryTerminate(false, false) && (int)prevCtl != 0 &&
1775	>	ctl == currentCtl) {
1776	>	int dc = -(short)(currentCtl >>> TC_SHIFT);
1777	>	long parkTime = dc < 0 ? FAST_IDLE_TIMEOUT: (dc + 1) * IDLE_TIMEOUT;
1778	>	long deadline = System.nanoTime() + parkTime - TIMEOUT_SLOP;
1779	>	Thread wt = Thread.currentThread();
1780	>	while (ctl == currentCtl) {
1781		Thread.interrupted();  // timed variant of version in scan()
1782		U.putObject(wt, PARKBLOCKER, this);
1783		w.parker = wt;
1784	<	if (ctl == c)
1785	<	U.park(false, SHRINK_RATE);
1784	>	if (ctl == currentCtl)
1785	>	U.park(false, parkTime);
1786		w.parker = null;
1787		U.putObject(wt, PARKBLOCKER, null);
1788	<	if (ctl != c)
1788	>	if (ctl != currentCtl)
1789		break;
1790	<	if (System.nanoTime() - startTime >= SHRINK_TIMEOUT &&
1791	<	U.compareAndSwapLong(this, CTL, c, nc)) {
1792	<	w.runState = -1;          // shrink
1793	<	w.eventCount = (ec + E_SEQ) | E_MASK;
1790	>	if (deadline - System.nanoTime() <= 0L &&
1791	>	U.compareAndSwapLong(this, CTL, currentCtl, prevCtl)) {
1792	>	w.eventCount = (w.eventCount + E_SEQ) | E_MASK;
1793	>	w.hint = -1;
1794	>	w.qlock = -1;   // shrink
1795		break;
1796		}
1797		}
#	Line 1586 | Line 1799 | public class ForkJoinPool extends Abstra
1799		}
1800
1801		/**
1802	+	* Scans through queues looking for work while joining a task; if
1803	+	* any present, signals. May return early if more signalling is
1804	+	* detectably unneeded.
1805	+	*
1806	+	* @param task return early if done
1807	+	* @param origin an index to start scan
1808	+	*/
1809	+	private void helpSignal(ForkJoinTask<?> task, int origin) {
1810	+	WorkQueue[] ws; WorkQueue w; Thread p; long c; int m, u, e, i, s;
1811	+	if (task != null && task.status >= 0 &&
1812	+	(u = (int)(ctl >>> 32)) < 0 && (u >> UAC_SHIFT) < 0 &&
1813	+	(ws = workQueues) != null && (m = ws.length - 1) >= 0) {
1814	+	outer: for (int k = origin, j = m; j >= 0; --j) {
1815	+	WorkQueue q = ws[k++ & m];
1816	+	for (int n = m;;) { // limit to at most m signals
1817	+	if (task.status < 0)
1818	+	break outer;
1819	+	if (q == null ||
1820	+	((s = -q.base + q.top) <= n && (n = s) <= 0))
1821	+	break;
1822	+	if ((u = (int)((c = ctl) >>> 32)) >= 0 ||
1823	+	(e = (int)c) <= 0 || m < (i = e & SMASK) ||
1824	+	(w = ws[i]) == null)
1825	+	break outer;
1826	+	long nc = (((long)(w.nextWait & E_MASK)) |
1827	+	((long)(u + UAC_UNIT) << 32));
1828	+	if (w.eventCount != (e | INT_SIGN))
1829	+	break outer;
1830	+	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1831	+	w.eventCount = (e + E_SEQ) & E_MASK;
1832	+	if ((p = w.parker) != null)
1833	+	U.unpark(p);
1834	+	if (--n <= 0)
1835	+	break;
1836	+	}
1837	+	}
1838	+	}
1839	+	}
1840	+	}
1841	+
1842	+	/**
1843		* Tries to locate and execute tasks for a stealer of the given
1844		* task, or in turn one of its stealers, Traces currentSteal ->
1845		* currentJoin links looking for a thread working on a descendant
#	Line 1596 | Line 1850 | public class ForkJoinPool extends Abstra
1850		* leaves hints in workers to speed up subsequent calls. The
1851		* implementation is very branchy to cope with potential
1852		* inconsistencies or loops encountering chains that are stale,
1853	<	* unknown, or of length greater than MAX_HELP_DEPTH links.  All
1600	<	* of these cases are dealt with by just retrying by caller.
1853	>	* unknown, or so long that they are likely cyclic.
1854		*
1855		* @param joiner the joining worker
1856		* @param task the task to join
1857	<	* @return true if found or ran a task (and so is immediately retryable)
1857	>	* @return 0 if no progress can be made, negative if task
1858	>	* known complete, else positive
1859		*/
1860	<	final boolean tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1861	<	ForkJoinTask<?> subtask;    // current target
1862	<	boolean progress = false;
1863	<	int depth = 0;              // current chain depth
1864	<	int m = runState & SMASK;
1865	<	WorkQueue[] ws = workQueues;
1866	<
1867	<	if (ws != null && ws.length > m && (subtask = task).status >= 0) {
1868	<	outer:for (WorkQueue j = joiner;;) {
1869	<	// Try to find the stealer of subtask, by first using hint
1870	<	WorkQueue stealer = null;
1871	<	WorkQueue v = ws[j.stealHint & m];
1872	<	if (v != null && v.currentSteal == subtask)
1873	<	stealer = v;
1874	<	else {
1875	<	for (int i = 1; i <= m; i += 2) {
1876	<	if ((v = ws[i]) != null && v.currentSteal == subtask) {
1877	<	stealer = v;
1878	<	j.stealHint = i; // save hint
1879	<	break;
1860	>	private int tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1861	>	int stat = 0, steps = 0;                    // bound to avoid cycles
1862	>	if (joiner != null && task != null) {       // hoist null checks
1863	>	restart: for (;;) {
1864	>	ForkJoinTask<?> subtask = task;     // current target
1865	>	for (WorkQueue j = joiner, v;;) {   // v is stealer of subtask
1866	>	WorkQueue[] ws; int m, s, h;
1867	>	if ((s = task.status) < 0) {
1868	>	stat = s;
1869	>	break restart;
1870	>	}
1871	>	if ((ws = workQueues) == null || (m = ws.length - 1) <= 0)
1872	>	break restart;              // shutting down
1873	>	if ((v = ws[h = (j.hint | 1) & m]) == null ||
1874	>	v.currentSteal != subtask) {
1875	>	for (int origin = h;;) {    // find stealer
1876	>	if (((h = (h + 2) & m) & 15) == 1 &&
1877	>	(subtask.status < 0 || j.currentJoin != subtask))
1878	>	continue restart;   // occasional staleness check
1879	>	if ((v = ws[h]) != null &&
1880	>	v.currentSteal == subtask) {
1881	>	j.hint = h;        // save hint
1882	>	break;
1883	>	}
1884	>	if (h == origin)
1885	>	break restart;      // cannot find stealer
1886	>	}
1887	>	}
1888	>	for (;;) { // help stealer or descend to its stealer
1889	>	ForkJoinTask[] a;  int b;
1890	>	if (subtask.status < 0)     // surround probes with
1891	>	continue restart;       //   consistency checks
1892	>	if ((b = v.base) - v.top < 0 && (a = v.array) != null) {
1893	>	int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1894	>	ForkJoinTask<?> t =
1895	>	(ForkJoinTask<?>)U.getObjectVolatile(a, i);
1896	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1897	>	v.currentSteal != subtask)
1898	>	continue restart;   // stale
1899	>	stat = 1;               // apparent progress
1900	>	if (t != null && v.base == b &&
1901	>	U.compareAndSwapObject(a, i, t, null)) {
1902	>	v.base = b + 1;     // help stealer
1903	>	joiner.runSubtask(t);
1904	>	}
1905	>	else if (v.base == b && ++steps == MAX_HELP)
1906	>	break restart;      // v apparently stalled
1907	>	}
1908	>	else {                      // empty -- try to descend
1909	>	ForkJoinTask<?> next = v.currentJoin;
1910	>	if (subtask.status < 0 || j.currentJoin != subtask ||
1911	>	v.currentSteal != subtask)
1912	>	continue restart;   // stale
1913	>	else if (next == null || ++steps == MAX_HELP)
1914	>	break restart;      // dead-end or maybe cyclic
1915	>	else {
1916	>	subtask = next;
1917	>	j = v;
1918	>	break;
1919	>	}
1920		}
1921		}
1922	<	if (stealer == null)
1922	>	}
1923	>	}
1924	>	}
1925	>	return stat;
1926	>	}
1927	>
1928	>	/**
1929	>	* Analog of tryHelpStealer for CountedCompleters. Tries to steal
1930	>	* and run tasks within the target's computation.
1931	>	*
1932	>	* @param task the task to join
1933	>	* @param mode if shared, exit upon completing any task
1934	>	* if all workers are active
1935	>	*/
1936	>	private int helpComplete(ForkJoinTask<?> task, int mode) {
1937	>	WorkQueue[] ws; WorkQueue q; int m, n, s, u;
1938	>	if (task != null && (ws = workQueues) != null &&
1939	>	(m = ws.length - 1) >= 0) {
1940	>	for (int j = 1, origin = j;;) {
1941	>	if ((s = task.status) < 0)
1942	>	return s;
1943	>	if ((q = ws[j & m]) != null && q.pollAndExecCC(task)) {
1944	>	origin = j;
1945	>	if (mode == SHARED_QUEUE &&
1946	>	((u = (int)(ctl >>> 32)) >= 0 || (u >> UAC_SHIFT) >= 0))
1947		break;
1948		}
1949	+	else if ((j = (j + 2) & m) == origin)
1950	+	break;
1951	+	}
1952	+	}
1953	+	return 0;
1954	+	}
1955
1956	<	for (WorkQueue q = stealer;;) { // Try to help stealer
1957	<	ForkJoinTask<?> t; int b;
1958	<	if (task.status < 0)
1959	<	break outer;
1960	<	if ((b = q.base) - q.top < 0) {
1961	<	progress = true;
1962	<	if (subtask.status < 0)
1963	<	break outer;               // stale
1964	<	if ((t = q.pollAt(b)) != null) {
1965	<	stealer.stealHint = joiner.poolIndex;
1966	<	joiner.runSubtask(t);
1956	>	/**
1957	>	* Tries to decrement active count (sometimes implicitly) and
1958	>	* possibly release or create a compensating worker in preparation
1959	>	* for blocking. Fails on contention or termination. Otherwise,
1960	>	* adds a new thread if no idle workers are available and pool
1961	>	* may become starved.
1962	>	*/
1963	>	final boolean tryCompensate() {
1964	>	int pc = config & SMASK, e, i, tc; long c;
1965	>	WorkQueue[] ws; WorkQueue w; Thread p;
1966	>	if ((ws = workQueues) != null && (e = (int)(c = ctl)) >= 0) {
1967	>	if (e != 0 && (i = e & SMASK) < ws.length &&
1968	>	(w = ws[i]) != null && w.eventCount == (e | INT_SIGN)) {
1969	>	long nc = ((long)(w.nextWait & E_MASK) |
1970	>	(c & (AC_MASK|TC_MASK)));
1971	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1972	>	w.eventCount = (e + E_SEQ) & E_MASK;
1973	>	if ((p = w.parker) != null)
1974	>	U.unpark(p);
1975	>	return true;   // replace with idle worker
1976	>	}
1977	>	}
1978	>	else if ((tc = (short)(c >>> TC_SHIFT)) >= 0 &&
1979	>	(int)(c >> AC_SHIFT) + pc > 1) {
1980	>	long nc = ((c - AC_UNIT) & AC_MASK) | (c & ~AC_MASK);
1981	>	if (U.compareAndSwapLong(this, CTL, c, nc))
1982	>	return true;   // no compensation
1983	>	}
1984	>	else if (tc + pc < MAX_CAP) {
1985	>	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
1986	>	if (U.compareAndSwapLong(this, CTL, c, nc)) {
1987	>	ForkJoinWorkerThreadFactory fac;
1988	>	Throwable ex = null;
1989	>	ForkJoinWorkerThread wt = null;
1990	>	try {
1991	>	if ((fac = factory) != null &&
1992	>	(wt = fac.newThread(this)) != null) {
1993	>	wt.start();
1994	>	return true;
1995		}
1996	+	} catch (Throwable rex) {
1997	+	ex = rex;
1998		}
1999	<	else { // empty - try to descend to find stealer's stealer
1646	<	ForkJoinTask<?> next = stealer.currentJoin;
1647	<	if (++depth == MAX_HELP_DEPTH || subtask.status < 0 ||
1648	<	next == null || next == subtask)
1649	<	break outer;  // max depth, stale, dead-end, cyclic
1650	<	subtask = next;
1651	<	j = stealer;
1652	<	break;
1653	<	}
1999	>	deregisterWorker(wt, ex); // clean up and return false
2000		}
2001		}
2002		}
2003	<	return progress;
2003	>	return false;
2004		}
2005
2006		/**
2007	<	* If task is at base of some steal queue, steals and executes it.
2007	>	* Helps and/or blocks until the given task is done.
2008		*
2009		* @param joiner the joining worker
2010		* @param task the task
2011	+	* @return task status on exit
2012		*/
2013	<	final void tryPollForAndExec(WorkQueue joiner, ForkJoinTask<?> task) {
2014	<	WorkQueue[] ws;
2015	<	int m = runState & SMASK;
2016	<	if ((ws = workQueues) != null && ws.length > m) {
2017	<	for (int j = 1; j <= m && task.status >= 0; j += 2) {
2018	<	WorkQueue q = ws[j];
2019	<	if (q != null && q.pollFor(task)) {
2020	<	joiner.runSubtask(task);
2021	<	break;
2013	>	final int awaitJoin(WorkQueue joiner, ForkJoinTask<?> task) {
2014	>	int s = 0;
2015	>	if (joiner != null && task != null && (s = task.status) >= 0) {
2016	>	ForkJoinTask<?> prevJoin = joiner.currentJoin;
2017	>	joiner.currentJoin = task;
2018	>	do {} while ((s = task.status) >= 0 && !joiner.isEmpty() &&
2019	>	joiner.tryRemoveAndExec(task)); // process local tasks
2020	>	if (s >= 0 && (s = task.status) >= 0) {
2021	>	helpSignal(task, joiner.poolIndex);
2022	>	if ((s = task.status) >= 0 &&
2023	>	(task instanceof CountedCompleter))
2024	>	s = helpComplete(task, LIFO_QUEUE);
2025	>	}
2026	>	while (s >= 0 && (s = task.status) >= 0) {
2027	>	if ((!joiner.isEmpty() ||           // try helping
2028	>	(s = tryHelpStealer(joiner, task)) == 0) &&
2029	>	(s = task.status) >= 0) {
2030	>	helpSignal(task, joiner.poolIndex);
2031	>	if ((s = task.status) >= 0 && tryCompensate()) {
2032	>	if (task.trySetSignal() && (s = task.status) >= 0) {
2033	>	synchronized (task) {
2034	>	if (task.status >= 0) {
2035	>	try {                // see ForkJoinTask
2036	>	task.wait();     //  for explanation
2037	>	} catch (InterruptedException ie) {
2038	>	}
2039	>	}
2040	>	else
2041	>	task.notifyAll();
2042	>	}
2043	>	}
2044	>	long c;                          // re-activate
2045	>	do {} while (!U.compareAndSwapLong
2046	>	(this, CTL, c = ctl, c + AC_UNIT));
2047	>	}
2048		}
2049		}
2050	+	joiner.currentJoin = prevJoin;
2051	+	}
2052	+	return s;
2053	+	}
2054	+
2055	+	/**
2056	+	* Stripped-down variant of awaitJoin used by timed joins. Tries
2057	+	* to help join only while there is continuous progress. (Caller
2058	+	* will then enter a timed wait.)
2059	+	*
2060	+	* @param joiner the joining worker
2061	+	* @param task the task
2062	+	*/
2063	+	final void helpJoinOnce(WorkQueue joiner, ForkJoinTask<?> task) {
2064	+	int s;
2065	+	if (joiner != null && task != null && (s = task.status) >= 0) {
2066	+	ForkJoinTask<?> prevJoin = joiner.currentJoin;
2067	+	joiner.currentJoin = task;
2068	+	do {} while ((s = task.status) >= 0 && !joiner.isEmpty() &&
2069	+	joiner.tryRemoveAndExec(task));
2070	+	if (s >= 0 && (s = task.status) >= 0) {
2071	+	helpSignal(task, joiner.poolIndex);
2072	+	if ((s = task.status) >= 0 &&
2073	+	(task instanceof CountedCompleter))
2074	+	s = helpComplete(task, LIFO_QUEUE);
2075	+	}
2076	+	if (s >= 0 && joiner.isEmpty()) {
2077	+	do {} while (task.status >= 0 &&
2078	+	tryHelpStealer(joiner, task) > 0);
2079	+	}
2080	+	joiner.currentJoin = prevJoin;
2081		}
2082		}
2083
2084		/**
2085	<	* Returns a non-empty steal queue, if one is found during a random,
2086	<	* then cyclic scan, else null.  This method must be retried by
2085	>	* Returns a (probably) non-empty steal queue, if one is found
2086	>	* during a scan, else null.  This method must be retried by
2087		* caller if, by the time it tries to use the queue, it is empty.
2088	+	* @param r a (random) seed for scanning
2089		*/
2090	<	private WorkQueue findNonEmptyStealQueue(WorkQueue w) {
2091	<	int r = w.seed;    // Same idea as scan(), but ignoring submissions
2092	<	for (WorkQueue[] ws;;) {
2093	<	int m = runState & SMASK;
2094	<	if ((ws = workQueues) == null)
2095	<	return null;
2096	<	if (ws.length > m) {
1692	<	WorkQueue q;
1693	<	for (int n = m << 2, k = r, j = -n;;) {
1694	<	r ^= r << 13; r ^= r >>> 17; r ^= r << 5;
1695	<	if ((q = ws[(k | 1) & m]) != null && q.base - q.top < 0) {
1696	<	w.seed = r;
2090	>	private WorkQueue findNonEmptyStealQueue(int r) {
2091	>	for (;;) {
2092	>	int ps = plock, m; WorkQueue[] ws; WorkQueue q;
2093	>	if ((ws = workQueues) != null && (m = ws.length - 1) >= 0) {
2094	>	for (int j = (m + 1) << 2; j >= 0; --j) {
2095	>	if ((q = ws[(((r + j) << 1) | 1) & m]) != null &&
2096	>	q.base - q.top < 0)
2097		return q;
1698	–	}
1699	–	else if (j > n)
1700	–	return null;
1701	–	else
1702	–	k = (j++ < 0) ? r : k + ((m >>> 1) | 1);
1703	–
2098		}
2099		}
2100	+	if (plock == ps)
2101	+	return null;
2102		}
2103		}
2104
#	Line 1714 | Line 2110 | public class ForkJoinPool extends Abstra
2110		*/
2111		final void helpQuiescePool(WorkQueue w) {
2112		for (boolean active = true;;) {
2113	<	w.runLocalTasks();      // exhaust local queue
2114	<	WorkQueue q = findNonEmptyStealQueue(w);
2115	<	if (q != null) {
2116	<	ForkJoinTask<?> t;
2113	>	long c; WorkQueue q; ForkJoinTask<?> t; int b;
2114	>	while ((t = w.nextLocalTask()) != null) {
2115	>	if (w.base - w.top < 0)
2116	>	signalWork(w);
2117	>	t.doExec();
2118	>	}
2119	>	if ((q = findNonEmptyStealQueue(w.nextSeed())) != null) {
2120		if (!active) {      // re-establish active count
1722	–	long c;
2121		active = true;
2122		do {} while (!U.compareAndSwapLong
2123		(this, CTL, c = ctl, c + AC_UNIT));
2124		}
2125	<	if ((t = q.poll()) != null)
2125	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null) {
2126	>	if (q.base - q.top < 0)
2127	>	signalWork(q);
2128		w.runSubtask(t);
2129	+	}
2130		}
2131	<	else {
2132	<	long c;
2133	<	if (active) {       // decrement active count without queuing
2131	>	else if (active) {       // decrement active count without queuing
2132	>	long nc = (c = ctl) - AC_UNIT;
2133	>	if ((int)(nc >> AC_SHIFT) + (config & SMASK) == 0)
2134	>	return;          // bypass decrement-then-increment
2135	>	if (U.compareAndSwapLong(this, CTL, c, nc))
2136		active = false;
1734	–	do {} while (!U.compareAndSwapLong
1735	–	(this, CTL, c = ctl, c -= AC_UNIT));
1736	–	}
1737	–	else
1738	–	c = ctl;        // re-increment on exit
1739	–	if ((int)(c >> AC_SHIFT) + parallelism == 0) {
1740	–	do {} while (!U.compareAndSwapLong
1741	–	(this, CTL, c = ctl, c + AC_UNIT));
1742	–	break;
1743	–	}
2137		}
2138	+	else if ((int)((c = ctl) >> AC_SHIFT) + (config & SMASK) == 0 &&
2139	+	U.compareAndSwapLong(this, CTL, c, c + AC_UNIT))
2140	+	return;
2141		}
2142		}
2143
#	Line 1752 | Line 2148 | public class ForkJoinPool extends Abstra
2148		*/
2149		final ForkJoinTask<?> nextTaskFor(WorkQueue w) {
2150		for (ForkJoinTask<?> t;;) {
2151	<	WorkQueue q;
2151	>	WorkQueue q; int b;
2152		if ((t = w.nextLocalTask()) != null)
2153		return t;
2154	<	if ((q = findNonEmptyStealQueue(w)) == null)
2154	>	if ((q = findNonEmptyStealQueue(w.nextSeed())) == null)
2155		return null;
2156	<	if ((t = q.poll()) != null)
2156	>	if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null) {
2157	>	if (q.base - q.top < 0)
2158	>	signalWork(q);
2159		return t;
2160	+	}
2161		}
2162		}
2163
2164		/**
2165	<	* Returns the approximate (non-atomic) number of idle threads per
2166	<	* active thread to offset steal queue size for method
2167	<	* ForkJoinTask.getSurplusQueuedTaskCount().
2168	<	*/
2169	<	final int idlePerActive() {
2170	<	// Approximate at powers of two for small values, saturate past 4
2171	<	int p = parallelism;
2172	<	int a = p + (int)(ctl >> AC_SHIFT);
2173	<	return (a > (p >>>= 1) ? 0 :
2174	<	a > (p >>>= 1) ? 1 :
2175	<	a > (p >>>= 1) ? 2 :
2176	<	a > (p >>>= 1) ? 4 :
2177	<	8);
2178	<	}
2179	<
2180	<	// Termination
2181	<
2182	<	/**
2183	<	* Sets SHUTDOWN bit of runState under lock
2184	<	*/
2185	<	private void enableShutdown() {
2186	<	ReentrantLock lock = this.lock;
2187	<	if (runState >= 0) {
2188	<	lock.lock();                       // don't need try/finally
2189	<	runState |= SHUTDOWN;
2190	<	lock.unlock();
2165	>	* Returns a cheap heuristic guide for task partitioning when
2166	>	* programmers, frameworks, tools, or languages have little or no
2167	>	* idea about task granularity.  In essence by offering this
2168	>	* method, we ask users only about tradeoffs in overhead vs
2169	>	* expected throughput and its variance, rather than how finely to
2170	>	* partition tasks.
2171	>	*
2172	>	* In a steady state strict (tree-structured) computation, each
2173	>	* thread makes available for stealing enough tasks for other
2174	>	* threads to remain active. Inductively, if all threads play by
2175	>	* the same rules, each thread should make available only a
2176	>	* constant number of tasks.
2177	>	*
2178	>	* The minimum useful constant is just 1. But using a value of 1
2179	>	* would require immediate replenishment upon each steal to
2180	>	* maintain enough tasks, which is infeasible.  Further,
2181	>	* partitionings/granularities of offered tasks should minimize
2182	>	* steal rates, which in general means that threads nearer the top
2183	>	* of computation tree should generate more than those nearer the
2184	>	* bottom. In perfect steady state, each thread is at
2185	>	* approximately the same level of computation tree. However,
2186	>	* producing extra tasks amortizes the uncertainty of progress and
2187	>	* diffusion assumptions.
2188	>	*
2189	>	* So, users will want to use values larger, but not much larger
2190	>	* than 1 to both smooth over transient shortages and hedge
2191	>	* against uneven progress; as traded off against the cost of
2192	>	* extra task overhead. We leave the user to pick a threshold
2193	>	* value to compare with the results of this call to guide
2194	>	* decisions, but recommend values such as 3.
2195	>	*
2196	>	* When all threads are active, it is on average OK to estimate
2197	>	* surplus strictly locally. In steady-state, if one thread is
2198	>	* maintaining say 2 surplus tasks, then so are others. So we can
2199	>	* just use estimated queue length.  However, this strategy alone
2200	>	* leads to serious mis-estimates in some non-steady-state
2201	>	* conditions (ramp-up, ramp-down, other stalls). We can detect
2202	>	* many of these by further considering the number of "idle"
2203	>	* threads, that are known to have zero queued tasks, so
2204	>	* compensate by a factor of (#idle/#active) threads.
2205	>	*
2206	>	* Note: The approximation of #busy workers as #active workers is
2207	>	* not very good under current signalling scheme, and should be
2208	>	* improved.
2209	>	*/
2210	>	static int getSurplusQueuedTaskCount() {
2211	>	Thread t; ForkJoinWorkerThread wt; ForkJoinPool pool; WorkQueue q;
2212	>	if (((t = Thread.currentThread()) instanceof ForkJoinWorkerThread)) {
2213	>	int p = (pool = (wt = (ForkJoinWorkerThread)t).pool).config & SMASK;
2214	>	int n = (q = wt.workQueue).top - q.base;
2215	>	int a = (int)(pool.ctl >> AC_SHIFT) + p;
2216	>	return n - (a > (p >>>= 1) ? 0 :
2217	>	a > (p >>>= 1) ? 1 :
2218	>	a > (p >>>= 1) ? 2 :
2219	>	a > (p >>>= 1) ? 4 :
2220	>	8);
2221		}
2222	+	return 0;
2223		}
2224
2225	+	//  Termination
2226	+
2227		/**
2228	<	* Possibly initiates and/or completes termination.  Upon
2229	<	* termination, cancels all queued tasks and then
2228	>	* Possibly initiates and/or completes termination.  The caller
2229	>	* triggering termination runs three passes through workQueues:
2230	>	* (0) Setting termination status, followed by wakeups of queued
2231	>	* workers; (1) cancelling all tasks; (2) interrupting lagging
2232	>	* threads (likely in external tasks, but possibly also blocked in
2233	>	* joins).  Each pass repeats previous steps because of potential
2234	>	* lagging thread creation.
2235		*
2236		* @param now if true, unconditionally terminate, else only
2237		* if no work and no active workers
2238	+	* @param enable if true, enable shutdown when next possible
2239		* @return true if now terminating or terminated
2240		*/
2241	<	private boolean tryTerminate(boolean now) {
2241	>	private boolean tryTerminate(boolean now, boolean enable) {
2242	>	int ps;
2243	>	if (this == common)                    // cannot shut down
2244	>	return false;
2245	>	if ((ps = plock) >= 0) {                   // enable by setting plock
2246	>	if (!enable)
2247	>	return false;
2248	>	if ((ps & PL_LOCK) != 0 ||
2249	>	!U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
2250	>	ps = acquirePlock();
2251	>	int nps = ((ps + PL_LOCK) & ~SHUTDOWN) | SHUTDOWN;
2252	>	if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
2253	>	releasePlock(nps);
2254	>	}
2255		for (long c;;) {
2256	<	if (((c = ctl) & STOP_BIT) != 0) {      // already terminating
2257	<	if ((short)(c >>> TC_SHIFT) == -parallelism) {
2258	<	ReentrantLock lock = this.lock; // signal when no workers
2259	<	lock.lock();                    // don't need try/finally
2260	<	termination.signalAll();        // signal when 0 workers
1810	<	lock.unlock();
2256	>	if (((c = ctl) & STOP_BIT) != 0) {     // already terminating
2257	>	if ((short)(c >>> TC_SHIFT) == -(config & SMASK)) {
2258	>	synchronized (this) {
2259	>	notifyAll();               // signal when 0 workers
2260	>	}
2261		}
2262		return true;
2263		}
2264	<	if (!now) {
2265	<	if ((int)(c >> AC_SHIFT) != -parallelism || runState >= 0 ||
2266	<	hasQueuedSubmissions())
2264	>	if (!now) {                            // check if idle & no tasks
2265	>	WorkQueue[] ws; WorkQueue w;
2266	>	if ((int)(c >> AC_SHIFT) != -(config & SMASK))
2267		return false;
2268	<	// Check for unqueued inactive workers. One pass suffices.
2269	<	WorkQueue[] ws = workQueues; WorkQueue w;
2270	<	if (ws != null) {
2271	<	int n = ws.length;
2272	<	for (int i = 1; i < n; i += 2) {
2273	<	if ((w = ws[i]) != null && w.eventCount >= 0)
2274	<	return false;
2268	>	if ((ws = workQueues) != null) {
2269	>	for (int i = 0; i < ws.length; ++i) {
2270	>	if ((w = ws[i]) != null) {
2271	>	if (!w.isEmpty()) {    // signal unprocessed tasks
2272	>	signalWork(w);
2273	>	return false;
2274	>	}
2275	>	if ((i & 1) != 0 && w.eventCount >= 0)
2276	>	return false;      // unqueued inactive worker
2277	>	}
2278	>	}
2279	>	}
2280	>	}
2281	>	if (U.compareAndSwapLong(this, CTL, c, c | STOP_BIT)) {
2282	>	for (int pass = 0; pass < 3; ++pass) {
2283	>	WorkQueue[] ws; WorkQueue w; Thread wt;
2284	>	if ((ws = workQueues) != null) {
2285	>	int n = ws.length;
2286	>	for (int i = 0; i < n; ++i) {
2287	>	if ((w = ws[i]) != null) {
2288	>	w.qlock = -1;
2289	>	if (pass > 0) {
2290	>	w.cancelAll();
2291	>	if (pass > 1 && (wt = w.owner) != null) {
2292	>	if (!wt.isInterrupted()) {
2293	>	try {
2294	>	wt.interrupt();
2295	>	} catch (Throwable ignore) {
2296	>	}
2297	>	}
2298	>	U.unpark(wt);
2299	>	}
2300	>	}
2301	>	}
2302	>	}
2303	>	// Wake up workers parked on event queue
2304	>	int i, e; long cc; Thread p;
2305	>	while ((e = (int)(cc = ctl) & E_MASK) != 0 &&
2306	>	(i = e & SMASK) < n && i >= 0 &&
2307	>	(w = ws[i]) != null) {
2308	>	long nc = ((long)(w.nextWait & E_MASK) |
2309	>	((cc + AC_UNIT) & AC_MASK) |
2310	>	(cc & (TC_MASK|STOP_BIT)));
2311	>	if (w.eventCount == (e | INT_SIGN) &&
2312	>	U.compareAndSwapLong(this, CTL, cc, nc)) {
2313	>	w.eventCount = (e + E_SEQ) & E_MASK;
2314	>	w.qlock = -1;
2315	>	if ((p = w.parker) != null)
2316	>	U.unpark(p);
2317	>	}
2318	>	}
2319		}
2320		}
2321		}
1828	–	if (U.compareAndSwapLong(this, CTL, c, c | STOP_BIT))
1829	–	startTerminating();
2322		}
2323		}
2324
2325	+	// external operations on common pool
2326	+
2327		/**
2328	<	* Initiates termination: Runs three passes through workQueues:
2329	<	* (0) Setting termination status, followed by wakeups of queued
1836	<	* workers; (1) cancelling all tasks; (2) interrupting lagging
1837	<	* threads (likely in external tasks, but possibly also blocked in
1838	<	* joins).  Each pass repeats previous steps because of potential
1839	<	* lagging thread creation.
2328	>	* Returns common pool queue for a thread that has submitted at
2329	>	* least one task.
2330		*/
2331	<	private void startTerminating() {
2332	<	for (int pass = 0; pass < 3; ++pass) {
2333	<	WorkQueue[] ws = workQueues;
2334	<	if (ws != null) {
2335	<	WorkQueue w; Thread wt;
2336	<	int n = ws.length;
2337	<	for (int j = 0; j < n; ++j) {
2338	<	if ((w = ws[j]) != null) {
2339	<	w.runState = -1;
2340	<	if (pass > 0) {
2341	<	w.cancelAll();
2342	<	if (pass > 1 && (wt = w.owner) != null &&
2343	<	!wt.isInterrupted()) {
2344	<	try {
2345	<	wt.interrupt();
2346	<	} catch (SecurityException ignore) {
2331	>	static WorkQueue commonSubmitterQueue() {
2332	>	ForkJoinPool p; WorkQueue[] ws; int m; Submitter z;
2333	>	return ((z = submitters.get()) != null &&
2334	>	(p = common) != null &&
2335	>	(ws = p.workQueues) != null &&
2336	>	(m = ws.length - 1) >= 0) ?
2337	>	ws[m & z.seed & SQMASK] : null;
2338	>	}
2339	>
2340	>	/**
2341	>	* Tries to pop the given task from submitter's queue in common pool.
2342	>	*/
2343	>	static boolean tryExternalUnpush(ForkJoinTask<?> t) {
2344	>	ForkJoinPool p; WorkQueue[] ws; WorkQueue q; Submitter z;
2345	>	ForkJoinTask<?>[] a;  int m, s;
2346	>	if (t != null &&
2347	>	(z = submitters.get()) != null &&
2348	>	(p = common) != null &&
2349	>	(ws = p.workQueues) != null &&
2350	>	(m = ws.length - 1) >= 0 &&
2351	>	(q = ws[m & z.seed & SQMASK]) != null &&
2352	>	(s = q.top) != q.base &&
2353	>	(a = q.array) != null) {
2354	>	long j = (((a.length - 1) & (s - 1)) << ASHIFT) + ABASE;
2355	>	if (U.getObject(a, j) == t &&
2356	>	U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2357	>	if (q.array == a && q.top == s && // recheck
2358	>	U.compareAndSwapObject(a, j, t, null)) {
2359	>	q.top = s - 1;
2360	>	q.qlock = 0;
2361	>	return true;
2362	>	}
2363	>	q.qlock = 0;
2364	>	}
2365	>	}
2366	>	return false;
2367	>	}
2368	>
2369	>	/**
2370	>	* Tries to pop and run local tasks within the same computation
2371	>	* as the given root. On failure, tries to help complete from
2372	>	* other queues via helpComplete.
2373	>	*/
2374	>	private void externalHelpComplete(WorkQueue q, ForkJoinTask<?> root) {
2375	>	ForkJoinTask<?>[] a; int m;
2376	>	if (q != null && (a = q.array) != null && (m = (a.length - 1)) >= 0 &&
2377	>	root != null && root.status >= 0) {
2378	>	for (;;) {
2379	>	int s, u; Object o; CountedCompleter<?> task = null;
2380	>	if ((s = q.top) - q.base > 0) {
2381	>	long j = ((m & (s - 1)) << ASHIFT) + ABASE;
2382	>	if ((o = U.getObject(a, j)) != null &&
2383	>	(o instanceof CountedCompleter)) {
2384	>	CountedCompleter<?> t = (CountedCompleter<?>)o, r = t;
2385	>	do {
2386	>	if (r == root) {
2387	>	if (U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2388	>	if (q.array == a && q.top == s &&
2389	>	U.compareAndSwapObject(a, j, t, null)) {
2390	>	q.top = s - 1;
2391	>	task = t;
2392	>	}
2393	>	q.qlock = 0;
2394		}
2395	+	break;
2396		}
2397	<	}
2397	>	} while ((r = r.completer) != null);
2398		}
2399		}
2400	<	// Wake up workers parked on event queue
2401	<	int i, e; long c; Thread p;
2402	<	while ((i = ((~(e = (int)(c = ctl)) << 1) | 1) & SMASK) < n &&
2403	<	(w = ws[i]) != null &&
2404	<	w.eventCount == (e | INT_SIGN)) {
2405	<	long nc = ((long)(w.nextWait & E_MASK) |
2406	<	((c + AC_UNIT) & AC_MASK) |
2407	<	(c & (TC_MASK|STOP_BIT)));
2408	<	if (U.compareAndSwapLong(this, CTL, c, nc)) {
2409	<	w.eventCount = (e + E_SEQ) & E_MASK;
2410	<	if ((p = w.parker) != null)
2411	<	U.unpark(p);
2400	>	if (task != null)
2401	>	task.doExec();
2402	>	if (root.status < 0 ||
2403	>	(u = (int)(ctl >>> 32)) >= 0 || (u >> UAC_SHIFT) >= 0)
2404	>	break;
2405	>	if (task == null) {
2406	>	helpSignal(root, q.poolIndex);
2407	>	if (root.status >= 0)
2408	>	helpComplete(root, SHARED_QUEUE);
2409	>	break;
2410	>	}
2411	>	}
2412	>	}
2413	>	}
2414	>
2415	>	/**
2416	>	* Tries to help execute or signal availability of the given task
2417	>	* from submitter's queue in common pool.
2418	>	*/
2419	>	static void externalHelpJoin(ForkJoinTask<?> t) {
2420	>	// Some hard-to-avoid overlap with tryExternalUnpush
2421	>	ForkJoinPool p; WorkQueue[] ws; WorkQueue q, w; Submitter z;
2422	>	ForkJoinTask<?>[] a;  int m, s, n;
2423	>	if (t != null &&
2424	>	(z = submitters.get()) != null &&
2425	>	(p = common) != null &&
2426	>	(ws = p.workQueues) != null &&
2427	>	(m = ws.length - 1) >= 0 &&
2428	>	(q = ws[m & z.seed & SQMASK]) != null &&
2429	>	(a = q.array) != null) {
2430	>	int am = a.length - 1;
2431	>	if ((s = q.top) != q.base) {
2432	>	long j = ((am & (s - 1)) << ASHIFT) + ABASE;
2433	>	if (U.getObject(a, j) == t &&
2434	>	U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2435	>	if (q.array == a && q.top == s &&
2436	>	U.compareAndSwapObject(a, j, t, null)) {
2437	>	q.top = s - 1;
2438	>	q.qlock = 0;
2439	>	t.doExec();
2440		}
2441	+	else
2442	+	q.qlock = 0;
2443		}
2444		}
2445	+	if (t.status >= 0) {
2446	+	if (t instanceof CountedCompleter)
2447	+	p.externalHelpComplete(q, t);
2448	+	else
2449	+	p.helpSignal(t, q.poolIndex);
2450	+	}
2451		}
2452		}
2453
#	Line 1946 | Line 2520 | public class ForkJoinPool extends Abstra
2520		checkPermission();
2521		if (factory == null)
2522		throw new NullPointerException();
2523	<	if (parallelism <= 0 || parallelism > MAX_ID)
2523	>	if (parallelism <= 0 || parallelism > MAX_CAP)
2524		throw new IllegalArgumentException();
1951	–	this.parallelism = parallelism;
2525		this.factory = factory;
2526		this.ueh = handler;
2527	<	this.localMode = asyncMode ? FIFO_QUEUE : LIFO_QUEUE;
1955	<	this.nextPoolIndex = 1;
2527	>	this.config = parallelism | (asyncMode ? (FIFO_QUEUE << 16) : 0);
2528		long np = (long)(-parallelism); // offset ctl counts
2529		this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2530	<	// initialize workQueues array with room for 2*parallelism if possible
1959	<	int n = parallelism << 1;
1960	<	if (n >= MAX_ID)
1961	<	n = MAX_ID;
1962	<	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();
2530	>	int pn = nextPoolId();
2531		StringBuilder sb = new StringBuilder("ForkJoinPool-");
2532	<	sb.append(poolNumberGenerator.incrementAndGet());
2532	>	sb.append(Integer.toString(pn));
2533		sb.append("-worker-");
2534		this.workerNamePrefix = sb.toString();
2535	<	// Create initial submission queue
2536	<	WorkQueue sq = tryAddSharedQueue(0);
2537	<	if (sq != null)
2538	<	sq.growArray(false);
2535	>	}
2536	>
2537	>	/**
2538	>	* Constructor for common pool, suitable only for static initialization.
2539	>	* Basically the same as above, but uses smallest possible initial footprint.
2540	>	*/
2541	>	ForkJoinPool(int parallelism, long ctl,
2542	>	ForkJoinWorkerThreadFactory factory,
2543	>	Thread.UncaughtExceptionHandler handler) {
2544	>	this.config = parallelism;
2545	>	this.ctl = ctl;
2546	>	this.factory = factory;
2547	>	this.ueh = handler;
2548	>	this.workerNamePrefix = "ForkJoinPool.commonPool-worker-";
2549	>	}
2550	>
2551	>	/**
2552	>	* Returns the common pool instance. This pool is statically
2553	>	* constructed; its run state is unaffected by attempts to {@link
2554	>	* #shutdown} or {@link #shutdownNow}. However this pool and any
2555	>	* ongoing processing are automatically terminated upon program
2556	>	* {@link System#exit}.  Any program that relies on asynchronous
2557	>	* task processing to complete before program termination should
2558	>	* invoke {@code commonPool().}{@link #awaitQuiescence}, before
2559	>	* exit.
2560	>	*
2561	>	* @return the common pool instance
2562	>	* @since 1.8
2563	>	*/
2564	>	public static ForkJoinPool commonPool() {
2565	>	// assert common != null : "static init error";
2566	>	return common;
2567		}
2568
2569		// Execution methods
#	Line 1996 | Line 2585 | public class ForkJoinPool extends Abstra
2585		*         scheduled for execution
2586		*/
2587		public <T> T invoke(ForkJoinTask<T> task) {
2588	<	doSubmit(task);
2588	>	if (task == null)
2589	>	throw new NullPointerException();
2590	>	externalPush(task);
2591		return task.join();
2592		}
2593
#	Line 2009 | Line 2600 | public class ForkJoinPool extends Abstra
2600		*         scheduled for execution
2601		*/
2602		public void execute(ForkJoinTask<?> task) {
2603	<	doSubmit(task);
2603	>	if (task == null)
2604	>	throw new NullPointerException();
2605	>	externalPush(task);
2606		}
2607
2608		// AbstractExecutorService methods
#	Line 2026 | Line 2619 | public class ForkJoinPool extends Abstra
2619		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2620		job = (ForkJoinTask<?>) task;
2621		else
2622	<	job = ForkJoinTask.adapt(task, null);
2623	<	doSubmit(job);
2622	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2623	>	externalPush(job);
2624		}
2625
2626		/**
#	Line 2040 | Line 2633 | public class ForkJoinPool extends Abstra
2633		*         scheduled for execution
2634		*/
2635		public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
2636	<	doSubmit(task);
2636	>	if (task == null)
2637	>	throw new NullPointerException();
2638	>	externalPush(task);
2639		return task;
2640		}
2641
#	Line 2050 | Line 2645 | public class ForkJoinPool extends Abstra
2645		*         scheduled for execution
2646		*/
2647		public <T> ForkJoinTask<T> submit(Callable<T> task) {
2648	<	if (task == null)
2649	<	throw new NullPointerException();
2055	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task);
2056	<	doSubmit(job);
2648	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedCallable<T>(task);
2649	>	externalPush(job);
2650		return job;
2651		}
2652
#	Line 2063 | Line 2656 | public class ForkJoinPool extends Abstra
2656		*         scheduled for execution
2657		*/
2658		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
2659	<	if (task == null)
2660	<	throw new NullPointerException();
2068	<	ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
2069	<	doSubmit(job);
2659	>	ForkJoinTask<T> job = new ForkJoinTask.AdaptedRunnable<T>(task, result);
2660	>	externalPush(job);
2661		return job;
2662		}
2663
#	Line 2082 | Line 2673 | public class ForkJoinPool extends Abstra
2673		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2674		job = (ForkJoinTask<?>) task;
2675		else
2676	<	job = ForkJoinTask.adapt(task, null);
2677	<	doSubmit(job);
2676	>	job = new ForkJoinTask.AdaptedRunnableAction(task);
2677	>	externalPush(job);
2678		return job;
2679		}
2680
#	Line 2092 | Line 2683 | public class ForkJoinPool extends Abstra
2683		* @throws RejectedExecutionException {@inheritDoc}
2684		*/
2685		public <T> List<Future<T>> invokeAll(Collection<? extends Callable<T>> tasks) {
2686	<	ArrayList<ForkJoinTask<T>> forkJoinTasks =
2687	<	new ArrayList<ForkJoinTask<T>>(tasks.size());
2688	<	for (Callable<T> task : tasks)
2689	<	forkJoinTasks.add(ForkJoinTask.adapt(task));
2690	<	invoke(new InvokeAll<T>(forkJoinTasks));
2691	<
2686	>	// In previous versions of this class, this method constructed
2687	>	// a task to run ForkJoinTask.invokeAll, but now external
2688	>	// invocation of multiple tasks is at least as efficient.
2689	>	List<ForkJoinTask<T>> fs = new ArrayList<ForkJoinTask<T>>(tasks.size());
2690	>	// Workaround needed because method wasn't declared with
2691	>	// wildcards in return type but should have been.
2692		@SuppressWarnings({"unchecked", "rawtypes"})
2693	<	List<Future<T>> futures = (List<Future<T>>) (List) forkJoinTasks;
2103	<	return futures;
2104	<	}
2693	>	List<Future<T>> futures = (List<Future<T>>) (List) fs;
2694
2695	<	static final class InvokeAll<T> extends RecursiveAction {
2696	<	final ArrayList<ForkJoinTask<T>> tasks;
2697	<	InvokeAll(ArrayList<ForkJoinTask<T>> tasks) { this.tasks = tasks; }
2698	<	public void compute() {
2699	<	try { invokeAll(tasks); }
2700	<	catch (Exception ignore) {}
2695	>	boolean done = false;
2696	>	try {
2697	>	for (Callable<T> t : tasks) {
2698	>	ForkJoinTask<T> f = new ForkJoinTask.AdaptedCallable<T>(t);
2699	>	externalPush(f);
2700	>	fs.add(f);
2701	>	}
2702	>	for (ForkJoinTask<T> f : fs)
2703	>	f.quietlyJoin();
2704	>	done = true;
2705	>	return futures;
2706	>	} finally {
2707	>	if (!done)
2708	>	for (ForkJoinTask<T> f : fs)
2709	>	f.cancel(false);
2710		}
2113	–	private static final long serialVersionUID = -7914297376763021607L;
2711		}
2712
2713		/**
#	Line 2138 | Line 2735 | public class ForkJoinPool extends Abstra
2735		* @return the targeted parallelism level of this pool
2736		*/
2737		public int getParallelism() {
2738	<	return parallelism;
2738	>	return config & SMASK;
2739	>	}
2740	>
2741	>	/**
2742	>	* Returns the targeted parallelism level of the common pool.
2743	>	*
2744	>	* @return the targeted parallelism level of the common pool
2745	>	* @since 1.8
2746	>	*/
2747	>	public static int getCommonPoolParallelism() {
2748	>	return commonParallelism;
2749		}
2750
2751		/**
#	Line 2150 | Line 2757 | public class ForkJoinPool extends Abstra
2757		* @return the number of worker threads
2758		*/
2759		public int getPoolSize() {
2760	<	return parallelism + (short)(ctl >>> TC_SHIFT);
2760	>	return (config & SMASK) + (short)(ctl >>> TC_SHIFT);
2761		}
2762
2763		/**
#	Line 2160 | Line 2767 | public class ForkJoinPool extends Abstra
2767		* @return {@code true} if this pool uses async mode
2768		*/
2769		public boolean getAsyncMode() {
2770	<	return localMode != 0;
2770	>	return (config >>> 16) == FIFO_QUEUE;
2771		}
2772
2773		/**
#	Line 2175 | Line 2782 | public class ForkJoinPool extends Abstra
2782		int rc = 0;
2783		WorkQueue[] ws; WorkQueue w;
2784		if ((ws = workQueues) != null) {
2785	<	int n = ws.length;
2786	<	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)
2785	>	for (int i = 1; i < ws.length; i += 2) {
2786	>	if ((w = ws[i]) != null && w.isApparentlyUnblocked())
2787		++rc;
2788		}
2789		}
#	Line 2197 | Line 2798 | public class ForkJoinPool extends Abstra
2798		* @return the number of active threads
2799		*/
2800		public int getActiveThreadCount() {
2801	<	int r = parallelism + (int)(ctl >> AC_SHIFT);
2801	>	int r = (config & SMASK) + (int)(ctl >> AC_SHIFT);
2802		return (r <= 0) ? 0 : r; // suppress momentarily negative values
2803		}
2804
#	Line 2213 | Line 2814 | public class ForkJoinPool extends Abstra
2814		* @return {@code true} if all threads are currently idle
2815		*/
2816		public boolean isQuiescent() {
2817	<	return (int)(ctl >> AC_SHIFT) + parallelism == 0;
2817	>	return (int)(ctl >> AC_SHIFT) + (config & SMASK) == 0;
2818		}
2819
2820		/**
#	Line 2228 | Line 2829 | public class ForkJoinPool extends Abstra
2829		* @return the number of steals
2830		*/
2831		public long getStealCount() {
2832	<	long count = stealCount.get();
2832	>	long count = stealCount;
2833		WorkQueue[] ws; WorkQueue w;
2834		if ((ws = workQueues) != null) {
2835	<	int n = ws.length;
2235	<	for (int i = 1; i < n; i += 2) {
2835	>	for (int i = 1; i < ws.length; i += 2) {
2836		if ((w = ws[i]) != null)
2837	<	count += w.totalSteals;
2837	>	count += w.nsteals;
2838		}
2839		}
2840		return count;
#	Line 2254 | Line 2854 | public class ForkJoinPool extends Abstra
2854		long count = 0;
2855		WorkQueue[] ws; WorkQueue w;
2856		if ((ws = workQueues) != null) {
2857	<	int n = ws.length;
2258	<	for (int i = 1; i < n; i += 2) {
2857	>	for (int i = 1; i < ws.length; i += 2) {
2858		if ((w = ws[i]) != null)
2859		count += w.queueSize();
2860		}
#	Line 2274 | Line 2873 | public class ForkJoinPool extends Abstra
2873		int count = 0;
2874		WorkQueue[] ws; WorkQueue w;
2875		if ((ws = workQueues) != null) {
2876	<	int n = ws.length;
2278	<	for (int i = 0; i < n; i += 2) {
2876	>	for (int i = 0; i < ws.length; i += 2) {
2877		if ((w = ws[i]) != null)
2878		count += w.queueSize();
2879		}
#	Line 2292 | Line 2890 | public class ForkJoinPool extends Abstra
2890		public boolean hasQueuedSubmissions() {
2891		WorkQueue[] ws; WorkQueue w;
2892		if ((ws = workQueues) != null) {
2893	<	int n = ws.length;
2894	<	for (int i = 0; i < n; i += 2) {
2297	<	if ((w = ws[i]) != null && w.queueSize() != 0)
2893	>	for (int i = 0; i < ws.length; i += 2) {
2894	>	if ((w = ws[i]) != null && !w.isEmpty())
2895		return true;
2896		}
2897		}
#	Line 2311 | Line 2908 | public class ForkJoinPool extends Abstra
2908		protected ForkJoinTask<?> pollSubmission() {
2909		WorkQueue[] ws; WorkQueue w; ForkJoinTask<?> t;
2910		if ((ws = workQueues) != null) {
2911	<	int n = ws.length;
2315	<	for (int i = 0; i < n; i += 2) {
2911	>	for (int i = 0; i < ws.length; i += 2) {
2912		if ((w = ws[i]) != null && (t = w.poll()) != null)
2913		return t;
2914		}
#	Line 2341 | Line 2937 | public class ForkJoinPool extends Abstra
2937		int count = 0;
2938		WorkQueue[] ws; WorkQueue w; ForkJoinTask<?> t;
2939		if ((ws = workQueues) != null) {
2940	<	int n = ws.length;
2345	<	for (int i = 0; i < n; ++i) {
2940	>	for (int i = 0; i < ws.length; ++i) {
2941		if ((w = ws[i]) != null) {
2942		while ((t = w.poll()) != null) {
2943		c.add(t);
#	Line 2362 | Line 2957 | public class ForkJoinPool extends Abstra
2957		* @return a string identifying this pool, as well as its state
2958		*/
2959		public String toString() {
2960	<	long st = getStealCount();
2961	<	long qt = getQueuedTaskCount();
2962	<	long qs = getQueuedSubmissionCount();
2368	<	int rc = getRunningThreadCount();
2369	<	int pc = parallelism;
2960	>	// Use a single pass through workQueues to collect counts
2961	>	long qt = 0L, qs = 0L; int rc = 0;
2962	>	long st = stealCount;
2963		long c = ctl;
2964	+	WorkQueue[] ws; WorkQueue w;
2965	+	if ((ws = workQueues) != null) {
2966	+	for (int i = 0; i < ws.length; ++i) {
2967	+	if ((w = ws[i]) != null) {
2968	+	int size = w.queueSize();
2969	+	if ((i & 1) == 0)
2970	+	qs += size;
2971	+	else {
2972	+	qt += size;
2973	+	st += w.nsteals;
2974	+	if (w.isApparentlyUnblocked())
2975	+	++rc;
2976	+	}
2977	+	}
2978	+	}
2979	+	}
2980	+	int pc = (config & SMASK);
2981		int tc = pc + (short)(c >>> TC_SHIFT);
2982		int ac = pc + (int)(c >> AC_SHIFT);
2983		if (ac < 0) // ignore transient negative
#	Line 2376 | Line 2986 | public class ForkJoinPool extends Abstra
2986		if ((c & STOP_BIT) != 0)
2987		level = (tc == 0) ? "Terminated" : "Terminating";
2988		else
2989	<	level = runState < 0 ? "Shutting down" : "Running";
2989	>	level = plock < 0 ? "Shutting down" : "Running";
2990		return super.toString() +
2991		"[" + level +
2992		", parallelism = " + pc +
#	Line 2390 | Line 3000 | public class ForkJoinPool extends Abstra
3000		}
3001
3002		/**
3003	<	* Initiates an orderly shutdown in which previously submitted
3004	<	* tasks are executed, but no new tasks will be accepted.
3005	<	* Invocation has no additional effect if already shut down.
3006	<	* Tasks that are in the process of being submitted concurrently
3007	<	* during the course of this method may or may not be rejected.
3003	>	* Possibly initiates an orderly shutdown in which previously
3004	>	* submitted tasks are executed, but no new tasks will be
3005	>	* accepted. Invocation has no effect on execution state if this
3006	>	* is the {@link #commonPool()}, and no additional effect if
3007	>	* already shut down.  Tasks that are in the process of being
3008	>	* submitted concurrently during the course of this method may or
3009	>	* may not be rejected.
3010		*
3011		* @throws SecurityException if a security manager exists and
3012		*         the caller is not permitted to modify threads
#	Line 2403 | Line 3015 | public class ForkJoinPool extends Abstra
3015		*/
3016		public void shutdown() {
3017		checkPermission();
3018	<	enableShutdown();
2407	<	tryTerminate(false);
3018	>	tryTerminate(false, true);
3019		}
3020
3021		/**
3022	<	* Attempts to cancel and/or stop all tasks, and reject all
3023	<	* subsequently submitted tasks.  Tasks that are in the process of
3024	<	* being submitted or executed concurrently during the course of
3025	<	* this method may or may not be rejected. This method cancels
3026	<	* both existing and unexecuted tasks, in order to permit
3027	<	* termination in the presence of task dependencies. So the method
3028	<	* always returns an empty list (unlike the case for some other
3029	<	* Executors).
3022	>	* Possibly attempts to cancel and/or stop all tasks, and reject
3023	>	* all subsequently submitted tasks.  Invocation has no effect on
3024	>	* execution state if this is the {@link #commonPool()}, and no
3025	>	* additional effect if already shut down. Otherwise, tasks that
3026	>	* are in the process of being submitted or executed concurrently
3027	>	* during the course of this method may or may not be
3028	>	* rejected. This method cancels both existing and unexecuted
3029	>	* tasks, in order to permit termination in the presence of task
3030	>	* dependencies. So the method always returns an empty list
3031	>	* (unlike the case for some other Executors).
3032		*
3033		* @return an empty list
3034		* @throws SecurityException if a security manager exists and
#	Line 2425 | Line 3038 | public class ForkJoinPool extends Abstra
3038		*/
3039		public List<Runnable> shutdownNow() {
3040		checkPermission();
3041	<	enableShutdown();
2429	<	tryTerminate(true);
3041	>	tryTerminate(true, true);
3042		return Collections.emptyList();
3043		}
3044
#	Line 2438 | Line 3050 | public class ForkJoinPool extends Abstra
3050		public boolean isTerminated() {
3051		long c = ctl;
3052		return ((c & STOP_BIT) != 0L &&
3053	<	(short)(c >>> TC_SHIFT) == -parallelism);
3053	>	(short)(c >>> TC_SHIFT) == -(config & SMASK));
3054		}
3055
3056		/**
#	Line 2446 | Line 3058 | public class ForkJoinPool extends Abstra
3058		* commenced but not yet completed.  This method may be useful for
3059		* debugging. A return of {@code true} reported a sufficient
3060		* period after shutdown may indicate that submitted tasks have
3061	<	* ignored or suppressed interruption, or are waiting for IO,
3061	>	* ignored or suppressed interruption, or are waiting for I/O,
3062		* causing this executor not to properly terminate. (See the
3063		* advisory notes for class {@link ForkJoinTask} stating that
3064		* tasks should not normally entail blocking operations.  But if
#	Line 2457 | Line 3069 | public class ForkJoinPool extends Abstra
3069		public boolean isTerminating() {
3070		long c = ctl;
3071		return ((c & STOP_BIT) != 0L &&
3072	<	(short)(c >>> TC_SHIFT) != -parallelism);
3072	>	(short)(c >>> TC_SHIFT) != -(config & SMASK));
3073		}
3074
3075		/**
#	Line 2466 | Line 3078 | public class ForkJoinPool extends Abstra
3078		* @return {@code true} if this pool has been shut down
3079		*/
3080		public boolean isShutdown() {
3081	<	return runState < 0;
3081	>	return plock < 0;
3082		}
3083
3084		/**
3085	<	* Blocks until all tasks have completed execution after a shutdown
3086	<	* request, or the timeout occurs, or the current thread is
3087	<	* interrupted, whichever happens first.
3085	>	* Blocks until all tasks have completed execution after a
3086	>	* shutdown request, or the timeout occurs, or the current thread
3087	>	* is interrupted, whichever happens first. Because the {@link
3088	>	* #commonPool()} never terminates until program shutdown, when
3089	>	* applied to the common pool, this method is equivalent to {@link
3090	>	* #awaitQuiescence} but always returns {@code false}.
3091		*
3092		* @param timeout the maximum time to wait
3093		* @param unit the time unit of the timeout argument
#	Line 2482 | Line 3097 | public class ForkJoinPool extends Abstra
3097		*/
3098		public boolean awaitTermination(long timeout, TimeUnit unit)
3099		throws InterruptedException {
3100	+	if (Thread.interrupted())
3101	+	throw new InterruptedException();
3102	+	if (this == common) {
3103	+	awaitQuiescence(timeout, unit);
3104	+	return false;
3105	+	}
3106		long nanos = unit.toNanos(timeout);
3107	<	final ReentrantLock lock = this.lock;
3108	<	lock.lock();
3109	<	try {
3110	<	for (;;) {
3111	<	if (isTerminated())
3112	<	return true;
3113	<	if (nanos <= 0)
3107	>	if (isTerminated())
3108	>	return true;
3109	>	long startTime = System.nanoTime();
3110	>	boolean terminated = false;
3111	>	synchronized (this) {
3112	>	for (long waitTime = nanos, millis = 0L;;) {
3113	>	if (terminated = isTerminated() ||
3114	>	waitTime <= 0L ||
3115	>	(millis = unit.toMillis(waitTime)) <= 0L)
3116	>	break;
3117	>	wait(millis);
3118	>	waitTime = nanos - (System.nanoTime() - startTime);
3119	>	}
3120	>	}
3121	>	return terminated;
3122	>	}
3123	>
3124	>	/**
3125	>	* If called by a ForkJoinTask operating in this pool, equivalent
3126	>	* in effect to {@link ForkJoinTask#helpQuiesce}. Otherwise,
3127	>	* waits and/or attempts to assist performing tasks until this
3128	>	* pool {@link #isQuiescent} or the indicated timeout elapses.
3129	>	*
3130	>	* @param timeout the maximum time to wait
3131	>	* @param unit the time unit of the timeout argument
3132	>	* @return {@code true} if quiescent; {@code false} if the
3133	>	* timeout elapsed.
3134	>	*/
3135	>	public boolean awaitQuiescence(long timeout, TimeUnit unit) {
3136	>	long nanos = unit.toNanos(timeout);
3137	>	ForkJoinWorkerThread wt;
3138	>	Thread thread = Thread.currentThread();
3139	>	if ((thread instanceof ForkJoinWorkerThread) &&
3140	>	(wt = (ForkJoinWorkerThread)thread).pool == this) {
3141	>	helpQuiescePool(wt.workQueue);
3142	>	return true;
3143	>	}
3144	>	long startTime = System.nanoTime();
3145	>	WorkQueue[] ws;
3146	>	int r = 0, m;
3147	>	boolean found = true;
3148	>	while (!isQuiescent() && (ws = workQueues) != null &&
3149	>	(m = ws.length - 1) >= 0) {
3150	>	if (!found) {
3151	>	if ((System.nanoTime() - startTime) > nanos)
3152		return false;
3153	<	nanos = termination.awaitNanos(nanos);
3153	>	Thread.yield(); // cannot block
3154	>	}
3155	>	found = false;
3156	>	for (int j = (m + 1) << 2; j >= 0; --j) {
3157	>	ForkJoinTask<?> t; WorkQueue q; int b;
3158	>	if ((q = ws[r++ & m]) != null && (b = q.base) - q.top < 0) {
3159	>	found = true;
3160	>	if ((t = q.pollAt(b)) != null) {
3161	>	if (q.base - q.top < 0)
3162	>	signalWork(q);
3163	>	t.doExec();
3164	>	}
3165	>	break;
3166	>	}
3167		}
2496	–	} finally {
2497	–	lock.unlock();
3168		}
3169	+	return true;
3170	+	}
3171	+
3172	+	/**
3173	+	* Waits and/or attempts to assist performing tasks indefinitely
3174	+	* until the {@link #commonPool()} {@link #isQuiescent}
3175	+	*/
3176	+	static void quiesceCommonPool() {
3177	+	common.awaitQuiescence(Long.MAX_VALUE, TimeUnit.NANOSECONDS);
3178		}
3179
3180		/**
#	Line 2594 | Line 3273 | public class ForkJoinPool extends Abstra
3273		public static void managedBlock(ManagedBlocker blocker)
3274		throws InterruptedException {
3275		Thread t = Thread.currentThread();
3276	<	ForkJoinPool p = ((t instanceof ForkJoinWorkerThread) ?
3277	<	((ForkJoinWorkerThread)t).pool : null);
3278	<	while (!blocker.isReleasable()) {
3279	<	if (p == null || p.tryCompensate()) {
3280	<	try {
3281	<	do {} while (!blocker.isReleasable() && !blocker.block());
3282	<	} finally {
3283	<	if (p != null)
3276	>	if (t instanceof ForkJoinWorkerThread) {
3277	>	ForkJoinPool p = ((ForkJoinWorkerThread)t).pool;
3278	>	while (!blocker.isReleasable()) { // variant of helpSignal
3279	>	WorkQueue[] ws; WorkQueue q; int m, u;
3280	>	if ((ws = p.workQueues) != null && (m = ws.length - 1) >= 0) {
3281	>	for (int i = 0; i <= m; ++i) {
3282	>	if (blocker.isReleasable())
3283	>	return;
3284	>	if ((q = ws[i]) != null && q.base - q.top < 0) {
3285	>	p.signalWork(q);
3286	>	if ((u = (int)(p.ctl >>> 32)) >= 0 ||
3287	>	(u >> UAC_SHIFT) >= 0)
3288	>	break;
3289	>	}
3290	>	}
3291	>	}
3292	>	if (p.tryCompensate()) {
3293	>	try {
3294	>	do {} while (!blocker.isReleasable() &&
3295	>	!blocker.block());
3296	>	} finally {
3297		p.incrementActiveCount();
3298	+	}
3299	+	break;
3300		}
2607	–	break;
3301		}
3302		}
3303	+	else {
3304	+	do {} while (!blocker.isReleasable() &&
3305	+	!blocker.block());
3306	+	}
3307		}
3308
3309		// AbstractExecutorService overrides.  These rely on undocumented
#	Line 2614 | Line 3311 | public class ForkJoinPool extends Abstra
3311		// implement RunnableFuture.
3312
3313		protected <T> RunnableFuture<T> newTaskFor(Runnable runnable, T value) {
3314	<	return (RunnableFuture<T>) ForkJoinTask.adapt(runnable, value);
3314	>	return new ForkJoinTask.AdaptedRunnable<T>(runnable, value);
3315		}
3316
3317		protected <T> RunnableFuture<T> newTaskFor(Callable<T> callable) {
3318	<	return (RunnableFuture<T>) ForkJoinTask.adapt(callable);
3318	>	return new ForkJoinTask.AdaptedCallable<T>(callable);
3319		}
3320
3321		// Unsafe mechanics
3322		private static final sun.misc.Unsafe U;
3323		private static final long CTL;
2627	–	private static final long RUNSTATE;
3324		private static final long PARKBLOCKER;
3325	+	private static final int ABASE;
3326	+	private static final int ASHIFT;
3327	+	private static final long STEALCOUNT;
3328	+	private static final long PLOCK;
3329	+	private static final long INDEXSEED;
3330	+	private static final long QLOCK;
3331
3332		static {
3333	<	poolNumberGenerator = new AtomicInteger();
2632	<	modifyThreadPermission = new RuntimePermission("modifyThread");
2633	<	defaultForkJoinWorkerThreadFactory =
2634	<	new DefaultForkJoinWorkerThreadFactory();
2635	<	int s;
3333	>	int s; // initialize field offsets for CAS etc
3334		try {
3335		U = getUnsafe();
3336		Class<?> k = ForkJoinPool.class;
2639	–	Class<?> tk = Thread.class;
3337		CTL = U.objectFieldOffset
3338		(k.getDeclaredField("ctl"));
3339	<	RUNSTATE = U.objectFieldOffset
3340	<	(k.getDeclaredField("runState"));
3339	>	STEALCOUNT = U.objectFieldOffset
3340	>	(k.getDeclaredField("stealCount"));
3341	>	PLOCK = U.objectFieldOffset
3342	>	(k.getDeclaredField("plock"));
3343	>	INDEXSEED = U.objectFieldOffset
3344	>	(k.getDeclaredField("indexSeed"));
3345	>	Class<?> tk = Thread.class;
3346		PARKBLOCKER = U.objectFieldOffset
3347		(tk.getDeclaredField("parkBlocker"));
3348	+	Class<?> wk = WorkQueue.class;
3349	+	QLOCK = U.objectFieldOffset
3350	+	(wk.getDeclaredField("qlock"));
3351	+	Class<?> ak = ForkJoinTask[].class;
3352	+	ABASE = U.arrayBaseOffset(ak);
3353	+	s = U.arrayIndexScale(ak);
3354	+	ASHIFT = 31 - Integer.numberOfLeadingZeros(s);
3355		} catch (Exception e) {
3356		throw new Error(e);
3357		}
3358	+	if ((s & (s-1)) != 0)
3359	+	throw new Error("data type scale not a power of two");
3360	+
3361	+	submitters = new ThreadLocal<Submitter>();
3362	+	ForkJoinWorkerThreadFactory fac = defaultForkJoinWorkerThreadFactory =
3363	+	new DefaultForkJoinWorkerThreadFactory();
3364	+	modifyThreadPermission = new RuntimePermission("modifyThread");
3365	+
3366	+	/*
3367	+	* Establish common pool parameters.  For extra caution,
3368	+	* computations to set up common pool state are here; the
3369	+	* constructor just assigns these values to fields.
3370	+	*/
3371	+
3372	+	int par = 0;
3373	+	Thread.UncaughtExceptionHandler handler = null;
3374	+	try {  // TBD: limit or report ignored exceptions?
3375	+	String pp = System.getProperty
3376	+	("java.util.concurrent.ForkJoinPool.common.parallelism");
3377	+	String hp = System.getProperty
3378	+	("java.util.concurrent.ForkJoinPool.common.exceptionHandler");
3379	+	String fp = System.getProperty
3380	+	("java.util.concurrent.ForkJoinPool.common.threadFactory");
3381	+	if (fp != null)
3382	+	fac = ((ForkJoinWorkerThreadFactory)ClassLoader.
3383	+	getSystemClassLoader().loadClass(fp).newInstance());
3384	+	if (hp != null)
3385	+	handler = ((Thread.UncaughtExceptionHandler)ClassLoader.
3386	+	getSystemClassLoader().loadClass(hp).newInstance());
3387	+	if (pp != null)
3388	+	par = Integer.parseInt(pp);
3389	+	} catch (Exception ignore) {
3390	+	}
3391	+
3392	+	if (par <= 0)
3393	+	par = Runtime.getRuntime().availableProcessors();
3394	+	if (par > MAX_CAP)
3395	+	par = MAX_CAP;
3396	+	commonParallelism = par;
3397	+	long np = (long)(-par); // precompute initial ctl value
3398	+	long ct = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
3399	+
3400	+	common = new ForkJoinPool(par, ct, fac, handler);
3401		}
3402
3403		/**
#	Line 2658 | Line 3410 | public class ForkJoinPool extends Abstra
3410		private static sun.misc.Unsafe getUnsafe() {
3411		try {
3412		return sun.misc.Unsafe.getUnsafe();
3413	<	} catch (SecurityException se) {
3414	<	try {
3415	<	return java.security.AccessController.doPrivileged
3416	<	(new java.security
3417	<	.PrivilegedExceptionAction<sun.misc.Unsafe>() {
3418	<	public sun.misc.Unsafe run() throws Exception {
3419	<	java.lang.reflect.Field f = sun.misc
3420	<	.Unsafe.class.getDeclaredField("theUnsafe");
3421	<	f.setAccessible(true);
3422	<	return (sun.misc.Unsafe) f.get(null);
3423	<	}});
3424	<	} catch (java.security.PrivilegedActionException e) {
3425	<	throw new RuntimeException("Could not initialize intrinsics",
3426	<	e.getCause());
3427	<	}
3413	>	} catch (SecurityException tryReflectionInstead) {}
3414	>	try {
3415	>	return java.security.AccessController.doPrivileged
3416	>	(new java.security.PrivilegedExceptionAction<sun.misc.Unsafe>() {
3417	>	public sun.misc.Unsafe run() throws Exception {
3418	>	Class<sun.misc.Unsafe> k = sun.misc.Unsafe.class;
3419	>	for (java.lang.reflect.Field f : k.getDeclaredFields()) {
3420	>	f.setAccessible(true);
3421	>	Object x = f.get(null);
3422	>	if (k.isInstance(x))
3423	>	return k.cast(x);
3424	>	}
3425	>	throw new NoSuchFieldError("the Unsafe");
3426	>	}});
3427	>	} catch (java.security.PrivilegedActionException e) {
3428	>	throw new RuntimeException("Could not initialize intrinsics",
3429	>	e.getCause());
3430		}
3431		}
2678	–
3432		}

Diff Legend

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