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 1.36 by jsr166, Sat Aug 1 22:09:13 2009 UTC vs.
Revision 1.62 by dl, Wed Aug 11 20:28:22 2010 UTC

#	Line 13 | Line 13 | import java.util.Arrays;
13		import java.util.Collection;
14		import java.util.Collections;
15		import java.util.List;
16	–	import java.util.concurrent.locks.Condition;
16		import java.util.concurrent.locks.LockSupport;
17		import java.util.concurrent.locks.ReentrantLock;
18		import java.util.concurrent.atomic.AtomicInteger;
19	<	import java.util.concurrent.atomic.AtomicLong;
19	>	import java.util.concurrent.CountDownLatch;
20
21		/**
22		* An {@link ExecutorService} for running {@link ForkJoinTask}s.
23	<	* A ForkJoinPool provides the entry point for submissions from
24	<	* non-ForkJoinTasks, as well as management and monitoring operations.
25	<	* Normally a single ForkJoinPool is used for a large number of
27	<	* submitted tasks. Otherwise, use would not usually outweigh the
28	<	* construction and bookkeeping overhead of creating a large set of
29	<	* threads.
23	>	* A {@code ForkJoinPool} provides the entry point for submissions
24	>	* from non-{@code ForkJoinTask} clients, as well as management and
25	>	* monitoring operations.
26		*
27	<	* <p>ForkJoinPools differ from other kinds of Executors mainly in
28	<	* that they provide <em>work-stealing</em>: all threads in the pool
29	<	* attempt to find and execute subtasks created by other active tasks
30	<	* (eventually blocking if none exist). This makes them efficient when
31	<	* most tasks spawn other subtasks (as do most ForkJoinTasks), as well
32	<	* as the mixed execution of some plain Runnable- or Callable- based
33	<	* activities along with ForkJoinTasks. When setting {@linkplain
34	<	* #setAsyncMode async mode}, a ForkJoinPool may also be appropriate
35	<	* for use with fine-grained tasks that are never joined. Otherwise,
40	<	* other ExecutorService implementations are typically more
41	<	* appropriate choices.
27	>	* <p>A {@code ForkJoinPool} differs from other kinds of {@link
28	>	* ExecutorService} mainly by virtue of employing
29	>	* <em>work-stealing</em>: all threads in the pool attempt to find and
30	>	* execute subtasks created by other active tasks (eventually blocking
31	>	* waiting for work if none exist). This enables efficient processing
32	>	* when most tasks spawn other subtasks (as do most {@code
33	>	* ForkJoinTask}s). When setting <em>asyncMode</em> to true in
34	>	* constructors, {@code ForkJoinPool}s may also be appropriate for use
35	>	* with event-style tasks that are never joined.
36		*
37	<	* <p>A ForkJoinPool may be constructed with a given parallelism level
38	<	* (target pool size), which it attempts to maintain by dynamically
39	<	* adding, suspending, or resuming threads, even if some tasks are
40	<	* waiting to join others. However, no such adjustments are performed
41	<	* in the face of blocked IO or other unmanaged synchronization. The
42	<	* nested {@link ManagedBlocker} interface enables extension of
43	<	* the kinds of synchronization accommodated.  The target parallelism
44	<	* level may also be changed dynamically ({@link #setParallelism})
45	<	* and thread construction can be limited using methods
52	<	* {@link #setMaximumPoolSize} and/or
53	<	* {@link #setMaintainsParallelism}.
37	>	* <p>A {@code ForkJoinPool} is constructed with a given target
38	>	* parallelism level; by default, equal to the number of available
39	>	* processors. The pool attempts to maintain enough active (or
40	>	* available) threads by dynamically adding, suspending, or resuming
41	>	* internal worker threads, even if some tasks are stalled waiting to
42	>	* join others. However, no such adjustments are guaranteed in the
43	>	* face of blocked IO or other unmanaged synchronization. The nested
44	>	* {@link ManagedBlocker} interface enables extension of the kinds of
45	>	* synchronization accommodated.
46		*
47		* <p>In addition to execution and lifecycle control methods, this
48		* class provides status check methods (for example
#	Line 59 | Line 51 | import java.util.concurrent.atomic.Atomi
51		* {@link #toString} returns indications of pool state in a
52		* convenient form for informal monitoring.
53		*
54	+	* <p> As is the case with other ExecutorServices, there are three
55	+	* main task execution methods summarized in the following
56	+	* table. These are designed to be used by clients not already engaged
57	+	* in fork/join computations in the current pool.  The main forms of
58	+	* these methods accept instances of {@code ForkJoinTask}, but
59	+	* overloaded forms also allow mixed execution of plain {@code
60	+	* Runnable}- or {@code Callable}- based activities as well.  However,
61	+	* tasks that are already executing in a pool should normally
62	+	* <em>NOT</em> use these pool execution methods, but instead use the
63	+	* within-computation forms listed in the table.
64	+	*
65	+	* <table BORDER CELLPADDING=3 CELLSPACING=1>
66	+	*  <tr>
67	+	*    <td></td>
68	+	*    <td ALIGN=CENTER> <b>Call from non-fork/join clients</b></td>
69	+	*    <td ALIGN=CENTER> <b>Call from within fork/join computations</b></td>
70	+	*  </tr>
71	+	*  <tr>
72	+	*    <td> <b>Arange async execution</td>
73	+	*    <td> {@link #execute(ForkJoinTask)}</td>
74	+	*    <td> {@link ForkJoinTask#fork}</td>
75	+	*  </tr>
76	+	*  <tr>
77	+	*    <td> <b>Await and obtain result</td>
78	+	*    <td> {@link #invoke(ForkJoinTask)}</td>
79	+	*    <td> {@link ForkJoinTask#invoke}</td>
80	+	*  </tr>
81	+	*  <tr>
82	+	*    <td> <b>Arrange exec and obtain Future</td>
83	+	*    <td> {@link #submit(ForkJoinTask)}</td>
84	+	*    <td> {@link ForkJoinTask#fork} (ForkJoinTasks <em>are</em> Futures)</td>
85	+	*  </tr>
86	+	* </table>
87	+	*
88	+	* <p><b>Sample Usage.</b> Normally a single {@code ForkJoinPool} is
89	+	* used for all parallel task execution in a program or subsystem.
90	+	* Otherwise, use would not usually outweigh the construction and
91	+	* bookkeeping overhead of creating a large set of threads. For
92	+	* example, a common pool could be used for the {@code SortTasks}
93	+	* illustrated in {@link RecursiveAction}. Because {@code
94	+	* ForkJoinPool} uses threads in {@linkplain java.lang.Thread#isDaemon
95	+	* daemon} mode, there is typically no need to explicitly {@link
96	+	* #shutdown} such a pool upon program exit.
97	+	*
98	+	* <pre>
99	+	* static final ForkJoinPool mainPool = new ForkJoinPool();
100	+	* ...
101	+	* public void sort(long[] array) {
102	+	*   mainPool.invoke(new SortTask(array, 0, array.length));
103	+	* }
104	+	* </pre>
105	+	*
106		* <p><b>Implementation notes</b>: This implementation restricts the
107		* maximum number of running threads to 32767. Attempts to create
108	<	* pools with greater than the maximum result in
109	<	* IllegalArgumentExceptions.
108	>	* pools with greater than the maximum number result in
109	>	* {@code IllegalArgumentException}.
110	>	*
111	>	* <p>This implementation rejects submitted tasks (that is, by throwing
112	>	* {@link RejectedExecutionException}) only when the pool is shut down
113	>	* or internal resources have been exhausted.
114		*
115		* @since 1.7
116		* @author Doug Lea
#	Line 70 | Line 118 | import java.util.concurrent.atomic.Atomi
118		public class ForkJoinPool extends AbstractExecutorService {
119
120		/*
121	<	* See the extended comments interspersed below for design,
122	<	* rationale, and walkthroughs.
121	>	* Implementation Overview
122	>	*
123	>	* This class provides the central bookkeeping and control for a
124	>	* set of worker threads: Submissions from non-FJ threads enter
125	>	* into a submission queue. Workers take these tasks and typically
126	>	* split them into subtasks that may be stolen by other workers.
127	>	* The main work-stealing mechanics implemented in class
128	>	* ForkJoinWorkerThread give first priority to processing tasks
129	>	* from their own queues (LIFO or FIFO, depending on mode), then
130	>	* to randomized FIFO steals of tasks in other worker queues, and
131	>	* lastly to new submissions. These mechanics do not consider
132	>	* affinities, loads, cache localities, etc, so rarely provide the
133	>	* best possible performance on a given machine, but portably
134	>	* provide good throughput by averaging over these factors.
135	>	* (Further, even if we did try to use such information, we do not
136	>	* usually have a basis for exploiting it. For example, some sets
137	>	* of tasks profit from cache affinities, but others are harmed by
138	>	* cache pollution effects.)
139	>	*
140	>	* Beyond work-stealing support and essential bookkeeping, the
141	>	* main responsibility of this framework is to take actions when
142	>	* one worker is waiting to join a task stolen (or always held by)
143	>	* another.  Becauae we are multiplexing many tasks on to a pool
144	>	* of workers, we can't just let them block (as in Thread.join).
145	>	* We also cannot just reassign the joiner's run-time stack with
146	>	* another and replace it later, which would be a form of
147	>	* "continuation", that even if possible is not necessarily a good
148	>	* idea. Given that the creation costs of most threads on most
149	>	* systems mainly surrounds setting up runtime stacks, thread
150	>	* creation and switching is usually not much more expensive than
151	>	* stack creation and switching, and is more flexible). Instead we
152	>	* combine two tactics:
153	>	*
154	>	*   Helping: Arranging for the joiner to execute some task that it
155	>	*      would be running if the steal had not occurred.  Method
156	>	*      ForkJoinWorkerThread.helpJoinTask tracks joining->stealing
157	>	*      links to try to find such a task.
158	>	*
159	>	*   Compensating: Unless there are already enough live threads,
160	>	*      method helpMaintainParallelism() may create or or
161	>	*      re-activate a spare thread to compensate for blocked
162	>	*      joiners until they unblock.
163	>	*
164	>	* Because the determining existence of conservatively safe
165	>	* helping targets, the availability of already-created spares,
166	>	* and the apparent need to create new spares are all racy and
167	>	* require heuristic guidance, we rely on multiple retries of
168	>	* each. Further, because it is impossible to keep exactly the
169	>	* target (parallelism) number of threads running at any given
170	>	* time, we allow compensation during joins to fail, and enlist
171	>	* all other threads to help out whenever they are not otherwise
172	>	* occupied (i.e., mainly in method preStep).
173	>	*
174	>	* The ManagedBlocker extension API can't use helping so relies
175	>	* only on compensation in method awaitBlocker.
176	>	*
177	>	* The main throughput advantages of work-stealing stem from
178	>	* decentralized control -- workers mostly steal tasks from each
179	>	* other. We do not want to negate this by creating bottlenecks
180	>	* implementing other management responsibilities. So we use a
181	>	* collection of techniques that avoid, reduce, or cope well with
182	>	* contention. These entail several instances of bit-packing into
183	>	* CASable fields to maintain only the minimally required
184	>	* atomicity. To enable such packing, we restrict maximum
185	>	* parallelism to (1<<15)-1 (enabling twice this (to accommodate
186	>	* unbalanced increments and decrements) to fit into a 16 bit
187	>	* field, which is far in excess of normal operating range.  Even
188	>	* though updates to some of these bookkeeping fields do sometimes
189	>	* contend with each other, they don't normally cache-contend with
190	>	* updates to others enough to warrant memory padding or
191	>	* isolation. So they are all held as fields of ForkJoinPool
192	>	* objects.  The main capabilities are as follows:
193	>	*
194	>	* 1. Creating and removing workers. Workers are recorded in the
195	>	* "workers" array. This is an array as opposed to some other data
196	>	* structure to support index-based random steals by workers.
197	>	* Updates to the array recording new workers and unrecording
198	>	* terminated ones are protected from each other by a lock
199	>	* (workerLock) but the array is otherwise concurrently readable,
200	>	* and accessed directly by workers. To simplify index-based
201	>	* operations, the array size is always a power of two, and all
202	>	* readers must tolerate null slots. Currently, all worker thread
203	>	* creation is on-demand, triggered by task submissions,
204	>	* replacement of terminated workers, and/or compensation for
205	>	* blocked workers. However, all other support code is set up to
206	>	* work with other policies.
207	>	*
208	>	* To ensure that we do not hold on to worker references that
209	>	* would prevent GC, ALL accesses to workers are via indices into
210	>	* the workers array (which is one source of some of the unusual
211	>	* code constructions here). In essence, the workers array serves
212	>	* as a WeakReference mechanism. Thus for example the event queue
213	>	* stores worker indices, not worker references. Access to the
214	>	* workers in associated methods (for example releaseEventWaiters)
215	>	* must both index-check and null-check the IDs. All such accesses
216	>	* ignore bad IDs by returning out early from what they are doing,
217	>	* since this can only be associated with shutdown, in which case
218	>	* it is OK to give up. On termination, we just clobber these
219	>	* data structures without trying to use them.
220	>	*
221	>	* 2. Bookkeeping for dynamically adding and removing workers. We
222	>	* aim to approximately maintain the given level of parallelism.
223	>	* When some workers are known to be blocked (on joins or via
224	>	* ManagedBlocker), we may create or resume others to take their
225	>	* place until they unblock (see below). Implementing this
226	>	* requires counts of the number of "running" threads (i.e., those
227	>	* that are neither blocked nor artifically suspended) as well as
228	>	* the total number.  These two values are packed into one field,
229	>	* "workerCounts" because we need accurate snapshots when deciding
230	>	* to create, resume or suspend.  Note however that the
231	>	* correspondance of these counts to reality is not guaranteed. In
232	>	* particular updates for unblocked threads may lag until they
233	>	* actually wake up.
234	>	*
235	>	* 3. Maintaining global run state. The run state of the pool
236	>	* consists of a runLevel (SHUTDOWN, TERMINATING, etc) similar to
237	>	* those in other Executor implementations, as well as a count of
238	>	* "active" workers -- those that are, or soon will be, or
239	>	* recently were executing tasks. The runLevel and active count
240	>	* are packed together in order to correctly trigger shutdown and
241	>	* termination. Without care, active counts can be subject to very
242	>	* high contention.  We substantially reduce this contention by
243	>	* relaxing update rules.  A worker must claim active status
244	>	* prospectively, by activating if it sees that a submitted or
245	>	* stealable task exists (it may find after activating that the
246	>	* task no longer exists). It stays active while processing this
247	>	* task (if it exists) and any other local subtasks it produces,
248	>	* until it cannot find any other tasks. It then tries
249	>	* inactivating (see method preStep), but upon update contention
250	>	* instead scans for more tasks, later retrying inactivation if it
251	>	* doesn't find any.
252	>	*
253	>	* 4. Managing idle workers waiting for tasks. We cannot let
254	>	* workers spin indefinitely scanning for tasks when none are
255	>	* available. On the other hand, we must quickly prod them into
256	>	* action when new tasks are submitted or generated.  We
257	>	* park/unpark these idle workers using an event-count scheme.
258	>	* Field eventCount is incremented upon events that may enable
259	>	* workers that previously could not find a task to now find one:
260	>	* Submission of a new task to the pool, or another worker pushing
261	>	* a task onto a previously empty queue.  (We also use this
262	>	* mechanism for termination actions that require wakeups of idle
263	>	* workers).  Each worker maintains its last known event count,
264	>	* and blocks when a scan for work did not find a task AND its
265	>	* lastEventCount matches the current eventCount. Waiting idle
266	>	* workers are recorded in a variant of Treiber stack headed by
267	>	* field eventWaiters which, when nonzero, encodes the thread
268	>	* index and count awaited for by the worker thread most recently
269	>	* calling eventSync. This thread in turn has a record (field
270	>	* nextEventWaiter) for the next waiting worker.  In addition to
271	>	* allowing simpler decisions about need for wakeup, the event
272	>	* count bits in eventWaiters serve the role of tags to avoid ABA
273	>	* errors in Treiber stacks.  To reduce delays in task diffusion,
274	>	* workers not otherwise occupied may invoke method
275	>	* releaseEventWaiters, that removes and signals (unparks) workers
276	>	* not waiting on current count. To reduce stalls, To minimize
277	>	* task production stalls associate with signalling, any worker
278	>	* pushing a task on an empty queue invokes the weaker method
279	>	* signalWork, that only releases idle workers until it detects
280	>	* interference by other threads trying to release, and lets them
281	>	* take over.  The net effect is a tree-like diffusion of signals,
282	>	* where released threads (and possibly others) help with unparks.
283	>	* To further reduce contention effects a bit, failed CASes to
284	>	* increment field eventCount are tolerated without retries.
285	>	* Conceptually they are merged into the same event, which is OK
286	>	* when their only purpose is to enable workers to scan for work.
287	>	*
288	>	* 5. Managing suspension of extra workers. When a worker is about
289	>	* to block waiting for a join (or via ManagedBlockers), we may
290	>	* create a new thread to maintain parallelism level, or at least
291	>	* avoid starvation. Usually, extra threads are needed for only
292	>	* very short periods, yet join dependencies are such that we
293	>	* sometimes need them in bursts. Rather than create new threads
294	>	* each time this happens, we suspend no-longer-needed extra ones
295	>	* as "spares". For most purposes, we don't distinguish "extra"
296	>	* spare threads from normal "core" threads: On each call to
297	>	* preStep (the only point at which we can do this) a worker
298	>	* checks to see if there are now too many running workers, and if
299	>	* so, suspends itself.  Method helpMaintainParallelism looks for
300	>	* suspended threads to resume before considering creating a new
301	>	* replacement. The spares themselves are encoded on another
302	>	* variant of a Treiber Stack, headed at field "spareWaiters".
303	>	* Note that the use of spares is intrinsically racy.  One thread
304	>	* may become a spare at about the same time as another is
305	>	* needlessly being created. We counteract this and related slop
306	>	* in part by requiring resumed spares to immediately recheck (in
307	>	* preStep) to see whether they they should re-suspend.  To avoid
308	>	* long-term build-up of spares, the oldest spare (see
309	>	* ForkJoinWorkerThread.suspendAsSpare) occasionally wakes up if
310	>	* not signalled and calls tryTrimSpare, which uses two different
311	>	* thresholds: Always killing if the number of spares is greater
312	>	* that 25% of total, and killing others only at a slower rate
313	>	* (UNUSED_SPARE_TRIM_RATE_NANOS).
314	>	*
315	>	* 6. Deciding when to create new workers. The main dynamic
316	>	* control in this class is deciding when to create extra threads
317	>	* in method helpMaintainParallelism. We would like to keep
318	>	* exactly #parallelism threads running, which is an impossble
319	>	* task. We always need to create one when the number of running
320	>	* threads would become zero and all workers are busy. Beyond
321	>	* this, we must rely on heuristics that work well in the the
322	>	* presence of transients phenomena such as GC stalls, dynamic
323	>	* compilation, and wake-up lags. These transients are extremely
324	>	* common -- we are normally trying to fully saturate the CPUs on
325	>	* a machine, so almost any activity other than running tasks
326	>	* impedes accuracy. Our main defense is to allow some slack in
327	>	* creation thresholds, using rules that reflect the fact that the
328	>	* more threads we have running, the more likely that we are
329	>	* underestimating the number running threads. The rules also
330	>	* better cope with the fact that some of the methods in this
331	>	* class tend to never become compiled (but are interpreted), so
332	>	* some components of the entire set of controls might execute 100
333	>	* times faster than others. And similarly for cases where the
334	>	* apparent lack of work is just due to GC stalls and other
335	>	* transient system activity.
336	>	*
337	>	* Beware that there is a lot of representation-level coupling
338	>	* among classes ForkJoinPool, ForkJoinWorkerThread, and
339	>	* ForkJoinTask.  For example, direct access to "workers" array by
340	>	* workers, and direct access to ForkJoinTask.status by both
341	>	* ForkJoinPool and ForkJoinWorkerThread.  There is little point
342	>	* trying to reduce this, since any associated future changes in
343	>	* representations will need to be accompanied by algorithmic
344	>	* changes anyway.
345	>	*
346	>	* Style notes: There are lots of inline assignments (of form
347	>	* "while ((local = field) != 0)") which are usually the simplest
348	>	* way to ensure the required read orderings (which are sometimes
349	>	* critical). Also several occurrences of the unusual "do {}
350	>	* while(!cas...)" which is the simplest way to force an update of
351	>	* a CAS'ed variable. There are also other coding oddities that
352	>	* help some methods perform reasonably even when interpreted (not
353	>	* compiled), at the expense of some messy constructions that
354	>	* reduce byte code counts.
355	>	*
356	>	* The order of declarations in this file is: (1) statics (2)
357	>	* fields (along with constants used when unpacking some of them)
358	>	* (3) internal control methods (4) callbacks and other support
359	>	* for ForkJoinTask and ForkJoinWorkerThread classes, (5) exported
360	>	* methods (plus a few little helpers).
361		*/
362
77	–	/** Mask for packing and unpacking shorts */
78	–	private static final int  shortMask = 0xffff;
79	–
80	–	/** Max pool size -- must be a power of two minus 1 */
81	–	private static final int MAX_THREADS =  0x7FFF;
82	–
363		/**
364		* Factory for creating new {@link ForkJoinWorkerThread}s.
365		* A {@code ForkJoinWorkerThreadFactory} must be defined and used
#	Line 91 | Line 371 | public class ForkJoinPool extends Abstra
371		* Returns a new worker thread operating in the given pool.
372		*
373		* @param pool the pool this thread works in
374	<	* @throws NullPointerException if pool is null
374	>	* @throws NullPointerException if the pool is null
375		*/
376		public ForkJoinWorkerThread newThread(ForkJoinPool pool);
377		}
#	Line 100 | Line 380 | public class ForkJoinPool extends Abstra
380		* Default ForkJoinWorkerThreadFactory implementation; creates a
381		* new ForkJoinWorkerThread.
382		*/
383	<	static class  DefaultForkJoinWorkerThreadFactory
383	>	static class DefaultForkJoinWorkerThreadFactory
384		implements ForkJoinWorkerThreadFactory {
385		public ForkJoinWorkerThread newThread(ForkJoinPool pool) {
386	<	try {
107	<	return new ForkJoinWorkerThread(pool);
108	<	} catch (OutOfMemoryError oom)  {
109	<	return null;
110	<	}
386	>	return new ForkJoinWorkerThread(pool);
387		}
388		}
389
#	Line 143 | Line 419 | public class ForkJoinPool extends Abstra
419		new AtomicInteger();
420
421		/**
422	<	* Array holding all worker threads in the pool. Initialized upon
423	<	* first use. Array size must be a power of two.  Updates and
424	<	* replacements are protected by workerLock, but it is always kept
425	<	* in a consistent enough state to be randomly accessed without
426	<	* locking by workers performing work-stealing.
422	>	* Absolute bound for parallelism level. Twice this number plus
423	>	* one (i.e., 0xfff) must fit into a 16bit field to enable
424	>	* word-packing for some counts and indices.
425	>	*/
426	>	private static final int MAX_WORKERS   = 0x7fff;
427	>
428	>	/**
429	>	* Array holding all worker threads in the pool.  Array size must
430	>	* be a power of two.  Updates and replacements are protected by
431	>	* workerLock, but the array is always kept in a consistent enough
432	>	* state to be randomly accessed without locking by workers
433	>	* performing work-stealing, as well as other traversal-based
434	>	* methods in this class. All readers must tolerate that some
435	>	* array slots may be null.
436		*/
437		volatile ForkJoinWorkerThread[] workers;
438
439		/**
440	<	* Lock protecting access to workers.
440	>	* Queue for external submissions.
441		*/
442	<	private final ReentrantLock workerLock;
442	>	private final LinkedTransferQueue<ForkJoinTask<?>> submissionQueue;
443
444		/**
445	<	* Condition for awaitTermination.
445	>	* Lock protecting updates to workers array.
446		*/
447	<	private final Condition termination;
447	>	private final ReentrantLock workerLock;
448
449		/**
450	<	* The uncaught exception handler used when any worker
166	<	* abruptly terminates
450	>	* Latch released upon termination.
451		*/
452	<	private Thread.UncaughtExceptionHandler ueh;
452	>	private final Phaser termination;
453
454		/**
455		* Creation factory for worker threads.
#	Line 173 | Line 457 | public class ForkJoinPool extends Abstra
457		private final ForkJoinWorkerThreadFactory factory;
458
459		/**
176	–	* Head of stack of threads that were created to maintain
177	–	* parallelism when other threads blocked, but have since
178	–	* suspended when the parallelism level rose.
179	–	*/
180	–	private volatile WaitQueueNode spareStack;
181	–
182	–	/**
460		* Sum of per-thread steal counts, updated only when threads are
461		* idle or terminating.
462		*/
463	<	private final AtomicLong stealCount;
463	>	private volatile long stealCount;
464
465		/**
466	<	* Queue for external submissions.
466	>	* The last nanoTime that a spare thread was trimmed
467		*/
468	<	private final LinkedTransferQueue<ForkJoinTask<?>> submissionQueue;
468	>	private volatile long trimTime;
469
470		/**
471	<	* Head of Treiber stack for barrier sync. See below for explanation.
471	>	* The rate at which to trim unused spares
472		*/
473	<	private volatile WaitQueueNode syncStack;
473	>	static final long UNUSED_SPARE_TRIM_RATE_NANOS =
474	>	1000L * 1000L * 1000L; // 1 sec
475
476		/**
477	<	* The count for event barrier
477	>	* Encoded record of top of treiber stack of threads waiting for
478	>	* events. The top 32 bits contain the count being waited for. The
479	>	* bottom 16 bits contains one plus the pool index of waiting
480	>	* worker thread. (Bits 16-31 are unused.)
481		*/
482	<	private volatile long eventCount;
482	>	private volatile long eventWaiters;
483	>
484	>	private static final int  EVENT_COUNT_SHIFT = 32;
485	>	private static final long WAITER_ID_MASK    = (1L << 16) - 1L;
486
487		/**
488	<	* Pool number, just for assigning useful names to worker threads
488	>	* A counter for events that may wake up worker threads:
489	>	*   - Submission of a new task to the pool
490	>	*   - A worker pushing a task on an empty queue
491	>	*   - termination
492		*/
493	<	private final int poolNumber;
493	>	private volatile int eventCount;
494
495		/**
496	<	* The maximum allowed pool size
496	>	* Encoded record of top of treiber stack of spare threads waiting
497	>	* for resumption. The top 16 bits contain an arbitrary count to
498	>	* avoid ABA effects. The bottom 16bits contains one plus the pool
499	>	* index of waiting worker thread.
500	>	*/
501	>	private volatile int spareWaiters;
502	>
503	>	private static final int SPARE_COUNT_SHIFT = 16;
504	>	private static final int SPARE_ID_MASK     = (1 << 16) - 1;
505	>
506	>	/**
507	>	* Lifecycle control. The low word contains the number of workers
508	>	* that are (probably) executing tasks. This value is atomically
509	>	* incremented before a worker gets a task to run, and decremented
510	>	* when worker has no tasks and cannot find any.  Bits 16-18
511	>	* contain runLevel value. When all are zero, the pool is
512	>	* running. Level transitions are monotonic (running -> shutdown
513	>	* -> terminating -> terminated) so each transition adds a bit.
514	>	* These are bundled together to ensure consistent read for
515	>	* termination checks (i.e., that runLevel is at least SHUTDOWN
516	>	* and active threads is zero).
517	>	*/
518	>	private volatile int runState;
519	>
520	>	// Note: The order among run level values matters.
521	>	private static final int RUNLEVEL_SHIFT     = 16;
522	>	private static final int SHUTDOWN           = 1 << RUNLEVEL_SHIFT;
523	>	private static final int TERMINATING        = 1 << (RUNLEVEL_SHIFT + 1);
524	>	private static final int TERMINATED         = 1 << (RUNLEVEL_SHIFT + 2);
525	>	private static final int ACTIVE_COUNT_MASK  = (1 << RUNLEVEL_SHIFT) - 1;
526	>	private static final int ONE_ACTIVE         = 1; // active update delta
527	>
528	>	/**
529	>	* Holds number of total (i.e., created and not yet terminated)
530	>	* and running (i.e., not blocked on joins or other managed sync)
531	>	* threads, packed together to ensure consistent snapshot when
532	>	* making decisions about creating and suspending spare
533	>	* threads. Updated only by CAS. Note that adding a new worker
534	>	* requires incrementing both counts, since workers start off in
535	>	* running state.
536		*/
537	<	private volatile int maxPoolSize;
537	>	private volatile int workerCounts;
538	>
539	>	private static final int TOTAL_COUNT_SHIFT  = 16;
540	>	private static final int RUNNING_COUNT_MASK = (1 << TOTAL_COUNT_SHIFT) - 1;
541	>	private static final int ONE_RUNNING        = 1;
542	>	private static final int ONE_TOTAL          = 1 << TOTAL_COUNT_SHIFT;
543
544		/**
545	<	* The desired parallelism level, updated only under workerLock.
545	>	* The target parallelism level.
546	>	* Accessed directly by ForkJoinWorkerThreads.
547		*/
548	<	private volatile int parallelism;
548	>	final int parallelism;
549
550		/**
551		* True if use local fifo, not default lifo, for local polling
552	+	* Read by, and replicated by ForkJoinWorkerThreads
553		*/
554	<	private volatile boolean locallyFifo;
554	>	final boolean locallyFifo;
555
556		/**
557	<	* Holds number of total (i.e., created and not yet terminated)
558	<	* and running (i.e., not blocked on joins or other managed sync)
226	<	* threads, packed into one int to ensure consistent snapshot when
227	<	* making decisions about creating and suspending spare
228	<	* threads. Updated only by CAS.  Note: CASes in
229	<	* updateRunningCount and preJoin assume that running active count
230	<	* is in low word, so need to be modified if this changes.
557	>	* The uncaught exception handler used when any worker abruptly
558	>	* terminates.
559		*/
560	<	private volatile int workerCounts;
560	>	private final Thread.UncaughtExceptionHandler ueh;
561	>
562	>	/**
563	>	* Pool number, just for assigning useful names to worker threads
564	>	*/
565	>	private final int poolNumber;
566	>
567
568	<	private static int totalCountOf(int s)           { return s >>> 16;  }
569	<	private static int runningCountOf(int s)         { return s & shortMask; }
236	<	private static int workerCountsFor(int t, int r) { return (t << 16) + r; }
568	>	// Utilities for CASing fields. Note that several of these
569	>	// are manually inlined by callers
570
571		/**
572	<	* Adds delta (which may be negative) to running count.  This must
240	<	* be called before (with negative arg) and after (with positive)
241	<	* any managed synchronization (i.e., mainly, joins).
242	<	*
243	<	* @param delta the number to add
572	>	* Increments running count part of workerCounts
573		*/
574	<	final void updateRunningCount(int delta) {
575	<	int s;
576	<	do {} while (!casWorkerCounts(s = workerCounts, s + delta));
574	>	final void incrementRunningCount() {
575	>	int c;
576	>	do {} while (!UNSAFE.compareAndSwapInt(this, workerCountsOffset,
577	>	c = workerCounts,
578	>	c + ONE_RUNNING));
579		}
580
581		/**
582	<	* Adds delta (which may be negative) to both total and running
252	<	* count.  This must be called upon creation and termination of
253	<	* worker threads.
254	<	*
255	<	* @param delta the number to add
582	>	* Tries to decrement running count unless already zero
583		*/
584	<	private void updateWorkerCount(int delta) {
585	<	int d = delta + (delta << 16); // add to both lo and hi parts
586	<	int s;
587	<	do {} while (!casWorkerCounts(s = workerCounts, s + d));
584	>	final boolean tryDecrementRunningCount() {
585	>	int wc = workerCounts;
586	>	if ((wc & RUNNING_COUNT_MASK) == 0)
587	>	return false;
588	>	return UNSAFE.compareAndSwapInt(this, workerCountsOffset,
589	>	wc, wc - ONE_RUNNING);
590		}
591
592		/**
593	<	* Lifecycle control. High word contains runState, low word
594	<	* contains the number of workers that are (probably) executing
595	<	* tasks. This value is atomically incremented before a worker
596	<	* gets a task to run, and decremented when worker has no tasks
597	<	* and cannot find any. These two fields are bundled together to
269	<	* support correct termination triggering.  Note: activeCount
270	<	* CAS'es cheat by assuming active count is in low word, so need
271	<	* to be modified if this changes
593	>	* Forces decrement of encoded workerCounts, awaiting nonzero if
594	>	* (rarely) necessary when other count updates lag.
595	>	*
596	>	* @param dr -- either zero or ONE_RUNNING
597	>	* @param dt == either zero or ONE_TOTAL
598		*/
599	<	private volatile int runControl;
600	<
601	<	// RunState values. Order among values matters
602	<	private static final int RUNNING     = 0;
603	<	private static final int SHUTDOWN    = 1;
604	<	private static final int TERMINATING = 2;
605	<	private static final int TERMINATED  = 3;
599	>	private void decrementWorkerCounts(int dr, int dt) {
600	>	for (;;) {
601	>	int wc = workerCounts;
602	>	if (wc == 0 && (runState & TERMINATED) != 0)
603	>	return; // lagging termination on a backout
604	>	if ((wc & RUNNING_COUNT_MASK)  - dr < 0 ||
605	>	(wc >>> TOTAL_COUNT_SHIFT) - dt < 0)
606	>	Thread.yield();
607	>	if (UNSAFE.compareAndSwapInt(this, workerCountsOffset,
608	>	wc, wc - (dr + dt)))
609	>	return;
610	>	}
611	>	}
612
613	<	private static int runStateOf(int c)             { return c >>> 16; }
614	<	private static int activeCountOf(int c)          { return c & shortMask; }
615	<	private static int runControlFor(int r, int a)   { return (r << 16) + a; }
613	>	/**
614	>	* Increments event count
615	>	*/
616	>	private void advanceEventCount() {
617	>	int c;
618	>	do {} while(!UNSAFE.compareAndSwapInt(this, eventCountOffset,
619	>	c = eventCount, c+1));
620	>	}
621
622		/**
623		* Tries incrementing active count; fails on contention.
624	<	* Called by workers before/during executing tasks.
624	>	* Called by workers before executing tasks.
625		*
626		* @return true on success
627		*/
628		final boolean tryIncrementActiveCount() {
629	<	int c = runControl;
630	<	return casRunControl(c, c+1);
629	>	int c;
630	>	return UNSAFE.compareAndSwapInt(this, runStateOffset,
631	>	c = runState, c + ONE_ACTIVE);
632		}
633
634		/**
635		* Tries decrementing active count; fails on contention.
636	<	* Possibly triggers termination on success.
299	<	* Called by workers when they can't find tasks.
300	<	*
301	<	* @return true on success
636	>	* Called when workers cannot find tasks to run.
637		*/
638		final boolean tryDecrementActiveCount() {
639	<	int c = runControl;
640	<	int nextc = c - 1;
641	<	if (!casRunControl(c, nextc))
307	<	return false;
308	<	if (canTerminateOnShutdown(nextc))
309	<	terminateOnShutdown();
310	<	return true;
639	>	int c;
640	>	return UNSAFE.compareAndSwapInt(this, runStateOffset,
641	>	c = runState, c - ONE_ACTIVE);
642		}
643
644		/**
645	<	* Returns {@code true} if argument represents zero active count
646	<	* and nonzero runstate, which is the triggering condition for
316	<	* terminating on shutdown.
645	>	* Advances to at least the given level. Returns true if not
646	>	* already in at least the given level.
647		*/
648	<	private static boolean canTerminateOnShutdown(int c) {
649	<	// i.e. least bit is nonzero runState bit
650	<	return ((c & -c) >>> 16) != 0;
648	>	private boolean advanceRunLevel(int level) {
649	>	for (;;) {
650	>	int s = runState;
651	>	if ((s & level) != 0)
652	>	return false;
653	>	if (UNSAFE.compareAndSwapInt(this, runStateOffset, s, s | level))
654	>	return true;
655	>	}
656		}
657
658	+	// workers array maintenance
659	+
660		/**
661	<	* Transition run state to at least the given state. Return true
325	<	* if not already at least given state.
661	>	* Records and returns a workers array index for new worker.
662		*/
663	<	private boolean transitionRunStateTo(int state) {
664	<	for (;;) {
665	<	int c = runControl;
666	<	if (runStateOf(c) >= state)
667	<	return false;
668	<	if (casRunControl(c, runControlFor(state, activeCountOf(c))))
669	<	return true;
663	>	private int recordWorker(ForkJoinWorkerThread w) {
664	>	// Try using slot totalCount-1. If not available, scan and/or resize
665	>	int k = (workerCounts >>> TOTAL_COUNT_SHIFT) - 1;
666	>	final ReentrantLock lock = this.workerLock;
667	>	lock.lock();
668	>	try {
669	>	ForkJoinWorkerThread[] ws = workers;
670	>	int n = ws.length;
671	>	if (k < 0 || k >= n || ws[k] != null) {
672	>	for (k = 0; k < n && ws[k] != null; ++k)
673	>	;
674	>	if (k == n)
675	>	ws = Arrays.copyOf(ws, n << 1);
676	>	}
677	>	ws[k] = w;
678	>	workers = ws; // volatile array write ensures slot visibility
679	>	} finally {
680	>	lock.unlock();
681		}
682	+	return k;
683		}
684
685		/**
686	<	* Controls whether to add spares to maintain parallelism
686	>	* Nulls out record of worker in workers array
687		*/
688	<	private volatile boolean maintainsParallelism;
688	>	private void forgetWorker(ForkJoinWorkerThread w) {
689	>	int idx = w.poolIndex;
690	>	// Locking helps method recordWorker avoid unecessary expansion
691	>	final ReentrantLock lock = this.workerLock;
692	>	lock.lock();
693	>	try {
694	>	ForkJoinWorkerThread[] ws = workers;
695	>	if (idx >= 0 && idx < ws.length && ws[idx] == w) // verify
696	>	ws[idx] = null;
697	>	} finally {
698	>	lock.unlock();
699	>	}
700	>	}
701
702	<	// Constructors
702	>	// adding and removing workers
703
704		/**
705	<	* Creates a ForkJoinPool with a pool size equal to the number of
706	<	* processors available on the system, using the default
707	<	* ForkJoinWorkerThreadFactory.
348	<	*
349	<	* @throws SecurityException if a security manager exists and
350	<	*         the caller is not permitted to modify threads
351	<	*         because it does not hold {@link
352	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
705	>	* Tries to create and add new worker. Assumes that worker counts
706	>	* are already updated to accommodate the worker, so adjusts on
707	>	* failure.
708		*/
709	<	public ForkJoinPool() {
710	<	this(Runtime.getRuntime().availableProcessors(),
711	<	defaultForkJoinWorkerThreadFactory);
709	>	private void addWorker() {
710	>	ForkJoinWorkerThread w = null;
711	>	try {
712	>	w = factory.newThread(this);
713	>	} finally { // Adjust on either null or exceptional factory return
714	>	if (w == null) {
715	>	decrementWorkerCounts(ONE_RUNNING, ONE_TOTAL);
716	>	tryTerminate(false); // in case of failure during shutdown
717	>	}
718	>	}
719	>	if (w != null)
720	>	w.start(recordWorker(w), ueh);
721		}
722
723		/**
724	<	* Creates a ForkJoinPool with the indicated parallelism level
725	<	* threads and using the default ForkJoinWorkerThreadFactory.
724	>	* Final callback from terminating worker.  Removes record of
725	>	* worker from array, and adjusts counts. If pool is shutting
726	>	* down, tries to complete terminatation.
727		*
728	<	* @param parallelism the number of worker threads
364	<	* @throws IllegalArgumentException if parallelism less than or
365	<	* equal to zero
366	<	* @throws SecurityException if a security manager exists and
367	<	*         the caller is not permitted to modify threads
368	<	*         because it does not hold {@link
369	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
728	>	* @param w the worker
729		*/
730	<	public ForkJoinPool(int parallelism) {
731	<	this(parallelism, defaultForkJoinWorkerThreadFactory);
730	>	final void workerTerminated(ForkJoinWorkerThread w) {
731	>	forgetWorker(w);
732	>	decrementWorkerCounts(w.isTrimmed()? 0 : ONE_RUNNING, ONE_TOTAL);
733	>	while (w.stealCount != 0) // collect final count
734	>	tryAccumulateStealCount(w);
735	>	tryTerminate(false);
736		}
737
738	+	// Waiting for and signalling events
739	+
740		/**
741	<	* Creates a ForkJoinPool with parallelism equal to the number of
742	<	* processors available on the system and using the given
743	<	* ForkJoinWorkerThreadFactory.
741	>	* Releases workers blocked on a count not equal to current count.
742	>	* Normally called after precheck that eventWaiters isn't zero to
743	>	* avoid wasted array checks.
744		*
745	<	* @param factory the factory for creating new threads
746	<	* @throws NullPointerException if factory is null
747	<	* @throws SecurityException if a security manager exists and
383	<	*         the caller is not permitted to modify threads
384	<	*         because it does not hold {@link
385	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
745	>	* @param signalling true if caller is a signalling worker so can
746	>	* exit upon (conservatively) detected contention by other threads
747	>	* who will continue to release
748		*/
749	<	public ForkJoinPool(ForkJoinWorkerThreadFactory factory) {
750	<	this(Runtime.getRuntime().availableProcessors(), factory);
749	>	private void releaseEventWaiters(boolean signalling) {
750	>	ForkJoinWorkerThread[] ws = workers;
751	>	int n = ws.length;
752	>	long h; // head of stack
753	>	ForkJoinWorkerThread w; int id, ec;
754	>	while ((id = ((int)((h = eventWaiters) & WAITER_ID_MASK)) - 1) >= 0 &&
755	>	(int)(h >>> EVENT_COUNT_SHIFT) != (ec = eventCount) &&
756	>	id < n && (w = ws[id]) != null) {
757	>	if (UNSAFE.compareAndSwapLong(this, eventWaitersOffset,
758	>	h, h = w.nextWaiter))
759	>	LockSupport.unpark(w);
760	>	if (signalling && (eventCount != ec || eventWaiters != h))
761	>	break;
762	>	}
763		}
764
765		/**
766	<	* Creates a ForkJoinPool with the given parallelism and factory.
767	<	*
394	<	* @param parallelism the targeted number of worker threads
395	<	* @param factory the factory for creating new threads
396	<	* @throws IllegalArgumentException if parallelism less than or
397	<	* equal to zero, or greater than implementation limit
398	<	* @throws NullPointerException if factory is null
399	<	* @throws SecurityException if a security manager exists and
400	<	*         the caller is not permitted to modify threads
401	<	*         because it does not hold {@link
402	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
766	>	* Tries to advance eventCount and releases waiters. Called only
767	>	* from workers.
768		*/
769	<	public ForkJoinPool(int parallelism, ForkJoinWorkerThreadFactory factory) {
770	<	if (parallelism <= 0 || parallelism > MAX_THREADS)
771	<	throw new IllegalArgumentException();
772	<	if (factory == null)
773	<	throw new NullPointerException();
409	<	checkPermission();
410	<	this.factory = factory;
411	<	this.parallelism = parallelism;
412	<	this.maxPoolSize = MAX_THREADS;
413	<	this.maintainsParallelism = true;
414	<	this.poolNumber = poolNumberGenerator.incrementAndGet();
415	<	this.workerLock = new ReentrantLock();
416	<	this.termination = workerLock.newCondition();
417	<	this.stealCount = new AtomicLong();
418	<	this.submissionQueue = new LinkedTransferQueue<ForkJoinTask<?>>();
419	<	// worker array and workers are lazily constructed
769	>	final void signalWork() {
770	>	int c; // try to increment event count -- CAS failure OK
771	>	UNSAFE.compareAndSwapInt(this, eventCountOffset, c = eventCount, c+1);
772	>	if (eventWaiters != 0L)
773	>	releaseEventWaiters(true);
774		}
775
776		/**
777	<	* Creates a new worker thread using factory.
777	>	* Blocks worker until terminating or event count
778	>	* advances from last value held by worker
779		*
780	<	* @param index the index to assign worker
426	<	* @return new worker, or null of factory failed
780	>	* @param w the calling worker thread
781		*/
782	<	private ForkJoinWorkerThread createWorker(int index) {
783	<	Thread.UncaughtExceptionHandler h = ueh;
784	<	ForkJoinWorkerThread w = factory.newThread(this);
785	<	if (w != null) {
786	<	w.poolIndex = index;
787	<	w.setDaemon(true);
788	<	w.setAsyncMode(locallyFifo);
789	<	w.setName("ForkJoinPool-" + poolNumber + "-worker-" + index);
790	<	if (h != null)
791	<	w.setUncaughtExceptionHandler(h);
782	>	private void eventSync(ForkJoinWorkerThread w) {
783	>	int wec = w.lastEventCount;
784	>	long nh = (((long)wec) << EVENT_COUNT_SHIFT) | ((long)(w.poolIndex+1));
785	>	long h;
786	>	while ((runState < SHUTDOWN || !tryTerminate(false)) &&
787	>	((h = eventWaiters) == 0L ||
788	>	(int)(h >>> EVENT_COUNT_SHIFT) == wec) &&
789	>	eventCount == wec) {
790	>	if (UNSAFE.compareAndSwapLong(this, eventWaitersOffset,
791	>	w.nextWaiter = h, nh)) {
792	>	while (runState < TERMINATING && eventCount == wec) {
793	>	if (!tryAccumulateStealCount(w))  // transfer while idle
794	>	continue;
795	>	Thread.interrupted();             // clear/ignore interrupt
796	>	if (eventCount != wec)
797	>	break;
798	>	LockSupport.park(w);
799	>	}
800	>	break;
801	>	}
802		}
803	<	return w;
803	>	w.lastEventCount = eventCount;
804		}
805
806	+	// Maintaining spares
807	+
808		/**
809	<	* Returns a good size for worker array given pool size.
444	<	* Currently requires size to be a power of two.
809	>	* Pushes worker onto the spare stack
810		*/
811	<	private static int arraySizeFor(int poolSize) {
812	<	return (poolSize <= 1) ? 1 :
813	<	(1 << (32 - Integer.numberOfLeadingZeros(poolSize-1)));
811	>	final void pushSpare(ForkJoinWorkerThread w) {
812	>	int ns = (++w.spareCount << SPARE_COUNT_SHIFT) | (w.poolIndex+1);
813	>	do {} while (!UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
814	>	w.nextSpare = spareWaiters,ns));
815		}
816
817		/**
818	<	* Creates or resizes array if necessary to hold newLength.
819	<	* Call only under exclusion.
454	<	*
455	<	* @return the array
818	>	* Tries (once) to resume a spare if running count is less than
819	>	* target parallelism. Fails on contention or stale workers.
820		*/
821	<	private ForkJoinWorkerThread[] ensureWorkerArrayCapacity(int newLength) {
822	<	ForkJoinWorkerThread[] ws = workers;
823	<	if (ws == null)
824	<	return workers = new ForkJoinWorkerThread[arraySizeFor(newLength)];
825	<	else if (newLength > ws.length)
826	<	return workers = Arrays.copyOf(ws, arraySizeFor(newLength));
827	<	else
828	<	return ws;
821	>	private void tryResumeSpare() {
822	>	int sw, id;
823	>	ForkJoinWorkerThread w;
824	>	ForkJoinWorkerThread[] ws;
825	>	if ((id = ((sw = spareWaiters) & SPARE_ID_MASK) - 1) >= 0 &&
826	>	id < (ws = workers).length && (w = ws[id]) != null &&
827	>	(workerCounts & RUNNING_COUNT_MASK) < parallelism &&
828	>	eventWaiters == 0L &&
829	>	spareWaiters == sw &&
830	>	UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
831	>	sw, w.nextSpare) &&
832	>	w.tryUnsuspend()) {
833	>	int c; // try increment; if contended, finish after unpark
834	>	boolean inc = UNSAFE.compareAndSwapInt(this, workerCountsOffset,
835	>	c = workerCounts,
836	>	c + ONE_RUNNING);
837	>	LockSupport.unpark(w);
838	>	if (!inc) {
839	>	do {} while(!UNSAFE.compareAndSwapInt(this, workerCountsOffset,
840	>	c = workerCounts,
841	>	c + ONE_RUNNING));
842	>	}
843	>	}
844		}
845
846		/**
847	<	* Tries to shrink workers into smaller array after one or more terminate.
847	>	* Callback from oldest spare occasionally waking up.  Tries
848	>	* (once) to shutdown a spare if more than 25% spare overage, or
849	>	* if UNUSED_SPARE_TRIM_RATE_NANOS have elapsed and there are at
850	>	* least #parallelism running threads. Note that we don't need CAS
851	>	* or locks here because the method is called only from the oldest
852	>	* suspended spare occasionally waking (and even misfires are OK).
853	>	*
854	>	* @param now the wake up nanoTime of caller
855		*/
856	<	private void tryShrinkWorkerArray() {
857	<	ForkJoinWorkerThread[] ws = workers;
858	<	if (ws != null) {
859	<	int len = ws.length;
860	<	int last = len - 1;
861	<	while (last >= 0 && ws[last] == null)
862	<	--last;
863	<	int newLength = arraySizeFor(last+1);
864	<	if (newLength < len)
865	<	workers = Arrays.copyOf(ws, newLength);
856	>	final void tryTrimSpare(long now) {
857	>	long lastTrim = trimTime;
858	>	trimTime = now;
859	>	helpMaintainParallelism(); // first, help wake up any needed spares
860	>	int sw, id;
861	>	ForkJoinWorkerThread w;
862	>	ForkJoinWorkerThread[] ws;
863	>	int pc = parallelism;
864	>	int wc = workerCounts;
865	>	if ((wc & RUNNING_COUNT_MASK) >= pc &&
866	>	(((wc >>> TOTAL_COUNT_SHIFT) - pc) > (pc >>> 2) + 1 ||// approx 25%
867	>	now - lastTrim >= UNUSED_SPARE_TRIM_RATE_NANOS) &&
868	>	(id = ((sw = spareWaiters) & SPARE_ID_MASK) - 1) >= 0 &&
869	>	id < (ws = workers).length && (w = ws[id]) != null &&
870	>	UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
871	>	sw, w.nextSpare))
872	>	w.shutdown(false);
873	>	}
874	>
875	>	/**
876	>	* Does at most one of:
877	>	*
878	>	* 1. Help wake up existing workers waiting for work via
879	>	*    releaseEventWaiters. (If any exist, then it probably doesn't
880	>	*    matter right now if under target parallelism level.)
881	>	*
882	>	* 2. If below parallelism level and a spare exists, try (once)
883	>	*    to resume it via tryResumeSpare.
884	>	*
885	>	* 3. If neither of the above, tries (once) to add a new
886	>	*    worker if either there are not enough total, or if all
887	>	*    existing workers are busy, there are either no running
888	>	*    workers or the deficit is at least twice the surplus.
889	>	*/
890	>	private void helpMaintainParallelism() {
891	>	// uglified to work better when not compiled
892	>	int pc, wc, rc, tc, rs; long h;
893	>	if ((h = eventWaiters) != 0L) {
894	>	if ((int)(h >>> EVENT_COUNT_SHIFT) != eventCount)
895	>	releaseEventWaiters(false); // avoid useless call
896	>	}
897	>	else if ((pc = parallelism) >
898	>	(rc = ((wc = workerCounts) & RUNNING_COUNT_MASK))) {
899	>	if (spareWaiters != 0)
900	>	tryResumeSpare();
901	>	else if ((rs = runState) < TERMINATING &&
902	>	((tc = wc >>> TOTAL_COUNT_SHIFT) < pc ||
903	>	(tc == (rs & ACTIVE_COUNT_MASK) && // all busy
904	>	(rc == 0 ||                       // must add
905	>	rc < pc - ((tc - pc) << 1)) &&   // within slack
906	>	tc < MAX_WORKERS && runState == rs)) && // recheck busy
907	>	workerCounts == wc &&
908	>	UNSAFE.compareAndSwapInt(this, workerCountsOffset, wc,
909	>	wc + (ONE_RUNNING|ONE_TOTAL)))
910	>	addWorker();
911	>	}
912	>	}
913	>
914	>	/**
915	>	* Callback from workers invoked upon each top-level action (i.e.,
916	>	* stealing a task or taking a submission and running
917	>	* it). Performs one or more of the following:
918	>	*
919	>	* 1. If the worker cannot find work (misses > 0), updates its
920	>	*    active status to inactive and updates activeCount unless
921	>	*    this is the first miss and there is contention, in which
922	>	*    case it may try again (either in this or a subsequent
923	>	*    call).
924	>	*
925	>	* 2. If there are at least 2 misses, awaits the next task event
926	>	*    via eventSync
927	>	*
928	>	* 3. If there are too many running threads, suspends this worker
929	>	*    (first forcing inactivation if necessary).  If it is not
930	>	*    needed, it may be killed while suspended via
931	>	*    tryTrimSpare. Otherwise, upon resume it rechecks to make
932	>	*    sure that it is still needed.
933	>	*
934	>	* 4. Helps release and/or reactivate other workers via
935	>	*    helpMaintainParallelism
936	>	*
937	>	* @param w the worker
938	>	* @param misses the number of scans by caller failing to find work
939	>	* (saturating at 2 just to avoid wraparound)
940	>	*/
941	>	final void preStep(ForkJoinWorkerThread w, int misses) {
942	>	boolean active = w.active;
943	>	int pc = parallelism;
944	>	for (;;) {
945	>	int wc = workerCounts;
946	>	int rc = wc & RUNNING_COUNT_MASK;
947	>	if (active && (misses > 0 || rc > pc)) {
948	>	int rs;                      // try inactivate
949	>	if (UNSAFE.compareAndSwapInt(this, runStateOffset,
950	>	rs = runState, rs - ONE_ACTIVE))
951	>	active = w.active = false;
952	>	else if (misses > 1 || rc > pc ||
953	>	(rs & ACTIVE_COUNT_MASK) >= pc)
954	>	continue;                // force inactivate
955	>	}
956	>	if (misses > 1) {
957	>	misses = 0;                  // don't re-sync
958	>	eventSync(w);                // continue loop to recheck rc
959	>	}
960	>	else if (rc > pc) {
961	>	if (workerCounts == wc &&   // try to suspend as spare
962	>	UNSAFE.compareAndSwapInt(this, workerCountsOffset,
963	>	wc, wc - ONE_RUNNING) &&
964	>	!w.suspendAsSpare())    // false if killed
965	>	break;
966	>	}
967	>	else {
968	>	if (rc < pc || eventWaiters != 0L)
969	>	helpMaintainParallelism();
970	>	break;
971	>	}
972		}
973		}
974
975		/**
976	<	* Initializes workers if necessary.
977	<	*/
978	<	final void ensureWorkerInitialization() {
979	<	ForkJoinWorkerThread[] ws = workers;
980	<	if (ws == null) {
981	<	final ReentrantLock lock = this.workerLock;
982	<	lock.lock();
983	<	try {
984	<	ws = workers;
985	<	if (ws == null) {
986	<	int ps = parallelism;
987	<	ws = ensureWorkerArrayCapacity(ps);
988	<	for (int i = 0; i < ps; ++i) {
989	<	ForkJoinWorkerThread w = createWorker(i);
990	<	if (w != null) {
991	<	ws[i] = w;
992	<	w.start();
993	<	updateWorkerCount(1);
994	<	}
995	<	}
996	<	}
997	<	} finally {
998	<	lock.unlock();
976	>	* Helps and/or blocks awaiting join of the given task.
977	>	* Alternates between helpJoinTask() and helpMaintainParallelism()
978	>	* as many times as there is a deficit in running count (or longer
979	>	* if running count would become zero), then blocks if task still
980	>	* not done.
981	>	*
982	>	* @param joinMe the task to join
983	>	*/
984	>	final void awaitJoin(ForkJoinTask<?> joinMe, ForkJoinWorkerThread worker) {
985	>	int threshold = parallelism;         // descend blocking thresholds
986	>	while (joinMe.status >= 0) {
987	>	boolean block; int wc;
988	>	worker.helpJoinTask(joinMe);
989	>	if (joinMe.status < 0)
990	>	break;
991	>	if (((wc = workerCounts) & RUNNING_COUNT_MASK) <= threshold) {
992	>	if (threshold > 0)
993	>	--threshold;
994	>	else
995	>	advanceEventCount(); // force release
996	>	block = false;
997	>	}
998	>	else
999	>	block = UNSAFE.compareAndSwapInt(this, workerCountsOffset,
1000	>	wc, wc - ONE_RUNNING);
1001	>	helpMaintainParallelism();
1002	>	if (block) {
1003	>	int c;
1004	>	joinMe.internalAwaitDone();
1005	>	do {} while (!UNSAFE.compareAndSwapInt
1006	>	(this, workerCountsOffset,
1007	>	c = workerCounts, c + ONE_RUNNING));
1008	>	break;
1009		}
1010		}
1011		}
1012
1013		/**
1014	<	* Worker creation and startup for threads added via setParallelism.
1014	>	* Same idea as awaitJoin, but no helping
1015		*/
1016	<	private void createAndStartAddedWorkers() {
1017	<	resumeAllSpares();  // Allow spares to convert to nonspare
1018	<	int ps = parallelism;
1019	<	ForkJoinWorkerThread[] ws = ensureWorkerArrayCapacity(ps);
1020	<	int len = ws.length;
1021	<	// Sweep through slots, to keep lowest indices most populated
1022	<	int k = 0;
1023	<	while (k < len) {
1024	<	if (ws[k] != null) {
1025	<	++k;
1026	<	continue;
1016	>	final void awaitBlocker(ManagedBlocker blocker)
1017	>	throws InterruptedException {
1018	>	int threshold = parallelism;
1019	>	while (!blocker.isReleasable()) {
1020	>	boolean block; int wc;
1021	>	if (((wc = workerCounts) & RUNNING_COUNT_MASK) <= threshold) {
1022	>	if (threshold > 0)
1023	>	--threshold;
1024	>	else
1025	>	advanceEventCount();
1026	>	block = false;
1027		}
1028	<	int s = workerCounts;
1029	<	int tc = totalCountOf(s);
1030	<	int rc = runningCountOf(s);
1031	<	if (rc >= ps || tc >= ps)
1028	>	else
1029	>	block = UNSAFE.compareAndSwapInt(this, workerCountsOffset,
1030	>	wc, wc - ONE_RUNNING);
1031	>	helpMaintainParallelism();
1032	>	if (block) {
1033	>	try {
1034	>	do {} while (!blocker.isReleasable() && !blocker.block());
1035	>	} finally {
1036	>	int c;
1037	>	do {} while (!UNSAFE.compareAndSwapInt
1038	>	(this, workerCountsOffset,
1039	>	c = workerCounts, c + ONE_RUNNING));
1040	>	}
1041		break;
1042	<	if (casWorkerCounts (s, workerCountsFor(tc+1, rc+1))) {
1043	<	ForkJoinWorkerThread w = createWorker(k);
1042	>	}
1043	>	}
1044	>	}
1045	>
1046	>	/**
1047	>	* Possibly initiates and/or completes termination.
1048	>	*
1049	>	* @param now if true, unconditionally terminate, else only
1050	>	* if shutdown and empty queue and no active workers
1051	>	* @return true if now terminating or terminated
1052	>	*/
1053	>	private boolean tryTerminate(boolean now) {
1054	>	if (now)
1055	>	advanceRunLevel(SHUTDOWN); // ensure at least SHUTDOWN
1056	>	else if (runState < SHUTDOWN ||
1057	>	!submissionQueue.isEmpty() ||
1058	>	(runState & ACTIVE_COUNT_MASK) != 0)
1059	>	return false;
1060	>
1061	>	if (advanceRunLevel(TERMINATING))
1062	>	startTerminating();
1063	>
1064	>	// Finish now if all threads terminated; else in some subsequent call
1065	>	if ((workerCounts >>> TOTAL_COUNT_SHIFT) == 0) {
1066	>	advanceRunLevel(TERMINATED);
1067	>	termination.arrive();
1068	>	}
1069	>	return true;
1070	>	}
1071	>
1072	>	/**
1073	>	* Actions on transition to TERMINATING
1074	>	*
1075	>	* Runs up to four passes through workers: (0) shutting down each
1076	>	* quietly (without waking up if parked) to quickly spread
1077	>	* notifications without unnecessary bouncing around event queues
1078	>	* etc (1) wake up and help cancel tasks (2) interrupt (3) mop up
1079	>	* races with interrupted workers
1080	>	*/
1081	>	private void startTerminating() {
1082	>	cancelSubmissions();
1083	>	for (int passes = 0; passes < 4 && workerCounts != 0; ++passes) {
1084	>	advanceEventCount();
1085	>	eventWaiters = 0L; // clobber lists
1086	>	spareWaiters = 0;
1087	>	ForkJoinWorkerThread[] ws = workers;
1088	>	int n = ws.length;
1089	>	for (int i = 0; i < n; ++i) {
1090	>	ForkJoinWorkerThread w = ws[i];
1091		if (w != null) {
1092	<	ws[k++] = w;
1093	<	w.start();
1094	<	}
1095	<	else {
1096	<	updateWorkerCount(-1); // back out on failed creation
1097	<	break;
1092	>	w.shutdown(true);
1093	>	if (passes > 0 && !w.isTerminated()) {
1094	>	w.cancelTasks();
1095	>	LockSupport.unpark(w);
1096	>	if (passes > 1) {
1097	>	try {
1098	>	w.interrupt();
1099	>	} catch (SecurityException ignore) {
1100	>	}
1101	>	}
1102	>	}
1103		}
1104		}
1105		}
1106		}
1107
1108	+	/**
1109	+	* Clear out and cancel submissions, ignoring exceptions
1110	+	*/
1111	+	private void cancelSubmissions() {
1112	+	ForkJoinTask<?> task;
1113	+	while ((task = submissionQueue.poll()) != null) {
1114	+	try {
1115	+	task.cancel(false);
1116	+	} catch (Throwable ignore) {
1117	+	}
1118	+	}
1119	+	}
1120	+
1121	+	// misc support for ForkJoinWorkerThread
1122	+
1123	+	/**
1124	+	* Returns pool number
1125	+	*/
1126	+	final int getPoolNumber() {
1127	+	return poolNumber;
1128	+	}
1129	+
1130	+	/**
1131	+	* Tries to accumulates steal count from a worker, clearing
1132	+	* the worker's value.
1133	+	*
1134	+	* @return true if worker steal count now zero
1135	+	*/
1136	+	final boolean tryAccumulateStealCount(ForkJoinWorkerThread w) {
1137	+	int sc = w.stealCount;
1138	+	long c = stealCount;
1139	+	// CAS even if zero, for fence effects
1140	+	if (UNSAFE.compareAndSwapLong(this, stealCountOffset, c, c + sc)) {
1141	+	if (sc != 0)
1142	+	w.stealCount = 0;
1143	+	return true;
1144	+	}
1145	+	return sc == 0;
1146	+	}
1147	+
1148	+	/**
1149	+	* Returns the approximate (non-atomic) number of idle threads per
1150	+	* active thread.
1151	+	*/
1152	+	final int idlePerActive() {
1153	+	int pc = parallelism; // use parallelism, not rc
1154	+	int ac = runState;    // no mask -- artifically boosts during shutdown
1155	+	// Use exact results for small values, saturate past 4
1156	+	return pc <= ac? 0 : pc >>> 1 <= ac? 1 : pc >>> 2 <= ac? 3 : pc >>> 3;
1157	+	}
1158	+
1159	+	// Public and protected methods
1160	+
1161	+	// Constructors
1162	+
1163	+	/**
1164	+	* Creates a {@code ForkJoinPool} with parallelism equal to {@link
1165	+	* java.lang.Runtime#availableProcessors}, using the {@linkplain
1166	+	* #defaultForkJoinWorkerThreadFactory default thread factory},
1167	+	* no UncaughtExceptionHandler, and non-async LIFO processing mode.
1168	+	*
1169	+	* @throws SecurityException if a security manager exists and
1170	+	*         the caller is not permitted to modify threads
1171	+	*         because it does not hold {@link
1172	+	*         java.lang.RuntimePermission}{@code ("modifyThread")}
1173	+	*/
1174	+	public ForkJoinPool() {
1175	+	this(Runtime.getRuntime().availableProcessors(),
1176	+	defaultForkJoinWorkerThreadFactory, null, false);
1177	+	}
1178	+
1179	+	/**
1180	+	* Creates a {@code ForkJoinPool} with the indicated parallelism
1181	+	* level, the {@linkplain
1182	+	* #defaultForkJoinWorkerThreadFactory default thread factory},
1183	+	* no UncaughtExceptionHandler, and non-async LIFO processing mode.
1184	+	*
1185	+	* @param parallelism the parallelism level
1186	+	* @throws IllegalArgumentException if parallelism less than or
1187	+	*         equal to zero, or greater than implementation limit
1188	+	* @throws SecurityException if a security manager exists and
1189	+	*         the caller is not permitted to modify threads
1190	+	*         because it does not hold {@link
1191	+	*         java.lang.RuntimePermission}{@code ("modifyThread")}
1192	+	*/
1193	+	public ForkJoinPool(int parallelism) {
1194	+	this(parallelism, defaultForkJoinWorkerThreadFactory, null, false);
1195	+	}
1196	+
1197	+	/**
1198	+	* Creates a {@code ForkJoinPool} with the given parameters.
1199	+	*
1200	+	* @param parallelism the parallelism level. For default value,
1201	+	* use {@link java.lang.Runtime#availableProcessors}.
1202	+	* @param factory the factory for creating new threads. For default value,
1203	+	* use {@link #defaultForkJoinWorkerThreadFactory}.
1204	+	* @param handler the handler for internal worker threads that
1205	+	* terminate due to unrecoverable errors encountered while executing
1206	+	* tasks. For default value, use <code>null</code>.
1207	+	* @param asyncMode if true,
1208	+	* establishes local first-in-first-out scheduling mode for forked
1209	+	* tasks that are never joined. This mode may be more appropriate
1210	+	* than default locally stack-based mode in applications in which
1211	+	* worker threads only process event-style asynchronous tasks.
1212	+	* For default value, use <code>false</code>.
1213	+	* @throws IllegalArgumentException if parallelism less than or
1214	+	*         equal to zero, or greater than implementation limit
1215	+	* @throws NullPointerException if the factory is null
1216	+	* @throws SecurityException if a security manager exists and
1217	+	*         the caller is not permitted to modify threads
1218	+	*         because it does not hold {@link
1219	+	*         java.lang.RuntimePermission}{@code ("modifyThread")}
1220	+	*/
1221	+	public ForkJoinPool(int parallelism,
1222	+	ForkJoinWorkerThreadFactory factory,
1223	+	Thread.UncaughtExceptionHandler handler,
1224	+	boolean asyncMode) {
1225	+	checkPermission();
1226	+	if (factory == null)
1227	+	throw new NullPointerException();
1228	+	if (parallelism <= 0 || parallelism > MAX_WORKERS)
1229	+	throw new IllegalArgumentException();
1230	+	this.parallelism = parallelism;
1231	+	this.factory = factory;
1232	+	this.ueh = handler;
1233	+	this.locallyFifo = asyncMode;
1234	+	int arraySize = initialArraySizeFor(parallelism);
1235	+	this.workers = new ForkJoinWorkerThread[arraySize];
1236	+	this.submissionQueue = new LinkedTransferQueue<ForkJoinTask<?>>();
1237	+	this.workerLock = new ReentrantLock();
1238	+	this.termination = new Phaser(1);
1239	+	this.poolNumber = poolNumberGenerator.incrementAndGet();
1240	+	this.trimTime = System.nanoTime();
1241	+	}
1242	+
1243	+	/**
1244	+	* Returns initial power of two size for workers array.
1245	+	* @param pc the initial parallelism level
1246	+	*/
1247	+	private static int initialArraySizeFor(int pc) {
1248	+	// See Hackers Delight, sec 3.2. We know MAX_WORKERS < (1 >>> 16)
1249	+	int size = pc < MAX_WORKERS ? pc + 1 : MAX_WORKERS;
1250	+	size |= size >>> 1;
1251	+	size |= size >>> 2;
1252	+	size |= size >>> 4;
1253	+	size |= size >>> 8;
1254	+	return size + 1;
1255	+	}
1256	+
1257		// Execution methods
1258
1259		/**
#	Line 550 | Line 1262 | public class ForkJoinPool extends Abstra
1262		private <T> void doSubmit(ForkJoinTask<T> task) {
1263		if (task == null)
1264		throw new NullPointerException();
1265	<	if (isShutdown())
1265	>	if (runState >= SHUTDOWN)
1266		throw new RejectedExecutionException();
555	–	if (workers == null)
556	–	ensureWorkerInitialization();
1267		submissionQueue.offer(task);
1268	<	signalIdleWorkers();
1268	>	advanceEventCount();
1269	>	helpMaintainParallelism();         // start or wake up workers
1270		}
1271
1272		/**
1273		* Performs the given task, returning its result upon completion.
1274	+	* If the caller is already engaged in a fork/join computation in
1275	+	* the current pool, this method is equivalent in effect to
1276	+	* {@link ForkJoinTask#invoke}.
1277		*
1278		* @param task the task
1279		* @return the task's result
1280	<	* @throws NullPointerException if task is null
1281	<	* @throws RejectedExecutionException if pool is shut down
1280	>	* @throws NullPointerException if the task is null
1281	>	* @throws RejectedExecutionException if the task cannot be
1282	>	*         scheduled for execution
1283		*/
1284		public <T> T invoke(ForkJoinTask<T> task) {
1285		doSubmit(task);
#	Line 573 | Line 1288 | public class ForkJoinPool extends Abstra
1288
1289		/**
1290		* Arranges for (asynchronous) execution of the given task.
1291	+	* If the caller is already engaged in a fork/join computation in
1292	+	* the current pool, this method is equivalent in effect to
1293	+	* {@link ForkJoinTask#fork}.
1294		*
1295		* @param task the task
1296	<	* @throws NullPointerException if task is null
1297	<	* @throws RejectedExecutionException if pool is shut down
1296	>	* @throws NullPointerException if the task is null
1297	>	* @throws RejectedExecutionException if the task cannot be
1298	>	*         scheduled for execution
1299		*/
1300	<	public <T> void execute(ForkJoinTask<T> task) {
1300	>	public void execute(ForkJoinTask<?> task) {
1301		doSubmit(task);
1302		}
1303
1304		// AbstractExecutorService methods
1305
1306	+	/**
1307	+	* @throws NullPointerException if the task is null
1308	+	* @throws RejectedExecutionException if the task cannot be
1309	+	*         scheduled for execution
1310	+	*/
1311		public void execute(Runnable task) {
1312		ForkJoinTask<?> job;
1313		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
#	Line 593 | Line 1317 | public class ForkJoinPool extends Abstra
1317		doSubmit(job);
1318		}
1319
1320	+	/**
1321	+	* Submits a ForkJoinTask for execution.
1322	+	* If the caller is already engaged in a fork/join computation in
1323	+	* the current pool, this method is equivalent in effect to
1324	+	* {@link ForkJoinTask#fork}.
1325	+	*
1326	+	* @param task the task to submit
1327	+	* @return the task
1328	+	* @throws NullPointerException if the task is null
1329	+	* @throws RejectedExecutionException if the task cannot be
1330	+	*         scheduled for execution
1331	+	*/
1332	+	public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
1333	+	doSubmit(task);
1334	+	return task;
1335	+	}
1336	+
1337	+	/**
1338	+	* @throws NullPointerException if the task is null
1339	+	* @throws RejectedExecutionException if the task cannot be
1340	+	*         scheduled for execution
1341	+	*/
1342		public <T> ForkJoinTask<T> submit(Callable<T> task) {
1343		ForkJoinTask<T> job = ForkJoinTask.adapt(task);
1344		doSubmit(job);
1345		return job;
1346		}
1347
1348	+	/**
1349	+	* @throws NullPointerException if the task is null
1350	+	* @throws RejectedExecutionException if the task cannot be
1351	+	*         scheduled for execution
1352	+	*/
1353		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
1354		ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
1355		doSubmit(job);
1356		return job;
1357		}
1358
1359	+	/**
1360	+	* @throws NullPointerException if the task is null
1361	+	* @throws RejectedExecutionException if the task cannot be
1362	+	*         scheduled for execution
1363	+	*/
1364		public ForkJoinTask<?> submit(Runnable task) {
1365		ForkJoinTask<?> job;
1366		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
#	Line 616 | Line 1372 | public class ForkJoinPool extends Abstra
1372		}
1373
1374		/**
1375	<	* Submits a ForkJoinTask for execution.
1376	<	*
621	<	* @param task the task to submit
622	<	* @return the task
623	<	* @throws RejectedExecutionException if the task cannot be
624	<	*         scheduled for execution
625	<	* @throws NullPointerException if the task is null
1375	>	* @throws NullPointerException       {@inheritDoc}
1376	>	* @throws RejectedExecutionException {@inheritDoc}
1377		*/
627	–	public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
628	–	doSubmit(task);
629	–	return task;
630	–	}
631	–
632	–
1378		public <T> List<Future<T>> invokeAll(Collection<? extends Callable<T>> tasks) {
1379		ArrayList<ForkJoinTask<T>> forkJoinTasks =
1380		new ArrayList<ForkJoinTask<T>>(tasks.size());
#	Line 638 | Line 1383 | public class ForkJoinPool extends Abstra
1383		invoke(new InvokeAll<T>(forkJoinTasks));
1384
1385		@SuppressWarnings({"unchecked", "rawtypes"})
1386	<	List<Future<T>> futures = (List<Future<T>>) (List) forkJoinTasks;
1386	>	List<Future<T>> futures = (List<Future<T>>) (List) forkJoinTasks;
1387		return futures;
1388		}
1389
#	Line 652 | Line 1397 | public class ForkJoinPool extends Abstra
1397		private static final long serialVersionUID = -7914297376763021607L;
1398		}
1399
655	–	// Configuration and status settings and queries
656	–
1400		/**
1401		* Returns the factory used for constructing new workers.
1402		*
#	Line 670 | Line 1413 | public class ForkJoinPool extends Abstra
1413		* @return the handler, or {@code null} if none
1414		*/
1415		public Thread.UncaughtExceptionHandler getUncaughtExceptionHandler() {
1416	<	Thread.UncaughtExceptionHandler h;
674	<	final ReentrantLock lock = this.workerLock;
675	<	lock.lock();
676	<	try {
677	<	h = ueh;
678	<	} finally {
679	<	lock.unlock();
680	<	}
681	<	return h;
1416	>	return ueh;
1417		}
1418
1419		/**
1420	<	* Sets the handler for internal worker threads that terminate due
686	<	* to unrecoverable errors encountered while executing tasks.
687	<	* Unless set, the current default or ThreadGroup handler is used
688	<	* as handler.
1420	>	* Returns the targeted parallelism level of this pool.
1421		*
1422	<	* @param h the new handler
691	<	* @return the old handler, or {@code null} if none
692	<	* @throws SecurityException if a security manager exists and
693	<	*         the caller is not permitted to modify threads
694	<	*         because it does not hold {@link
695	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
696	<	*/
697	<	public Thread.UncaughtExceptionHandler
698	<	setUncaughtExceptionHandler(Thread.UncaughtExceptionHandler h) {
699	<	checkPermission();
700	<	Thread.UncaughtExceptionHandler old = null;
701	<	final ReentrantLock lock = this.workerLock;
702	<	lock.lock();
703	<	try {
704	<	old = ueh;
705	<	ueh = h;
706	<	ForkJoinWorkerThread[] ws = workers;
707	<	if (ws != null) {
708	<	for (int i = 0; i < ws.length; ++i) {
709	<	ForkJoinWorkerThread w = ws[i];
710	<	if (w != null)
711	<	w.setUncaughtExceptionHandler(h);
712	<	}
713	<	}
714	<	} finally {
715	<	lock.unlock();
716	<	}
717	<	return old;
718	<	}
719	<
720	<
721	<	/**
722	<	* Sets the target parallelism level of this pool.
723	<	*
724	<	* @param parallelism the target parallelism
725	<	* @throws IllegalArgumentException if parallelism less than or
726	<	* equal to zero or greater than maximum size bounds
727	<	* @throws SecurityException if a security manager exists and
728	<	*         the caller is not permitted to modify threads
729	<	*         because it does not hold {@link
730	<	*         java.lang.RuntimePermission}{@code ("modifyThread")}
731	<	*/
732	<	public void setParallelism(int parallelism) {
733	<	checkPermission();
734	<	if (parallelism <= 0 || parallelism > maxPoolSize)
735	<	throw new IllegalArgumentException();
736	<	final ReentrantLock lock = this.workerLock;
737	<	lock.lock();
738	<	try {
739	<	if (!isTerminating()) {
740	<	int p = this.parallelism;
741	<	this.parallelism = parallelism;
742	<	if (parallelism > p)
743	<	createAndStartAddedWorkers();
744	<	else
745	<	trimSpares();
746	<	}
747	<	} finally {
748	<	lock.unlock();
749	<	}
750	<	signalIdleWorkers();
751	<	}
752	<
753	<	/**
754	<	* Returns the targeted number of worker threads in this pool.
755	<	*
756	<	* @return the targeted number of worker threads in this pool
1422	>	* @return the targeted parallelism level of this pool
1423		*/
1424		public int getParallelism() {
1425		return parallelism;
#	Line 768 | Line 1434 | public class ForkJoinPool extends Abstra
1434		* @return the number of worker threads
1435		*/
1436		public int getPoolSize() {
1437	<	return totalCountOf(workerCounts);
772	<	}
773	<
774	<	/**
775	<	* Returns the maximum number of threads allowed to exist in the
776	<	* pool, even if there are insufficient unblocked running threads.
777	<	*
778	<	* @return the maximum
779	<	*/
780	<	public int getMaximumPoolSize() {
781	<	return maxPoolSize;
782	<	}
783	<
784	<	/**
785	<	* Sets the maximum number of threads allowed to exist in the
786	<	* pool, even if there are insufficient unblocked running threads.
787	<	* Setting this value has no effect on current pool size. It
788	<	* controls construction of new threads.
789	<	*
790	<	* @throws IllegalArgumentException if negative or greater than
791	<	* internal implementation limit
792	<	*/
793	<	public void setMaximumPoolSize(int newMax) {
794	<	if (newMax < 0 || newMax > MAX_THREADS)
795	<	throw new IllegalArgumentException();
796	<	maxPoolSize = newMax;
797	<	}
798	<
799	<
800	<	/**
801	<	* Returns {@code true} if this pool dynamically maintains its
802	<	* target parallelism level. If false, new threads are added only
803	<	* to avoid possible starvation.  This setting is by default true.
804	<	*
805	<	* @return {@code true} if maintains parallelism
806	<	*/
807	<	public boolean getMaintainsParallelism() {
808	<	return maintainsParallelism;
809	<	}
810	<
811	<	/**
812	<	* Sets whether this pool dynamically maintains its target
813	<	* parallelism level. If false, new threads are added only to
814	<	* avoid possible starvation.
815	<	*
816	<	* @param enable {@code true} to maintain parallelism
817	<	*/
818	<	public void setMaintainsParallelism(boolean enable) {
819	<	maintainsParallelism = enable;
820	<	}
821	<
822	<	/**
823	<	* Establishes local first-in-first-out scheduling mode for forked
824	<	* tasks that are never joined. This mode may be more appropriate
825	<	* than default locally stack-based mode in applications in which
826	<	* worker threads only process asynchronous tasks.  This method is
827	<	* designed to be invoked only when the pool is quiescent, and
828	<	* typically only before any tasks are submitted. The effects of
829	<	* invocations at other times may be unpredictable.
830	<	*
831	<	* @param async if {@code true}, use locally FIFO scheduling
832	<	* @return the previous mode
833	<	* @see #getAsyncMode
834	<	*/
835	<	public boolean setAsyncMode(boolean async) {
836	<	boolean oldMode = locallyFifo;
837	<	locallyFifo = async;
838	<	ForkJoinWorkerThread[] ws = workers;
839	<	if (ws != null) {
840	<	for (int i = 0; i < ws.length; ++i) {
841	<	ForkJoinWorkerThread t = ws[i];
842	<	if (t != null)
843	<	t.setAsyncMode(async);
844	<	}
845	<	}
846	<	return oldMode;
1437	>	return workerCounts >>> TOTAL_COUNT_SHIFT;
1438		}
1439
1440		/**
#	Line 851 | Line 1442 | public class ForkJoinPool extends Abstra
1442		* scheduling mode for forked tasks that are never joined.
1443		*
1444		* @return {@code true} if this pool uses async mode
854	–	* @see #setAsyncMode
1445		*/
1446		public boolean getAsyncMode() {
1447		return locallyFifo;
#	Line 860 | Line 1450 | public class ForkJoinPool extends Abstra
1450		/**
1451		* Returns an estimate of the number of worker threads that are
1452		* not blocked waiting to join tasks or for other managed
1453	<	* synchronization.
1453	>	* synchronization. This method may overestimate the
1454	>	* number of running threads.
1455		*
1456		* @return the number of worker threads
1457		*/
1458		public int getRunningThreadCount() {
1459	<	return runningCountOf(workerCounts);
1459	>	return workerCounts & RUNNING_COUNT_MASK;
1460		}
1461
1462		/**
#	Line 876 | Line 1467 | public class ForkJoinPool extends Abstra
1467		* @return the number of active threads
1468		*/
1469		public int getActiveThreadCount() {
1470	<	return activeCountOf(runControl);
880	<	}
881	<
882	<	/**
883	<	* Returns an estimate of the number of threads that are currently
884	<	* idle waiting for tasks. This method may underestimate the
885	<	* number of idle threads.
886	<	*
887	<	* @return the number of idle threads
888	<	*/
889	<	final int getIdleThreadCount() {
890	<	int c = runningCountOf(workerCounts) - activeCountOf(runControl);
891	<	return (c <= 0) ? 0 : c;
1470	>	return runState & ACTIVE_COUNT_MASK;
1471		}
1472
1473		/**
#	Line 903 | Line 1482 | public class ForkJoinPool extends Abstra
1482		* @return {@code true} if all threads are currently idle
1483		*/
1484		public boolean isQuiescent() {
1485	<	return activeCountOf(runControl) == 0;
1485	>	return (runState & ACTIVE_COUNT_MASK) == 0;
1486		}
1487
1488		/**
#	Line 918 | Line 1497 | public class ForkJoinPool extends Abstra
1497		* @return the number of steals
1498		*/
1499		public long getStealCount() {
1500	<	return stealCount.get();
922	<	}
923	<
924	<	/**
925	<	* Accumulates steal count from a worker.
926	<	* Call only when worker known to be idle.
927	<	*/
928	<	private void updateStealCount(ForkJoinWorkerThread w) {
929	<	int sc = w.getAndClearStealCount();
930	<	if (sc != 0)
931	<	stealCount.addAndGet(sc);
1500	>	return stealCount;
1501		}
1502
1503		/**
#	Line 944 | Line 1513 | public class ForkJoinPool extends Abstra
1513		public long getQueuedTaskCount() {
1514		long count = 0;
1515		ForkJoinWorkerThread[] ws = workers;
1516	<	if (ws != null) {
1517	<	for (int i = 0; i < ws.length; ++i) {
1518	<	ForkJoinWorkerThread t = ws[i];
1519	<	if (t != null)
1520	<	count += t.getQueueSize();
952	<	}
1516	>	int n = ws.length;
1517	>	for (int i = 0; i < n; ++i) {
1518	>	ForkJoinWorkerThread w = ws[i];
1519	>	if (w != null)
1520	>	count += w.getQueueSize();
1521		}
1522		return count;
1523		}
1524
1525		/**
1526	<	* Returns an estimate of the number tasks submitted to this pool
1527	<	* that have not yet begun executing. This method takes time
1526	>	* Returns an estimate of the number of tasks submitted to this
1527	>	* pool that have not yet begun executing.  This method takes time
1528		* proportional to the number of submissions.
1529		*
1530		* @return the number of queued submissions
#	Line 990 | Line 1558 | public class ForkJoinPool extends Abstra
1558		* Removes all available unexecuted submitted and forked tasks
1559		* from scheduling queues and adds them to the given collection,
1560		* without altering their execution status. These may include
1561	<	* artificially generated or wrapped tasks. This method is designed
1562	<	* to be invoked only when the pool is known to be
1561	>	* artificially generated or wrapped tasks. This method is
1562	>	* designed to be invoked only when the pool is known to be
1563		* quiescent. Invocations at other times may not remove all
1564		* tasks. A failure encountered while attempting to add elements
1565		* to collection {@code c} may result in elements being in
#	Line 1004 | Line 1572 | public class ForkJoinPool extends Abstra
1572		* @return the number of elements transferred
1573		*/
1574		protected int drainTasksTo(Collection<? super ForkJoinTask<?>> c) {
1575	<	int n = submissionQueue.drainTo(c);
1575	>	int count = submissionQueue.drainTo(c);
1576		ForkJoinWorkerThread[] ws = workers;
1577	<	if (ws != null) {
1578	<	for (int i = 0; i < ws.length; ++i) {
1579	<	ForkJoinWorkerThread w = ws[i];
1580	<	if (w != null)
1581	<	n += w.drainTasksTo(c);
1014	<	}
1577	>	int n = ws.length;
1578	>	for (int i = 0; i < n; ++i) {
1579	>	ForkJoinWorkerThread w = ws[i];
1580	>	if (w != null)
1581	>	count += w.drainTasksTo(c);
1582		}
1583	<	return n;
1583	>	return count;
1584		}
1585
1586		/**
#	Line 1024 | Line 1591 | public class ForkJoinPool extends Abstra
1591		* @return a string identifying this pool, as well as its state
1592		*/
1593		public String toString() {
1027	–	int ps = parallelism;
1028	–	int wc = workerCounts;
1029	–	int rc = runControl;
1594		long st = getStealCount();
1595		long qt = getQueuedTaskCount();
1596		long qs = getQueuedSubmissionCount();
1597	+	int wc = workerCounts;
1598	+	int tc = wc >>> TOTAL_COUNT_SHIFT;
1599	+	int rc = wc & RUNNING_COUNT_MASK;
1600	+	int pc = parallelism;
1601	+	int rs = runState;
1602	+	int ac = rs & ACTIVE_COUNT_MASK;
1603		return super.toString() +
1604	<	"[" + runStateToString(runStateOf(rc)) +
1605	<	", parallelism = " + ps +
1606	<	", size = " + totalCountOf(wc) +
1607	<	", active = " + activeCountOf(rc) +
1608	<	", running = " + runningCountOf(wc) +
1604	>	"[" + runLevelToString(rs) +
1605	>	", parallelism = " + pc +
1606	>	", size = " + tc +
1607	>	", active = " + ac +
1608	>	", running = " + rc +
1609		", steals = " + st +
1610		", tasks = " + qt +
1611		", submissions = " + qs +
1612		"]";
1613		}
1614
1615	<	private static String runStateToString(int rs) {
1616	<	switch(rs) {
1617	<	case RUNNING: return "Running";
1618	<	case SHUTDOWN: return "Shutting down";
1619	<	case TERMINATING: return "Terminating";
1050	<	case TERMINATED: return "Terminated";
1051	<	default: throw new Error("Unknown run state");
1052	<	}
1615	>	private static String runLevelToString(int s) {
1616	>	return ((s & TERMINATED) != 0 ? "Terminated" :
1617	>	((s & TERMINATING) != 0 ? "Terminating" :
1618	>	((s & SHUTDOWN) != 0 ? "Shutting down" :
1619	>	"Running")));
1620		}
1621
1055	–	// lifecycle control
1056	–
1622		/**
1623		* Initiates an orderly shutdown in which previously submitted
1624		* tasks are executed, but no new tasks will be accepted.
#	Line 1068 | Line 1633 | public class ForkJoinPool extends Abstra
1633		*/
1634		public void shutdown() {
1635		checkPermission();
1636	<	transitionRunStateTo(SHUTDOWN);
1637	<	if (canTerminateOnShutdown(runControl)) {
1073	<	if (workers == null) { // shutting down before workers created
1074	<	final ReentrantLock lock = this.workerLock;
1075	<	lock.lock();
1076	<	try {
1077	<	if (workers == null) {
1078	<	terminate();
1079	<	transitionRunStateTo(TERMINATED);
1080	<	termination.signalAll();
1081	<	}
1082	<	} finally {
1083	<	lock.unlock();
1084	<	}
1085	<	}
1086	<	terminateOnShutdown();
1087	<	}
1636	>	advanceRunLevel(SHUTDOWN);
1637	>	tryTerminate(false);
1638		}
1639
1640		/**
1641	<	* Attempts to stop all actively executing tasks, and cancels all
1642	<	* waiting tasks.  Tasks that are in the process of being
1643	<	* submitted or executed concurrently during the course of this
1644	<	* method may or may not be rejected. Unlike some other executors,
1645	<	* this method cancels rather than collects non-executed tasks
1646	<	* upon termination, so always returns an empty list. However, you
1647	<	* can use method {@link #drainTasksTo} before invoking this
1648	<	* method to transfer unexecuted tasks to another collection.
1641	>	* Attempts to cancel and/or stop all tasks, and reject all
1642	>	* subsequently submitted tasks.  Tasks that are in the process of
1643	>	* being submitted or executed concurrently during the course of
1644	>	* this method may or may not be rejected. This method cancels
1645	>	* both existing and unexecuted tasks, in order to permit
1646	>	* termination in the presence of task dependencies. So the method
1647	>	* always returns an empty list (unlike the case for some other
1648	>	* Executors).
1649		*
1650		* @return an empty list
1651		* @throws SecurityException if a security manager exists and
#	Line 1105 | Line 1655 | public class ForkJoinPool extends Abstra
1655		*/
1656		public List<Runnable> shutdownNow() {
1657		checkPermission();
1658	<	terminate();
1658	>	tryTerminate(true);
1659		return Collections.emptyList();
1660		}
1661
#	Line 1115 | Line 1665 | public class ForkJoinPool extends Abstra
1665		* @return {@code true} if all tasks have completed following shut down
1666		*/
1667		public boolean isTerminated() {
1668	<	return runStateOf(runControl) == TERMINATED;
1668	>	return runState >= TERMINATED;
1669		}
1670
1671		/**
1672		* Returns {@code true} if the process of termination has
1673	<	* commenced but possibly not yet completed.
1673	>	* commenced but not yet completed.  This method may be useful for
1674	>	* debugging. A return of {@code true} reported a sufficient
1675	>	* period after shutdown may indicate that submitted tasks have
1676	>	* ignored or suppressed interruption, causing this executor not
1677	>	* to properly terminate.
1678		*
1679	<	* @return {@code true} if terminating
1679	>	* @return {@code true} if terminating but not yet terminated
1680		*/
1681		public boolean isTerminating() {
1682	<	return runStateOf(runControl) >= TERMINATING;
1682	>	return (runState & (TERMINATING|TERMINATED)) == TERMINATING;
1683		}
1684
1685		/**
#	Line 1134 | Line 1688 | public class ForkJoinPool extends Abstra
1688		* @return {@code true} if this pool has been shut down
1689		*/
1690		public boolean isShutdown() {
1691	<	return runStateOf(runControl) >= SHUTDOWN;
1691	>	return runState >= SHUTDOWN;
1692		}
1693
1694		/**
#	Line 1150 | Line 1704 | public class ForkJoinPool extends Abstra
1704		*/
1705		public boolean awaitTermination(long timeout, TimeUnit unit)
1706		throws InterruptedException {
1153	–	long nanos = unit.toNanos(timeout);
1154	–	final ReentrantLock lock = this.workerLock;
1155	–	lock.lock();
1707		try {
1708	<	for (;;) {
1709	<	if (isTerminated())
1159	<	return true;
1160	<	if (nanos <= 0)
1161	<	return false;
1162	<	nanos = termination.awaitNanos(nanos);
1163	<	}
1164	<	} finally {
1165	<	lock.unlock();
1166	<	}
1167	<	}
1168	<
1169	<	// Shutdown and termination support
1170	<
1171	<	/**
1172	<	* Callback from terminating worker. Nulls out the corresponding
1173	<	* workers slot, and if terminating, tries to terminate; else
1174	<	* tries to shrink workers array.
1175	<	*
1176	<	* @param w the worker
1177	<	*/
1178	<	final void workerTerminated(ForkJoinWorkerThread w) {
1179	<	updateStealCount(w);
1180	<	updateWorkerCount(-1);
1181	<	final ReentrantLock lock = this.workerLock;
1182	<	lock.lock();
1183	<	try {
1184	<	ForkJoinWorkerThread[] ws = workers;
1185	<	if (ws != null) {
1186	<	int idx = w.poolIndex;
1187	<	if (idx >= 0 && idx < ws.length && ws[idx] == w)
1188	<	ws[idx] = null;
1189	<	if (totalCountOf(workerCounts) == 0) {
1190	<	terminate(); // no-op if already terminating
1191	<	transitionRunStateTo(TERMINATED);
1192	<	termination.signalAll();
1193	<	}
1194	<	else if (!isTerminating()) {
1195	<	tryShrinkWorkerArray();
1196	<	tryResumeSpare(true); // allow replacement
1197	<	}
1198	<	}
1199	<	} finally {
1200	<	lock.unlock();
1201	<	}
1202	<	signalIdleWorkers();
1203	<	}
1204	<
1205	<	/**
1206	<	* Initiates termination.
1207	<	*/
1208	<	private void terminate() {
1209	<	if (transitionRunStateTo(TERMINATING)) {
1210	<	stopAllWorkers();
1211	<	resumeAllSpares();
1212	<	signalIdleWorkers();
1213	<	cancelQueuedSubmissions();
1214	<	cancelQueuedWorkerTasks();
1215	<	interruptUnterminatedWorkers();
1216	<	signalIdleWorkers(); // resignal after interrupt
1217	<	}
1218	<	}
1219	<
1220	<	/**
1221	<	* Possibly terminates when on shutdown state.
1222	<	*/
1223	<	private void terminateOnShutdown() {
1224	<	if (!hasQueuedSubmissions() && canTerminateOnShutdown(runControl))
1225	<	terminate();
1226	<	}
1227	<
1228	<	/**
1229	<	* Clears out and cancels submissions.
1230	<	*/
1231	<	private void cancelQueuedSubmissions() {
1232	<	ForkJoinTask<?> task;
1233	<	while ((task = pollSubmission()) != null)
1234	<	task.cancel(false);
1235	<	}
1236	<
1237	<	/**
1238	<	* Cleans out worker queues.
1239	<	*/
1240	<	private void cancelQueuedWorkerTasks() {
1241	<	final ReentrantLock lock = this.workerLock;
1242	<	lock.lock();
1243	<	try {
1244	<	ForkJoinWorkerThread[] ws = workers;
1245	<	if (ws != null) {
1246	<	for (int i = 0; i < ws.length; ++i) {
1247	<	ForkJoinWorkerThread t = ws[i];
1248	<	if (t != null)
1249	<	t.cancelTasks();
1250	<	}
1251	<	}
1252	<	} finally {
1253	<	lock.unlock();
1254	<	}
1255	<	}
1256	<
1257	<	/**
1258	<	* Sets each worker's status to terminating. Requires lock to avoid
1259	<	* conflicts with add/remove.
1260	<	*/
1261	<	private void stopAllWorkers() {
1262	<	final ReentrantLock lock = this.workerLock;
1263	<	lock.lock();
1264	<	try {
1265	<	ForkJoinWorkerThread[] ws = workers;
1266	<	if (ws != null) {
1267	<	for (int i = 0; i < ws.length; ++i) {
1268	<	ForkJoinWorkerThread t = ws[i];
1269	<	if (t != null)
1270	<	t.shutdownNow();
1271	<	}
1272	<	}
1273	<	} finally {
1274	<	lock.unlock();
1275	<	}
1276	<	}
1277	<
1278	<	/**
1279	<	* Interrupts all unterminated workers.  This is not required for
1280	<	* sake of internal control, but may help unstick user code during
1281	<	* shutdown.
1282	<	*/
1283	<	private void interruptUnterminatedWorkers() {
1284	<	final ReentrantLock lock = this.workerLock;
1285	<	lock.lock();
1286	<	try {
1287	<	ForkJoinWorkerThread[] ws = workers;
1288	<	if (ws != null) {
1289	<	for (int i = 0; i < ws.length; ++i) {
1290	<	ForkJoinWorkerThread t = ws[i];
1291	<	if (t != null && !t.isTerminated()) {
1292	<	try {
1293	<	t.interrupt();
1294	<	} catch (SecurityException ignore) {
1295	<	}
1296	<	}
1297	<	}
1298	<	}
1299	<	} finally {
1300	<	lock.unlock();
1301	<	}
1302	<	}
1303	<
1304	<
1305	<	/*
1306	<	* Nodes for event barrier to manage idle threads.  Queue nodes
1307	<	* are basic Treiber stack nodes, also used for spare stack.
1308	<	*
1309	<	* The event barrier has an event count and a wait queue (actually
1310	<	* a Treiber stack).  Workers are enabled to look for work when
1311	<	* the eventCount is incremented. If they fail to find work, they
1312	<	* may wait for next count. Upon release, threads help others wake
1313	<	* up.
1314	<	*
1315	<	* Synchronization events occur only in enough contexts to
1316	<	* maintain overall liveness:
1317	<	*
1318	<	*   - Submission of a new task to the pool
1319	<	*   - Resizes or other changes to the workers array
1320	<	*   - pool termination
1321	<	*   - A worker pushing a task on an empty queue
1322	<	*
1323	<	* The case of pushing a task occurs often enough, and is heavy
1324	<	* enough compared to simple stack pushes, to require special
1325	<	* handling: Method signalWork returns without advancing count if
1326	<	* the queue appears to be empty.  This would ordinarily result in
1327	<	* races causing some queued waiters not to be woken up. To avoid
1328	<	* this, the first worker enqueued in method sync (see
1329	<	* syncIsReleasable) rescans for tasks after being enqueued, and
1330	<	* helps signal if any are found. This works well because the
1331	<	* worker has nothing better to do, and so might as well help
1332	<	* alleviate the overhead and contention on the threads actually
1333	<	* doing work.  Also, since event counts increments on task
1334	<	* availability exist to maintain liveness (rather than to force
1335	<	* refreshes etc), it is OK for callers to exit early if
1336	<	* contending with another signaller.
1337	<	*/
1338	<	static final class WaitQueueNode {
1339	<	WaitQueueNode next; // only written before enqueued
1340	<	volatile ForkJoinWorkerThread thread; // nulled to cancel wait
1341	<	final long count; // unused for spare stack
1342	<
1343	<	WaitQueueNode(long c, ForkJoinWorkerThread w) {
1344	<	count = c;
1345	<	thread = w;
1346	<	}
1347	<
1348	<	/**
1349	<	* Wakes up waiter, returning false if known to already
1350	<	*/
1351	<	boolean signal() {
1352	<	ForkJoinWorkerThread t = thread;
1353	<	if (t == null)
1354	<	return false;
1355	<	thread = null;
1356	<	LockSupport.unpark(t);
1357	<	return true;
1358	<	}
1359	<
1360	<	/**
1361	<	* Awaits release on sync.
1362	<	*/
1363	<	void awaitSyncRelease(ForkJoinPool p) {
1364	<	while (thread != null && !p.syncIsReleasable(this))
1365	<	LockSupport.park(this);
1366	<	}
1367	<
1368	<	/**
1369	<	* Awaits resumption as spare.
1370	<	*/
1371	<	void awaitSpareRelease() {
1372	<	while (thread != null) {
1373	<	if (!Thread.interrupted())
1374	<	LockSupport.park(this);
1375	<	}
1376	<	}
1377	<	}
1378	<
1379	<	/**
1380	<	* Ensures that no thread is waiting for count to advance from the
1381	<	* current value of eventCount read on entry to this method, by
1382	<	* releasing waiting threads if necessary.
1383	<	*
1384	<	* @return the count
1385	<	*/
1386	<	final long ensureSync() {
1387	<	long c = eventCount;
1388	<	WaitQueueNode q;
1389	<	while ((q = syncStack) != null && q.count < c) {
1390	<	if (casBarrierStack(q, null)) {
1391	<	do {
1392	<	q.signal();
1393	<	} while ((q = q.next) != null);
1394	<	break;
1395	<	}
1396	<	}
1397	<	return c;
1398	<	}
1399	<
1400	<	/**
1401	<	* Increments event count and releases waiting threads.
1402	<	*/
1403	<	private void signalIdleWorkers() {
1404	<	long c;
1405	<	do {} while (!casEventCount(c = eventCount, c+1));
1406	<	ensureSync();
1407	<	}
1408	<
1409	<	/**
1410	<	* Signals threads waiting to poll a task. Because method sync
1411	<	* rechecks availability, it is OK to only proceed if queue
1412	<	* appears to be non-empty, and OK to skip under contention to
1413	<	* increment count (since some other thread succeeded).
1414	<	*/
1415	<	final void signalWork() {
1416	<	long c;
1417	<	WaitQueueNode q;
1418	<	if (syncStack != null &&
1419	<	casEventCount(c = eventCount, c+1) &&
1420	<	(((q = syncStack) != null && q.count <= c) &&
1421	<	(!casBarrierStack(q, q.next) || !q.signal())))
1422	<	ensureSync();
1423	<	}
1424	<
1425	<	/**
1426	<	* Waits until event count advances from last value held by
1427	<	* caller, or if excess threads, caller is resumed as spare, or
1428	<	* caller or pool is terminating. Updates caller's event on exit.
1429	<	*
1430	<	* @param w the calling worker thread
1431	<	*/
1432	<	final void sync(ForkJoinWorkerThread w) {
1433	<	updateStealCount(w); // Transfer w's count while it is idle
1434	<
1435	<	while (!w.isShutdown() && !isTerminating() && !suspendIfSpare(w)) {
1436	<	long prev = w.lastEventCount;
1437	<	WaitQueueNode node = null;
1438	<	WaitQueueNode h;
1439	<	while (eventCount == prev &&
1440	<	((h = syncStack) == null || h.count == prev)) {
1441	<	if (node == null)
1442	<	node = new WaitQueueNode(prev, w);
1443	<	if (casBarrierStack(node.next = h, node)) {
1444	<	node.awaitSyncRelease(this);
1445	<	break;
1446	<	}
1447	<	}
1448	<	long ec = ensureSync();
1449	<	if (ec != prev) {
1450	<	w.lastEventCount = ec;
1451	<	break;
1452	<	}
1453	<	}
1454	<	}
1455	<
1456	<	/**
1457	<	* Returns {@code true} if worker waiting on sync can proceed:
1458	<	*  - on signal (thread == null)
1459	<	*  - on event count advance (winning race to notify vs signaller)
1460	<	*  - on interrupt
1461	<	*  - if the first queued node, we find work available
1462	<	* If node was not signalled and event count not advanced on exit,
1463	<	* then we also help advance event count.
1464	<	*
1465	<	* @return {@code true} if node can be released
1466	<	*/
1467	<	final boolean syncIsReleasable(WaitQueueNode node) {
1468	<	long prev = node.count;
1469	<	if (!Thread.interrupted() && node.thread != null &&
1470	<	(node.next != null ||
1471	<	!ForkJoinWorkerThread.hasQueuedTasks(workers)) &&
1472	<	eventCount == prev)
1473	<	return false;
1474	<	if (node.thread != null) {
1475	<	node.thread = null;
1476	<	long ec = eventCount;
1477	<	if (prev <= ec) // help signal
1478	<	casEventCount(ec, ec+1);
1479	<	}
1480	<	return true;
1481	<	}
1482	<
1483	<	/**
1484	<	* Returns {@code true} if a new sync event occurred since last
1485	<	* call to sync or this method, if so, updating caller's count.
1486	<	*/
1487	<	final boolean hasNewSyncEvent(ForkJoinWorkerThread w) {
1488	<	long lc = w.lastEventCount;
1489	<	long ec = ensureSync();
1490	<	if (ec == lc)
1708	>	return termination.awaitAdvanceInterruptibly(0, timeout, unit) > 0;
1709	>	} catch(TimeoutException ex) {
1710		return false;
1492	–	w.lastEventCount = ec;
1493	–	return true;
1494	–	}
1495	–
1496	–	//  Parallelism maintenance
1497	–
1498	–	/**
1499	–	* Decrements running count; if too low, adds spare.
1500	–	*
1501	–	* Conceptually, all we need to do here is add or resume a
1502	–	* spare thread when one is about to block (and remove or
1503	–	* suspend it later when unblocked -- see suspendIfSpare).
1504	–	* However, implementing this idea requires coping with
1505	–	* several problems: we have imperfect information about the
1506	–	* states of threads. Some count updates can and usually do
1507	–	* lag run state changes, despite arrangements to keep them
1508	–	* accurate (for example, when possible, updating counts
1509	–	* before signalling or resuming), especially when running on
1510	–	* dynamic JVMs that don't optimize the infrequent paths that
1511	–	* update counts. Generating too many threads can make these
1512	–	* problems become worse, because excess threads are more
1513	–	* likely to be context-switched with others, slowing them all
1514	–	* down, especially if there is no work available, so all are
1515	–	* busy scanning or idling.  Also, excess spare threads can
1516	–	* only be suspended or removed when they are idle, not
1517	–	* immediately when they aren't needed. So adding threads will
1518	–	* raise parallelism level for longer than necessary.  Also,
1519	–	* FJ applications often encounter highly transient peaks when
1520	–	* many threads are blocked joining, but for less time than it
1521	–	* takes to create or resume spares.
1522	–	*
1523	–	* @param joinMe if non-null, return early if done
1524	–	* @param maintainParallelism if true, try to stay within
1525	–	* target counts, else create only to avoid starvation
1526	–	* @return true if joinMe known to be done
1527	–	*/
1528	–	final boolean preJoin(ForkJoinTask<?> joinMe,
1529	–	boolean maintainParallelism) {
1530	–	maintainParallelism &= maintainsParallelism; // overrride
1531	–	boolean dec = false;  // true when running count decremented
1532	–	while (spareStack == null || !tryResumeSpare(dec)) {
1533	–	int counts = workerCounts;
1534	–	if (dec || (dec = casWorkerCounts(counts, --counts))) {
1535	–	// CAS cheat
1536	–	if (!needSpare(counts, maintainParallelism))
1537	–	break;
1538	–	if (joinMe.status < 0)
1539	–	return true;
1540	–	if (tryAddSpare(counts))
1541	–	break;
1542	–	}
1543	–	}
1544	–	return false;
1545	–	}
1546	–
1547	–	/**
1548	–	* Same idea as preJoin
1549	–	*/
1550	–	final boolean preBlock(ManagedBlocker blocker,
1551	–	boolean maintainParallelism) {
1552	–	maintainParallelism &= maintainsParallelism;
1553	–	boolean dec = false;
1554	–	while (spareStack == null || !tryResumeSpare(dec)) {
1555	–	int counts = workerCounts;
1556	–	if (dec || (dec = casWorkerCounts(counts, --counts))) {
1557	–	if (!needSpare(counts, maintainParallelism))
1558	–	break;
1559	–	if (blocker.isReleasable())
1560	–	return true;
1561	–	if (tryAddSpare(counts))
1562	–	break;
1563	–	}
1564	–	}
1565	–	return false;
1566	–	}
1567	–
1568	–	/**
1569	–	* Returns {@code true} if a spare thread appears to be needed.
1570	–	* If maintaining parallelism, returns true when the deficit in
1571	–	* running threads is more than the surplus of total threads, and
1572	–	* there is apparently some work to do.  This self-limiting rule
1573	–	* means that the more threads that have already been added, the
1574	–	* less parallelism we will tolerate before adding another.
1575	–	*
1576	–	* @param counts current worker counts
1577	–	* @param maintainParallelism try to maintain parallelism
1578	–	*/
1579	–	private boolean needSpare(int counts, boolean maintainParallelism) {
1580	–	int ps = parallelism;
1581	–	int rc = runningCountOf(counts);
1582	–	int tc = totalCountOf(counts);
1583	–	int runningDeficit = ps - rc;
1584	–	int totalSurplus = tc - ps;
1585	–	return (tc < maxPoolSize &&
1586	–	(rc == 0 || totalSurplus < 0 ||
1587	–	(maintainParallelism &&
1588	–	runningDeficit > totalSurplus &&
1589	–	ForkJoinWorkerThread.hasQueuedTasks(workers))));
1590	–	}
1591	–
1592	–	/**
1593	–	* Adds a spare worker if lock available and no more than the
1594	–	* expected numbers of threads exist.
1595	–	*
1596	–	* @return true if successful
1597	–	*/
1598	–	private boolean tryAddSpare(int expectedCounts) {
1599	–	final ReentrantLock lock = this.workerLock;
1600	–	int expectedRunning = runningCountOf(expectedCounts);
1601	–	int expectedTotal = totalCountOf(expectedCounts);
1602	–	boolean success = false;
1603	–	boolean locked = false;
1604	–	// confirm counts while locking; CAS after obtaining lock
1605	–	try {
1606	–	for (;;) {
1607	–	int s = workerCounts;
1608	–	int tc = totalCountOf(s);
1609	–	int rc = runningCountOf(s);
1610	–	if (rc > expectedRunning || tc > expectedTotal)
1611	–	break;
1612	–	if (!locked && !(locked = lock.tryLock()))
1613	–	break;
1614	–	if (casWorkerCounts(s, workerCountsFor(tc+1, rc+1))) {
1615	–	createAndStartSpare(tc);
1616	–	success = true;
1617	–	break;
1618	–	}
1619	–	}
1620	–	} finally {
1621	–	if (locked)
1622	–	lock.unlock();
1623	–	}
1624	–	return success;
1625	–	}
1626	–
1627	–	/**
1628	–	* Adds the kth spare worker. On entry, pool counts are already
1629	–	* adjusted to reflect addition.
1630	–	*/
1631	–	private void createAndStartSpare(int k) {
1632	–	ForkJoinWorkerThread w = null;
1633	–	ForkJoinWorkerThread[] ws = ensureWorkerArrayCapacity(k + 1);
1634	–	int len = ws.length;
1635	–	// Probably, we can place at slot k. If not, find empty slot
1636	–	if (k < len && ws[k] != null) {
1637	–	for (k = 0; k < len && ws[k] != null; ++k)
1638	–	;
1639	–	}
1640	–	if (k < len && !isTerminating() && (w = createWorker(k)) != null) {
1641	–	ws[k] = w;
1642	–	w.start();
1643	–	}
1644	–	else
1645	–	updateWorkerCount(-1); // adjust on failure
1646	–	signalIdleWorkers();
1647	–	}
1648	–
1649	–	/**
1650	–	* Suspends calling thread w if there are excess threads.  Called
1651	–	* only from sync.  Spares are enqueued in a Treiber stack using
1652	–	* the same WaitQueueNodes as barriers.  They are resumed mainly
1653	–	* in preJoin, but are also woken on pool events that require all
1654	–	* threads to check run state.
1655	–	*
1656	–	* @param w the caller
1657	–	*/
1658	–	private boolean suspendIfSpare(ForkJoinWorkerThread w) {
1659	–	WaitQueueNode node = null;
1660	–	int s;
1661	–	while (parallelism < runningCountOf(s = workerCounts)) {
1662	–	if (node == null)
1663	–	node = new WaitQueueNode(0, w);
1664	–	if (casWorkerCounts(s, s-1)) { // representation-dependent
1665	–	// push onto stack
1666	–	do {} while (!casSpareStack(node.next = spareStack, node));
1667	–	// block until released by resumeSpare
1668	–	node.awaitSpareRelease();
1669	–	return true;
1670	–	}
1671	–	}
1672	–	return false;
1673	–	}
1674	–
1675	–	/**
1676	–	* Tries to pop and resume a spare thread.
1677	–	*
1678	–	* @param updateCount if true, increment running count on success
1679	–	* @return true if successful
1680	–	*/
1681	–	private boolean tryResumeSpare(boolean updateCount) {
1682	–	WaitQueueNode q;
1683	–	while ((q = spareStack) != null) {
1684	–	if (casSpareStack(q, q.next)) {
1685	–	if (updateCount)
1686	–	updateRunningCount(1);
1687	–	q.signal();
1688	–	return true;
1689	–	}
1690	–	}
1691	–	return false;
1692	–	}
1693	–
1694	–	/**
1695	–	* Pops and resumes all spare threads. Same idea as ensureSync.
1696	–	*
1697	–	* @return true if any spares released
1698	–	*/
1699	–	private boolean resumeAllSpares() {
1700	–	WaitQueueNode q;
1701	–	while ( (q = spareStack) != null) {
1702	–	if (casSpareStack(q, null)) {
1703	–	do {
1704	–	updateRunningCount(1);
1705	–	q.signal();
1706	–	} while ((q = q.next) != null);
1707	–	return true;
1708	–	}
1709	–	}
1710	–	return false;
1711	–	}
1712	–
1713	–	/**
1714	–	* Pops and shuts down excessive spare threads. Call only while
1715	–	* holding lock. This is not guaranteed to eliminate all excess
1716	–	* threads, only those suspended as spares, which are the ones
1717	–	* unlikely to be needed in the future.
1718	–	*/
1719	–	private void trimSpares() {
1720	–	int surplus = totalCountOf(workerCounts) - parallelism;
1721	–	WaitQueueNode q;
1722	–	while (surplus > 0 && (q = spareStack) != null) {
1723	–	if (casSpareStack(q, null)) {
1724	–	do {
1725	–	updateRunningCount(1);
1726	–	ForkJoinWorkerThread w = q.thread;
1727	–	if (w != null && surplus > 0 &&
1728	–	runningCountOf(workerCounts) > 0 && w.shutdown())
1729	–	--surplus;
1730	–	q.signal();
1731	–	} while ((q = q.next) != null);
1732	–	}
1711		}
1712		}
1713
#	Line 1737 | Line 1715 | public class ForkJoinPool extends Abstra
1715		* Interface for extending managed parallelism for tasks running
1716		* in {@link ForkJoinPool}s.
1717		*
1718	<	* <p>A {@code ManagedBlocker} provides two methods.
1719	<	* Method {@code isReleasable} must return {@code true} if
1720	<	* blocking is not necessary. Method {@code block} blocks the
1721	<	* current thread if necessary (perhaps internally invoking
1722	<	* {@code isReleasable} before actually blocking.).
1718	>	* <p>A {@code ManagedBlocker} provides two methods.  Method
1719	>	* {@code isReleasable} must return {@code true} if blocking is
1720	>	* not necessary. Method {@code block} blocks the current thread
1721	>	* if necessary (perhaps internally invoking {@code isReleasable}
1722	>	* before actually blocking). The unusual methods in this API
1723	>	* accommodate synchronizers that may, but don't usually, block
1724	>	* for long periods. Similarly, they allow more efficient internal
1725	>	* handling of cases in which additional workers may be, but
1726	>	* usually are not, needed to ensure sufficient parallelism.
1727	>	* Toward this end, implementations of method {@code isReleasable}
1728	>	* must be amenable to repeated invocation.
1729		*
1730		* <p>For example, here is a ManagedBlocker based on a
1731		* ReentrantLock:
#	Line 1759 | Line 1743 | public class ForkJoinPool extends Abstra
1743		*     return hasLock || (hasLock = lock.tryLock());
1744		*   }
1745		* }}</pre>
1746	+	*
1747	+	* <p>Here is a class that possibly blocks waiting for an
1748	+	* item on a given queue:
1749	+	*  <pre> {@code
1750	+	* class QueueTaker<E> implements ManagedBlocker {
1751	+	*   final BlockingQueue<E> queue;
1752	+	*   volatile E item = null;
1753	+	*   QueueTaker(BlockingQueue<E> q) { this.queue = q; }
1754	+	*   public boolean block() throws InterruptedException {
1755	+	*     if (item == null)
1756	+	*       item = queue.take
1757	+	*     return true;
1758	+	*   }
1759	+	*   public boolean isReleasable() {
1760	+	*     return item != null || (item = queue.poll) != null;
1761	+	*   }
1762	+	*   public E getItem() { // call after pool.managedBlock completes
1763	+	*     return item;
1764	+	*   }
1765	+	* }}</pre>
1766		*/
1767		public static interface ManagedBlocker {
1768		/**
#	Line 1780 | Line 1784 | public class ForkJoinPool extends Abstra
1784
1785		/**
1786		* Blocks in accord with the given blocker.  If the current thread
1787	<	* is a ForkJoinWorkerThread, this method possibly arranges for a
1788	<	* spare thread to be activated if necessary to ensure parallelism
1789	<	* while the current thread is blocked.  If
1786	<	* {@code maintainParallelism} is {@code true} and the pool supports
1787	<	* it ({@link #getMaintainsParallelism}), this method attempts to
1788	<	* maintain the pool's nominal parallelism. Otherwise it activates
1789	<	* a thread only if necessary to avoid complete starvation. This
1790	<	* option may be preferable when blockages use timeouts, or are
1791	<	* almost always brief.
1787	>	* is a {@link ForkJoinWorkerThread}, this method possibly
1788	>	* arranges for a spare thread to be activated if necessary to
1789	>	* ensure sufficient parallelism while the current thread is blocked.
1790		*
1791	<	* <p> If the caller is not a ForkJoinTask, this method is behaviorally
1792	<	* equivalent to
1791	>	* <p>If the caller is not a {@link ForkJoinTask}, this method is
1792	>	* behaviorally equivalent to
1793		*  <pre> {@code
1794		* while (!blocker.isReleasable())
1795		*   if (blocker.block())
1796		*     return;
1797		* }</pre>
1798	<	* If the caller is a ForkJoinTask, then the pool may first
1799	<	* be expanded to ensure parallelism, and later adjusted.
1798	>	*
1799	>	* If the caller is a {@code ForkJoinTask}, then the pool may
1800	>	* first be expanded to ensure parallelism, and later adjusted.
1801		*
1802		* @param blocker the blocker
1804	–	* @param maintainParallelism if {@code true} and supported by
1805	–	* this pool, attempt to maintain the pool's nominal parallelism;
1806	–	* otherwise activate a thread only if necessary to avoid
1807	–	* complete starvation.
1803		* @throws InterruptedException if blocker.block did so
1804		*/
1805	<	public static void managedBlock(ManagedBlocker blocker,
1811	<	boolean maintainParallelism)
1805	>	public static void managedBlock(ManagedBlocker blocker)
1806		throws InterruptedException {
1807		Thread t = Thread.currentThread();
1808	<	ForkJoinPool pool = ((t instanceof ForkJoinWorkerThread) ?
1809	<	((ForkJoinWorkerThread) t).pool : null);
1810	<	if (!blocker.isReleasable()) {
1811	<	try {
1812	<	if (pool == null ||
1813	<	!pool.preBlock(blocker, maintainParallelism))
1820	<	awaitBlocker(blocker);
1821	<	} finally {
1822	<	if (pool != null)
1823	<	pool.updateRunningCount(1);
1824	<	}
1808	>	if (t instanceof ForkJoinWorkerThread) {
1809	>	ForkJoinWorkerThread w = (ForkJoinWorkerThread) t;
1810	>	w.pool.awaitBlocker(blocker);
1811	>	}
1812	>	else {
1813	>	do {} while (!blocker.isReleasable() && !blocker.block());
1814		}
1826	–	}
1827	–
1828	–	private static void awaitBlocker(ManagedBlocker blocker)
1829	–	throws InterruptedException {
1830	–	do {} while (!blocker.isReleasable() && !blocker.block());
1815		}
1816
1817		// AbstractExecutorService overrides.  These rely on undocumented
#	Line 1845 | Line 1829 | public class ForkJoinPool extends Abstra
1829		// Unsafe mechanics
1830
1831		private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1848	–	private static final long eventCountOffset =
1849	–	objectFieldOffset("eventCount", ForkJoinPool.class);
1832		private static final long workerCountsOffset =
1833		objectFieldOffset("workerCounts", ForkJoinPool.class);
1834	<	private static final long runControlOffset =
1835	<	objectFieldOffset("runControl", ForkJoinPool.class);
1836	<	private static final long syncStackOffset =
1837	<	objectFieldOffset("syncStack",ForkJoinPool.class);
1838	<	private static final long spareStackOffset =
1839	<	objectFieldOffset("spareStack", ForkJoinPool.class);
1840	<
1841	<	private boolean casEventCount(long cmp, long val) {
1842	<	return UNSAFE.compareAndSwapLong(this, eventCountOffset, cmp, val);
1843	<	}
1862	<	private boolean casWorkerCounts(int cmp, int val) {
1863	<	return UNSAFE.compareAndSwapInt(this, workerCountsOffset, cmp, val);
1864	<	}
1865	<	private boolean casRunControl(int cmp, int val) {
1866	<	return UNSAFE.compareAndSwapInt(this, runControlOffset, cmp, val);
1867	<	}
1868	<	private boolean casSpareStack(WaitQueueNode cmp, WaitQueueNode val) {
1869	<	return UNSAFE.compareAndSwapObject(this, spareStackOffset, cmp, val);
1870	<	}
1871	<	private boolean casBarrierStack(WaitQueueNode cmp, WaitQueueNode val) {
1872	<	return UNSAFE.compareAndSwapObject(this, syncStackOffset, cmp, val);
1873	<	}
1834	>	private static final long runStateOffset =
1835	>	objectFieldOffset("runState", ForkJoinPool.class);
1836	>	private static final long eventCountOffset =
1837	>	objectFieldOffset("eventCount", ForkJoinPool.class);
1838	>	private static final long eventWaitersOffset =
1839	>	objectFieldOffset("eventWaiters",ForkJoinPool.class);
1840	>	private static final long stealCountOffset =
1841	>	objectFieldOffset("stealCount",ForkJoinPool.class);
1842	>	private static final long spareWaitersOffset =
1843	>	objectFieldOffset("spareWaiters",ForkJoinPool.class);
1844
1845		private static long objectFieldOffset(String field, Class<?> klazz) {
1846		try {

Diff Legend

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