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 1.74 by jsr166, Tue Sep 7 06:32:45 2010 UTC vs.
Revision 1.94 by dl, Tue Mar 1 10:59:04 2011 UTC

#	Line 6 | Line 6
6
7		package jsr166y;
8
9	–	import java.util.concurrent.*;
9		import java.util.ArrayList;
10		import java.util.Arrays;
11		import java.util.Collection;
12		import java.util.Collections;
13		import java.util.List;
14	+	import java.util.Random;
15	+	import java.util.concurrent.AbstractExecutorService;
16	+	import java.util.concurrent.Callable;
17	+	import java.util.concurrent.ExecutorService;
18	+	import java.util.concurrent.Future;
19	+	import java.util.concurrent.RejectedExecutionException;
20	+	import java.util.concurrent.RunnableFuture;
21	+	import java.util.concurrent.TimeUnit;
22	+	import java.util.concurrent.TimeoutException;
23	+	import java.util.concurrent.atomic.AtomicInteger;
24		import java.util.concurrent.locks.LockSupport;
25		import java.util.concurrent.locks.ReentrantLock;
26	<	import java.util.concurrent.atomic.AtomicInteger;
18	<	import java.util.concurrent.CountDownLatch;
26	>	import java.util.concurrent.locks.Condition;
27
28		/**
29		* An {@link ExecutorService} for running {@link ForkJoinTask}s.
#	Line 123 | Line 131 | public class ForkJoinPool extends Abstra
131		* set of worker threads: Submissions from non-FJ threads enter
132		* into a submission queue. Workers take these tasks and typically
133		* split them into subtasks that may be stolen by other workers.
134	<	* The main work-stealing mechanics implemented in class
135	<	* ForkJoinWorkerThread give first priority to processing tasks
136	<	* from their own queues (LIFO or FIFO, depending on mode), then
137	<	* to randomized FIFO steals of tasks in other worker queues, and
138	<	* lastly to new submissions. These mechanics do not consider
139	<	* affinities, loads, cache localities, etc, so rarely provide the
140	<	* best possible performance on a given machine, but portably
141	<	* provide good throughput by averaging over these factors.
142	<	* (Further, even if we did try to use such information, we do not
143	<	* usually have a basis for exploiting it. For example, some sets
144	<	* of tasks profit from cache affinities, but others are harmed by
145	<	* cache pollution effects.)
146	<	*
147	<	* Beyond work-stealing support and essential bookkeeping, the
148	<	* main responsibility of this framework is to take actions when
149	<	* one worker is waiting to join a task stolen (or always held by)
150	<	* another.  Because we are multiplexing many tasks on to a pool
151	<	* of workers, we can't just let them block (as in Thread.join).
152	<	* We also cannot just reassign the joiner's run-time stack with
153	<	* another and replace it later, which would be a form of
154	<	* "continuation", that even if possible is not necessarily a good
155	<	* idea. Given that the creation costs of most threads on most
156	<	* systems mainly surrounds setting up runtime stacks, thread
157	<	* creation and switching is usually not much more expensive than
158	<	* stack creation and switching, and is more flexible). Instead we
134	>	* Preference rules give first priority to processing tasks from
135	>	* their own queues (LIFO or FIFO, depending on mode), then to
136	>	* randomized FIFO steals of tasks in other worker queues, and
137	>	* lastly to new submissions.
138	>	*
139	>	* The main throughput advantages of work-stealing stem from
140	>	* decentralized control -- workers mostly take tasks from
141	>	* themselves or each other. We cannot negate this in the
142	>	* implementation of other management responsibilities. The main
143	>	* tactic for avoiding bottlenecks is packing nearly all
144	>	* essentially atomic control state into a single 64bit volatile
145	>	* variable ("ctl"). This variable is read on the order of 10-100
146	>	* times as often as it is modified (always via CAS). (There is
147	>	* some additional control state, for example variable "shutdown"
148	>	* for which we can cope with uncoordinated updates.)  This
149	>	* streamlines synchronization and control at the expense of messy
150	>	* constructions needed to repack status bits upon updates.
151	>	* Updates tend not to contend with each other except during
152	>	* bursts while submitted tasks begin or end.  In some cases when
153	>	* they do contend, threads can instead do something else
154	>	* (usually, scan for tesks) until contention subsides.
155	>	*
156	>	* To enable packing, we restrict maximum parallelism to (1<<15)-1
157	>	* (which is far in excess of normal operating range) to allow
158	>	* ids, counts, and their negations (used for thresholding) to fit
159	>	* into 16bit fields.
160	>	*
161	>	* Recording Workers.  Workers are recorded in the "workers" array
162	>	* that is created upon pool construction and expanded if (rarely)
163	>	* necessary.  This is an array as opposed to some other data
164	>	* structure to support index-based random steals by workers.
165	>	* Updates to the array recording new workers and unrecording
166	>	* terminated ones are protected from each other by a seqLock
167	>	* (scanGuard) but the array is otherwise concurrently readable,
168	>	* and accessed directly by workers. To simplify index-based
169	>	* operations, the array size is always a power of two, and all
170	>	* readers must tolerate null slots. To avoid flailing during
171	>	* start-up, the array is presized to hold twice #parallelism
172	>	* workers (which is unlikely to need further resizing during
173	>	* execution). But to avoid dealing with so many null slots,
174	>	* variable scanGuard includes a mask for the nearest power of two
175	>	* that contains all current workers.  All worker thread creation
176	>	* is on-demand, triggered by task submissions, replacement of
177	>	* terminated workers, and/or compensation for blocked
178	>	* workers. However, all other support code is set up to work with
179	>	* other policies.  To ensure that we do not hold on to worker
180	>	* references that would prevent GC, ALL accesses to workers are
181	>	* via indices into the workers array (which is one source of some
182	>	* of the messy code constructions here). In essence, the workers
183	>	* array serves as a weak reference mechanism. Thus for example
184	>	* the wait queue field of ctl stores worker indices, not worker
185	>	* references.  Access to the workers in associated methods (for
186	>	* example signalWork) must both index-check and null-check the
187	>	* IDs. All such accesses ignore bad IDs by returning out early
188	>	* from what they are doing, since this can only be associated
189	>	* with termination, in which case it is OK to give up.
190	>	*
191	>	* All uses of the workers array, as well as queue arrays, check
192	>	* that the array is non-null (even if previously non-null). This
193	>	* allows nulling during termination, which is currently not
194	>	* necessary, but remains an option for resource-revocation-based
195	>	* shutdown schemes.
196	>	*
197	>	* Wait Queuing. Unlike HPC work-stealing frameworks, we cannot
198	>	* let workers spin indefinitely scanning for tasks when none are
199	>	* can be immediately found, and we cannot start/resume workers
200	>	* unless there appear to be tasks available.  On the other hand,
201	>	* we must quickly prod them into action when new tasks are
202	>	* submitted or generated.  We park/unpark workers after placing
203	>	* in an event wait queue when they cannot find work. This "queue"
204	>	* is actually a simple Treiber stack, headed by the "id" field of
205	>	* ctl, plus a 15bit counter value to both wake up waiters (by
206	>	* advancing their count) and avoid ABA effects. Successors are
207	>	* held in worker field "nextWait".  Queuing deals with several
208	>	* intrinsic races, mainly that a task-producing thread can miss
209	>	* seeing (and signalling) another thread that gave up looking for
210	>	* work but has not yet entered the wait queue. We solve this by
211	>	* requiring a full sweep of all workers both before (in scan())
212	>	* and after (in awaitWork()) a newly waiting worker is added to
213	>	* the wait queue. During a rescan, the worker might release some
214	>	* other queued worker rather than itself, which has the same net
215	>	* effect.
216	>	*
217	>	* Signalling.  We create or wake up workers only when there
218	>	* appears to be at least one task they might be able to find and
219	>	* execute.  When a submission is added or another worker adds a
220	>	* task to a queue that previously had two or fewer tasks, they
221	>	* signal waiting workers (or trigger creation of new ones if
222	>	* fewer than the given parallelism level -- see signalWork).
223	>	* These primary signals are buttressed by signals during rescans
224	>	* as well as those performed when a worker steals a task and
225	>	* notices that there are more tasks too; together these cover the
226	>	* signals needed in cases when more than two tasks are pushed
227	>	* but untaken.
228	>	*
229	>	* Trimming workers. To release resources after periods of lack of
230	>	* use, a worker starting to wait when the pool is quiescent will
231	>	* time out and terminate if the pool has remained quiescent for
232	>	* SHRINK_RATE nanosecs.
233	>	*
234	>	* Submissions. External submissions are maintained in an
235	>	* array-based queue that is structured identically to
236	>	* ForkJoinWorkerThread queues (which see) except for the use of
237	>	* submissionLock in method addSubmission. Unlike worker queues,
238	>	* multiple external threads can add new submissions.
239	>	*
240	>	* Compensation. Beyond work-stealing support and lifecycle
241	>	* control, the main responsibility of this framework is to take
242	>	* actions when one worker is waiting to join a task stolen (or
243	>	* always held by) another.  Because we are multiplexing many
244	>	* tasks on to a pool of workers, we can't just let them block (as
245	>	* in Thread.join).  We also cannot just reassign the joiner's
246	>	* run-time stack with another and replace it later, which would
247	>	* be a form of "continuation", that even if possible is not
248	>	* necessarily a good idea since we sometimes need both an
249	>	* unblocked task and its continuation to progress. Instead we
250		* combine two tactics:
251		*
252		*   Helping: Arranging for the joiner to execute some task that it
253		*      would be running if the steal had not occurred.  Method
254	<	*      ForkJoinWorkerThread.helpJoinTask tracks joining->stealing
254	>	*      ForkJoinWorkerThread.joinTask tracks joining->stealing
255		*      links to try to find such a task.
256		*
257		*   Compensating: Unless there are already enough live threads,
258	<	*      method helpMaintainParallelism() may create or
259	<	*      re-activate a spare thread to compensate for blocked
260	<	*      joiners until they unblock.
162	<	*
163	<	* It is impossible to keep exactly the target (parallelism)
164	<	* number of threads running at any given time.  Determining
165	<	* existence of conservatively safe helping targets, the
166	<	* availability of already-created spares, and the apparent need
167	<	* to create new spares are all racy and require heuristic
168	<	* guidance, so we rely on multiple retries of each.  Compensation
169	<	* occurs in slow-motion. It is triggered only upon timeouts of
170	<	* Object.wait used for joins. This reduces poor decisions that
171	<	* would otherwise be made when threads are waiting for others
172	<	* that are stalled because of unrelated activities such as
173	<	* garbage collection.
258	>	*      method tryPreBlock() may create or re-activate a spare
259	>	*      thread to compensate for blocked joiners until they
260	>	*      unblock.
261		*
262		* The ManagedBlocker extension API can't use helping so relies
263		* only on compensation in method awaitBlocker.
264		*
265	<	* The main throughput advantages of work-stealing stem from
266	<	* decentralized control -- workers mostly steal tasks from each
267	<	* other. We do not want to negate this by creating bottlenecks
268	<	* implementing other management responsibilities. So we use a
269	<	* collection of techniques that avoid, reduce, or cope well with
270	<	* contention. These entail several instances of bit-packing into
271	<	* CASable fields to maintain only the minimally required
272	<	* atomicity. To enable such packing, we restrict maximum
273	<	* parallelism to (1<<15)-1 (enabling twice this (to accommodate
274	<	* unbalanced increments and decrements) to fit into a 16 bit
275	<	* field, which is far in excess of normal operating range.  Even
276	<	* though updates to some of these bookkeeping fields do sometimes
277	<	* contend with each other, they don't normally cache-contend with
278	<	* updates to others enough to warrant memory padding or
279	<	* isolation. So they are all held as fields of ForkJoinPool
280	<	* objects.  The main capabilities are as follows:
281	<	*
282	<	* 1. Creating and removing workers. Workers are recorded in the
283	<	* "workers" array. This is an array as opposed to some other data
284	<	* structure to support index-based random steals by workers.
285	<	* Updates to the array recording new workers and unrecording
286	<	* terminated ones are protected from each other by a lock
287	<	* (workerLock) but the array is otherwise concurrently readable,
288	<	* and accessed directly by workers. To simplify index-based
289	<	* operations, the array size is always a power of two, and all
290	<	* readers must tolerate null slots. Currently, all worker thread
291	<	* creation is on-demand, triggered by task submissions,
292	<	* replacement of terminated workers, and/or compensation for
293	<	* blocked workers. However, all other support code is set up to
294	<	* work with other policies.
208	<	*
209	<	* To ensure that we do not hold on to worker references that
210	<	* would prevent GC, ALL accesses to workers are via indices into
211	<	* the workers array (which is one source of some of the unusual
212	<	* code constructions here). In essence, the workers array serves
213	<	* as a WeakReference mechanism. Thus for example the event queue
214	<	* stores worker indices, not worker references. Access to the
215	<	* workers in associated methods (for example releaseEventWaiters)
216	<	* must both index-check and null-check the IDs. All such accesses
217	<	* ignore bad IDs by returning out early from what they are doing,
218	<	* since this can only be associated with shutdown, in which case
219	<	* it is OK to give up. On termination, we just clobber these
220	<	* data structures without trying to use them.
221	<	*
222	<	* 2. Bookkeeping for dynamically adding and removing workers. We
223	<	* aim to approximately maintain the given level of parallelism.
224	<	* When some workers are known to be blocked (on joins or via
225	<	* ManagedBlocker), we may create or resume others to take their
226	<	* place until they unblock (see below). Implementing this
227	<	* requires counts of the number of "running" threads (i.e., those
228	<	* that are neither blocked nor artificially suspended) as well as
229	<	* the total number.  These two values are packed into one field,
230	<	* "workerCounts" because we need accurate snapshots when deciding
231	<	* to create, resume or suspend.  Note however that the
232	<	* correspondence of these counts to reality is not guaranteed. In
233	<	* particular updates for unblocked threads may lag until they
234	<	* actually wake up.
235	<	*
236	<	* 3. Maintaining global run state. The run state of the pool
237	<	* consists of a runLevel (SHUTDOWN, TERMINATING, etc) similar to
238	<	* those in other Executor implementations, as well as a count of
239	<	* "active" workers -- those that are, or soon will be, or
240	<	* recently were executing tasks. The runLevel and active count
241	<	* are packed together in order to correctly trigger shutdown and
242	<	* termination. Without care, active counts can be subject to very
243	<	* high contention.  We substantially reduce this contention by
244	<	* relaxing update rules.  A worker must claim active status
245	<	* prospectively, by activating if it sees that a submitted or
246	<	* stealable task exists (it may find after activating that the
247	<	* task no longer exists). It stays active while processing this
248	<	* task (if it exists) and any other local subtasks it produces,
249	<	* until it cannot find any other tasks. It then tries
250	<	* inactivating (see method preStep), but upon update contention
251	<	* instead scans for more tasks, later retrying inactivation if it
252	<	* doesn't find any.
253	<	*
254	<	* 4. Managing idle workers waiting for tasks. We cannot let
255	<	* workers spin indefinitely scanning for tasks when none are
256	<	* available. On the other hand, we must quickly prod them into
257	<	* action when new tasks are submitted or generated.  We
258	<	* park/unpark these idle workers using an event-count scheme.
259	<	* Field eventCount is incremented upon events that may enable
260	<	* workers that previously could not find a task to now find one:
261	<	* Submission of a new task to the pool, or another worker pushing
262	<	* a task onto a previously empty queue.  (We also use this
263	<	* mechanism for configuration and termination actions that
264	<	* require wakeups of idle workers).  Each worker maintains its
265	<	* last known event count, and blocks when a scan for work did not
266	<	* find a task AND its lastEventCount matches the current
267	<	* eventCount. Waiting idle workers are recorded in a variant of
268	<	* Treiber stack headed by field eventWaiters which, when nonzero,
269	<	* encodes the thread index and count awaited for by the worker
270	<	* thread most recently calling eventSync. This thread in turn has
271	<	* a record (field nextEventWaiter) for the next waiting worker.
272	<	* In addition to allowing simpler decisions about need for
273	<	* wakeup, the event count bits in eventWaiters serve the role of
274	<	* tags to avoid ABA errors in Treiber stacks. Upon any wakeup,
275	<	* released threads also try to release at most two others.  The
276	<	* net effect is a tree-like diffusion of signals, where released
277	<	* threads (and possibly others) help with unparks.  To further
278	<	* reduce contention effects a bit, failed CASes to increment
279	<	* field eventCount are tolerated without retries in signalWork.
280	<	* Conceptually they are merged into the same event, which is OK
281	<	* when their only purpose is to enable workers to scan for work.
282	<	*
283	<	* 5. Managing suspension of extra workers. When a worker notices
284	<	* (usually upon timeout of a wait()) that there are too few
285	<	* running threads, we may create a new thread to maintain
286	<	* parallelism level, or at least avoid starvation. Usually, extra
287	<	* threads are needed for only very short periods, yet join
288	<	* dependencies are such that we sometimes need them in
289	<	* bursts. Rather than create new threads each time this happens,
290	<	* we suspend no-longer-needed extra ones as "spares". For most
291	<	* purposes, we don't distinguish "extra" spare threads from
292	<	* normal "core" threads: On each call to preStep (the only point
293	<	* at which we can do this) a worker checks to see if there are
294	<	* now too many running workers, and if so, suspends itself.
295	<	* Method helpMaintainParallelism looks for suspended threads to
296	<	* resume before considering creating a new replacement. The
297	<	* spares themselves are encoded on another variant of a Treiber
298	<	* Stack, headed at field "spareWaiters".  Note that the use of
299	<	* spares is intrinsically racy.  One thread may become a spare at
300	<	* about the same time as another is needlessly being created. We
301	<	* counteract this and related slop in part by requiring resumed
302	<	* spares to immediately recheck (in preStep) to see whether they
303	<	* should re-suspend.
304	<	*
305	<	* 6. Killing off unneeded workers. A timeout mechanism is used to
306	<	* shed unused workers: The oldest (first) event queue waiter uses
307	<	* a timed rather than hard wait. When this wait times out without
308	<	* a normal wakeup, it tries to shutdown any one (for convenience
309	<	* the newest) other spare or event waiter via
310	<	* tryShutdownUnusedWorker. This eventually reduces the number of
311	<	* worker threads to a minimum of one after a long enough period
312	<	* without use.
313	<	*
314	<	* 7. Deciding when to create new workers. The main dynamic
315	<	* control in this class is deciding when to create extra threads
316	<	* in method helpMaintainParallelism. We would like to keep
317	<	* exactly #parallelism threads running, which is an impossible
318	<	* task. We always need to create one when the number of running
319	<	* threads would become zero and all workers are busy. Beyond
320	<	* this, we must rely on heuristics that work well in the
321	<	* presence of transient phenomena such as GC stalls, dynamic
322	<	* compilation, and wake-up lags. These transients are extremely
323	<	* common -- we are normally trying to fully saturate the CPUs on
324	<	* a machine, so almost any activity other than running tasks
325	<	* impedes accuracy. Our main defense is to allow parallelism to
326	<	* lapse for a while during joins, and use a timeout to see if,
327	<	* after the resulting settling, there is still a need for
328	<	* additional workers.  This also better copes with the fact that
329	<	* some of the methods in this class tend to never become compiled
330	<	* (but are interpreted), so some components of the entire set of
331	<	* controls might execute 100 times faster than others. And
332	<	* similarly for cases where the apparent lack of work is just due
333	<	* to GC stalls and other transient system activity.
265	>	* It is impossible to keep exactly the target parallelism number
266	>	* of threads running at any given time.  Determining the
267	>	* existence of conservatively safe helping targets, the
268	>	* availability of already-created spares, and the apparent need
269	>	* to create new spares are all racy and require heuristic
270	>	* guidance, so we rely on multiple retries of each.  Currently,
271	>	* in keeping with on-demand signalling policy, we compensate only
272	>	* if blocking would leave less than one active (non-waiting,
273	>	* non-blocked) worker. Additionally, to avoid some false alarms
274	>	* due to GC, lagging counters, system activity, etc, compensated
275	>	* blocking for joins is only attempted after a number of rechecks
276	>	* proportional to the current apparent deficit (where retries are
277	>	* interspersed with Thread.yield, for good citizenship).  The
278	>	* variable blockedCount, incremented before blocking and
279	>	* decremented after, is sometimes needed to distinguish cases of
280	>	* waiting for work vs blocking on joins or other managed sync,
281	>	* but both the cases are equivalent for most pool control, so we
282	>	* can update non-atomically. (Additionally, contention on
283	>	* blockedCount alleviates some contention on ctl).
284	>	*
285	>	* Shutdown and Termination. A call to shutdownNow atomically sets
286	>	* the ctl stop bit and then (non-atomically) sets each workers
287	>	* "terminate" status, cancels all unprocessed tasks, and wakes up
288	>	* all waiting workers.  Detecting whether termination should
289	>	* commence after a non-abrupt shutdown() call requires more work
290	>	* and bookkeeping. We need consensus about quiesence (i.e., that
291	>	* there is no more work) which is reflected in active counts so
292	>	* long as there are no current blockers, as well as possible
293	>	* re-evaluations during independent changes in blocking or
294	>	* quiescing workers.
295		*
296	<	* Beware that there is a lot of representation-level coupling
296	>	* Style notes: There is a lot of representation-level coupling
297		* among classes ForkJoinPool, ForkJoinWorkerThread, and
298	<	* ForkJoinTask.  For example, direct access to "workers" array by
298	>	* ForkJoinTask.  Most fields of ForkJoinWorkerThread maintain
299	>	* data structures managed by ForkJoinPool, so are directly
300	>	* accessed.  Conversely we allow access to "workers" array by
301		* workers, and direct access to ForkJoinTask.status by both
302		* ForkJoinPool and ForkJoinWorkerThread.  There is little point
303		* trying to reduce this, since any associated future changes in
304		* representations will need to be accompanied by algorithmic
305	<	* changes anyway.
306	<	*
307	<	* Style notes: There are lots of inline assignments (of form
308	<	* "while ((local = field) != 0)") which are usually the simplest
309	<	* way to ensure the required read orderings (which are sometimes
310	<	* critical). Also several occurrences of the unusual "do {}
311	<	* while (!cas...)" which is the simplest way to force an update of
312	<	* a CAS'ed variable. There are also other coding oddities that
313	<	* help some methods perform reasonably even when interpreted (not
314	<	* compiled), at the expense of some messy constructions that
315	<	* reduce byte code counts.
316	<	*
317	<	* The order of declarations in this file is: (1) statics (2)
318	<	* fields (along with constants used when unpacking some of them)
319	<	* (3) internal control methods (4) callbacks and other support
320	<	* for ForkJoinTask and ForkJoinWorkerThread classes, (5) exported
321	<	* methods (plus a few little helpers).
305	>	* changes anyway. All together, these low-level implementation
306	>	* choices produce as much as a factor of 4 performance
307	>	* improvement compared to naive implementations, and enable the
308	>	* processing of billions of tasks per second, at the expense of
309	>	* some ugliness.
310	>	*
311	>	* Methods signalWork() and scan() are the main bottlenecks so are
312	>	* especially heavily micro-optimized/mangled.  There are lots of
313	>	* inline assignments (of form "while ((local = field) != 0)")
314	>	* which are usually the simplest way to ensure the required read
315	>	* orderings (which are sometimes critical). This leads to a
316	>	* "C"-like style of listing declarations of these locals at the
317	>	* heads of methods or blocks.  There are several occurrences of
318	>	* the unusual "do {} while (!cas...)"  which is the simplest way
319	>	* to force an update of a CAS'ed variable. There are also other
320	>	* coding oddities that help some methods perform reasonably even
321	>	* when interpreted (not compiled).
322	>	*
323	>	* The order of declarations in this file is: (1) declarations of
324	>	* statics (2) fields (along with constants used when unpacking
325	>	* some of them), listed in an order that tends to reduce
326	>	* contention among them a bit under most JVMs.  (3) internal
327	>	* control methods (4) callbacks and other support for
328	>	* ForkJoinTask and ForkJoinWorkerThread classes, (5) exported
329	>	* methods (plus a few little helpers). (6) static block
330	>	* initializing all statics in a minimally dependent order.
331		*/
332
333		/**
#	Line 390 | Line 362 | public class ForkJoinPool extends Abstra
362		* overridden in ForkJoinPool constructors.
363		*/
364		public static final ForkJoinWorkerThreadFactory
365	<	defaultForkJoinWorkerThreadFactory =
394	<	new DefaultForkJoinWorkerThreadFactory();
365	>	defaultForkJoinWorkerThreadFactory;
366
367		/**
368		* Permission required for callers of methods that may start or
369		* kill threads.
370		*/
371	<	private static final RuntimePermission modifyThreadPermission =
401	<	new RuntimePermission("modifyThread");
371	>	private static final RuntimePermission modifyThreadPermission;
372
373		/**
374		* If there is a security manager, makes sure caller has
#	Line 413 | Line 383 | public class ForkJoinPool extends Abstra
383		/**
384		* Generator for assigning sequence numbers as pool names.
385		*/
386	<	private static final AtomicInteger poolNumberGenerator =
417	<	new AtomicInteger();
386	>	private static final AtomicInteger poolNumberGenerator;
387
388		/**
389	<	* The time to block in a join (see awaitJoin) before checking if
390	<	* a new worker should be (re)started to maintain parallelism
391	<	* level. The value should be short enough to maintain global
392	<	* responsiveness and progress but long enough to avoid
393	<	* counterproductive firings during GC stalls or unrelated system
425	<	* activity, and to not bog down systems with continual re-firings
426	<	* on GCs or legitimately long waits.
389	>	* Generator for initial random seeds for worker victim
390	>	* selection. This is used only to create initial seeds. Random
391	>	* steals use a cheaper xorshift generator per steal attempt. We
392	>	* don't expect much contention on seedGenerator, so just use a
393	>	* plain Random.
394		*/
395	<	private static final long JOIN_TIMEOUT_MILLIS = 250L; // 4 per second
395	>	static final Random workerSeedGenerator;
396
397		/**
398	<	* The wakeup interval (in nanoseconds) for the oldest worker
399	<	* waiting for an event invokes tryShutdownUnusedWorker to shrink
400	<	* the number of workers.  The exact value does not matter too
401	<	* much, but should be long enough to slowly release resources
402	<	* during long periods without use without disrupting normal use.
398	>	* Array holding all worker threads in the pool.  Initialized upon
399	>	* construction. Array size must be a power of two.  Updates and
400	>	* replacements are protected by scanGuard, but the array is
401	>	* always kept in a consistent enough state to be randomly
402	>	* accessed without locking by workers performing work-stealing,
403	>	* as well as other traversal-based methods in this class, so long
404	>	* as reads memory-acquire by first reading ctl. All readers must
405	>	* tolerate that some array slots may be null.
406		*/
407	<	private static final long SHRINK_RATE_NANOS =
438	<	30L * 1000L * 1000L * 1000L; // 2 per minute
407	>	ForkJoinWorkerThread[] workers;
408
409		/**
410	<	* Absolute bound for parallelism level. Twice this number plus
411	<	* one (i.e., 0xfff) must fit into a 16bit field to enable
412	<	* word-packing for some counts and indices.
410	>	* Initial size for submission queue array. Must be a power of
411	>	* two.  In many applications, these always stay small so we use a
412	>	* small initial cap.
413		*/
414	<	private static final int MAX_WORKERS   = 0x7fff;
414	>	private static final int INITIAL_QUEUE_CAPACITY = 8;
415
416		/**
417	<	* Array holding all worker threads in the pool.  Array size must
418	<	* be a power of two.  Updates and replacements are protected by
419	<	* workerLock, but the array is always kept in a consistent enough
420	<	* state to be randomly accessed without locking by workers
452	<	* performing work-stealing, as well as other traversal-based
453	<	* methods in this class. All readers must tolerate that some
454	<	* array slots may be null.
417	>	* Maximum size for submission queue array. Must be a power of two
418	>	* less than or equal to 1 << (31 - width of array entry) to
419	>	* ensure lack of index wraparound, but is capped at a lower
420	>	* value to help users trap runaway computations.
421		*/
422	<	volatile ForkJoinWorkerThread[] workers;
422	>	private static final int MAXIMUM_QUEUE_CAPACITY = 1 << 24; // 16M
423
424		/**
425	<	* Queue for external submissions.
425	>	* Array serving as submission queue. Initialized upon construction.
426		*/
427	<	private final LinkedTransferQueue<ForkJoinTask<?>> submissionQueue;
427	>	private ForkJoinTask<?>[] submissionQueue;
428
429		/**
430	<	* Lock protecting updates to workers array.
430	>	* Lock protecting submissions array for addSubmission
431		*/
432	<	private final ReentrantLock workerLock;
432	>	private final ReentrantLock submissionLock;
433
434		/**
435	<	* Latch released upon termination.
435	>	* Condition for awaitTermination, using submissionLock for
436	>	* convenience.
437		*/
438	<	private final Phaser termination;
438	>	private final Condition termination;
439
440		/**
441		* Creation factory for worker threads.
#	Line 476 | Line 443 | public class ForkJoinPool extends Abstra
443		private final ForkJoinWorkerThreadFactory factory;
444
445		/**
446	<	* Sum of per-thread steal counts, updated only when threads are
447	<	* idle or terminating.
446	>	* The uncaught exception handler used when any worker abruptly
447	>	* terminates.
448		*/
449	<	private volatile long stealCount;
449	>	final Thread.UncaughtExceptionHandler ueh;
450
451		/**
452	<	* Encoded record of top of Treiber stack of threads waiting for
486	<	* events. The top 32 bits contain the count being waited for. The
487	<	* bottom 16 bits contains one plus the pool index of waiting
488	<	* worker thread. (Bits 16-31 are unused.)
452	>	* Prefix for assigning names to worker threads
453		*/
454	<	private volatile long eventWaiters;
491	<
492	<	private static final int  EVENT_COUNT_SHIFT = 32;
493	<	private static final long WAITER_ID_MASK    = (1L << 16) - 1L;
454	>	private final String workerNamePrefix;
455
456		/**
457	<	* A counter for events that may wake up worker threads:
458	<	*   - Submission of a new task to the pool
498	<	*   - A worker pushing a task on an empty queue
499	<	*   - termination
457	>	* Sum of per-thread steal counts, updated only when threads are
458	>	* idle or terminating.
459		*/
460	<	private volatile int eventCount;
460	>	private volatile long stealCount;
461
462		/**
463	<	* Encoded record of top of Treiber stack of spare threads waiting
464	<	* for resumption. The top 16 bits contain an arbitrary count to
465	<	* avoid ABA effects. The bottom 16bits contains one plus the pool
466	<	* index of waiting worker thread.
467	<	*/
468	<	private volatile int spareWaiters;
469	<
470	<	private static final int SPARE_COUNT_SHIFT = 16;
471	<	private static final int SPARE_ID_MASK     = (1 << 16) - 1;
463	>	* Main pool control -- a long packed with:
464	>	* AC: Number of active running workers minus target parallelism (16 bits)
465	>	* TC: Number of total workers minus target parallelism (16bits)
466	>	* ST: true if pool is terminating (1 bit)
467	>	* EC: the wait count of top waiting thread (15 bits)
468	>	* ID: ~poolIndex of top of Treiber stack of waiting threads (16 bits)
469	>	*
470	>	* When convenient, we can extract the upper 32 bits of counts and
471	>	* the lower 32 bits of queue state, u = (int)(ctl >>> 32) and e =
472	>	* (int)ctl.  The ec field is never accessed alone, but always
473	>	* together with id and st. The offsets of counts by the target
474	>	* parallelism and the positionings of fields makes it possible to
475	>	* perform the most common checks via sign tests of fields: When
476	>	* ac is negative, there are not enough active workers, when tc is
477	>	* negative, there are not enough total workers, when id is
478	>	* negative, there is at least one waiting worker, and when e is
479	>	* negative, the pool is terminating.  To deal with these possibly
480	>	* negative fields, we use casts in and out of "short" and/or
481	>	* signed shifts to maintain signedness.
482	>	*/
483	>	volatile long ctl;
484	>
485	>	// bit positions/shifts for fields
486	>	private static final int  AC_SHIFT   = 48;
487	>	private static final int  TC_SHIFT   = 32;
488	>	private static final int  ST_SHIFT   = 31;
489	>	private static final int  EC_SHIFT   = 16;
490	>
491	>	// bounds
492	>	private static final int  MAX_ID     = 0x7fff;  // max poolIndex
493	>	private static final int  SMASK      = 0xffff;  // mask short bits
494	>	private static final int  SHORT_SIGN = 1 << 15;
495	>	private static final int  INT_SIGN   = 1 << 31;
496	>
497	>	// masks
498	>	private static final long STOP_BIT   = 0x0001L << ST_SHIFT;
499	>	private static final long AC_MASK    = ((long)SMASK) << AC_SHIFT;
500	>	private static final long TC_MASK    = ((long)SMASK) << TC_SHIFT;
501	>
502	>	// units for incrementing and decrementing
503	>	private static final long TC_UNIT    = 1L << TC_SHIFT;
504	>	private static final long AC_UNIT    = 1L << AC_SHIFT;
505	>
506	>	// masks and units for dealing with u = (int)(ctl >>> 32)
507	>	private static final int  UAC_SHIFT  = AC_SHIFT - 32;
508	>	private static final int  UTC_SHIFT  = TC_SHIFT - 32;
509	>	private static final int  UAC_MASK   = SMASK << UAC_SHIFT;
510	>	private static final int  UTC_MASK   = SMASK << UTC_SHIFT;
511	>	private static final int  UAC_UNIT   = 1 << UAC_SHIFT;
512	>	private static final int  UTC_UNIT   = 1 << UTC_SHIFT;
513	>
514	>	// masks and units for dealing with e = (int)ctl
515	>	private static final int  E_MASK     = 0x7fffffff; // no STOP_BIT
516	>	private static final int  EC_UNIT    = 1 << EC_SHIFT;
517
518		/**
519	<	* Lifecycle control. The low word contains the number of workers
516	<	* that are (probably) executing tasks. This value is atomically
517	<	* incremented before a worker gets a task to run, and decremented
518	<	* when worker has no tasks and cannot find any.  Bits 16-18
519	<	* contain runLevel value. When all are zero, the pool is
520	<	* running. Level transitions are monotonic (running -> shutdown
521	<	* -> terminating -> terminated) so each transition adds a bit.
522	<	* These are bundled together to ensure consistent read for
523	<	* termination checks (i.e., that runLevel is at least SHUTDOWN
524	<	* and active threads is zero).
525	<	*
526	<	* Notes: Most direct CASes are dependent on these bitfield
527	<	* positions.  Also, this field is non-private to enable direct
528	<	* performance-sensitive CASes in ForkJoinWorkerThread.
519	>	* The target parallelism level.
520		*/
521	<	volatile int runState;
531	<
532	<	// Note: The order among run level values matters.
533	<	private static final int RUNLEVEL_SHIFT     = 16;
534	<	private static final int SHUTDOWN           = 1 << RUNLEVEL_SHIFT;
535	<	private static final int TERMINATING        = 1 << (RUNLEVEL_SHIFT + 1);
536	<	private static final int TERMINATED         = 1 << (RUNLEVEL_SHIFT + 2);
537	<	private static final int ACTIVE_COUNT_MASK  = (1 << RUNLEVEL_SHIFT) - 1;
521	>	final int parallelism;
522
523		/**
524	<	* Holds number of total (i.e., created and not yet terminated)
525	<	* and running (i.e., not blocked on joins or other managed sync)
542	<	* threads, packed together to ensure consistent snapshot when
543	<	* making decisions about creating and suspending spare
544	<	* threads. Updated only by CAS. Note that adding a new worker
545	<	* requires incrementing both counts, since workers start off in
546	<	* running state.
524	>	* Index (mod submission queue length) of next element to take
525	>	* from submission queue.
526		*/
527	<	private volatile int workerCounts;
527	>	volatile int queueBase;
528
529	<	private static final int TOTAL_COUNT_SHIFT  = 16;
530	<	private static final int RUNNING_COUNT_MASK = (1 << TOTAL_COUNT_SHIFT) - 1;
531	<	private static final int ONE_RUNNING        = 1;
532	<	private static final int ONE_TOTAL          = 1 << TOTAL_COUNT_SHIFT;
529	>	/**
530	>	* Index (mod submission queue length) of next element to add
531	>	* in submission queue.
532	>	*/
533	>	int queueTop;
534
535		/**
536	<	* The target parallelism level.
557	<	* Accessed directly by ForkJoinWorkerThreads.
536	>	* True when shutdown() has been called.
537		*/
538	<	final int parallelism;
538	>	volatile boolean shutdown;
539
540		/**
541		* True if use local fifo, not default lifo, for local polling
#	Line 565 | Line 544 | public class ForkJoinPool extends Abstra
544		final boolean locallyFifo;
545
546		/**
547	<	* The uncaught exception handler used when any worker abruptly
548	<	* terminates.
547	>	* The number of threads in ForkJoinWorkerThreads.helpQuiescePool.
548	>	* When non-zero, suppresses automatic shutdown when active
549	>	* counts become zero.
550		*/
551	<	private final Thread.UncaughtExceptionHandler ueh;
551	>	volatile int quiescerCount;
552
553		/**
554	<	* Pool number, just for assigning useful names to worker threads
554	>	* The number of threads blocked in join.
555		*/
556	<	private final int poolNumber;
577	<
578	<	// Utilities for CASing fields. Note that most of these
579	<	// are usually manually inlined by callers
556	>	volatile int blockedCount;
557
558		/**
559	<	* Increments running count part of workerCounts
559	>	* Counter for worker Thread names (unrelated to their poolIndex)
560		*/
561	<	final void incrementRunningCount() {
585	<	int c;
586	<	do {} while (!UNSAFE.compareAndSwapInt(this, workerCountsOffset,
587	<	c = workerCounts,
588	<	c + ONE_RUNNING));
589	<	}
561	>	private volatile int nextWorkerNumber;
562
563		/**
564	<	* Tries to decrement running count unless already zero
564	>	* The index for the next created worker. Accessed under scanGuard.
565		*/
566	<	final boolean tryDecrementRunningCount() {
595	<	int wc = workerCounts;
596	<	if ((wc & RUNNING_COUNT_MASK) == 0)
597	<	return false;
598	<	return UNSAFE.compareAndSwapInt(this, workerCountsOffset,
599	<	wc, wc - ONE_RUNNING);
600	<	}
566	>	private int nextWorkerIndex;
567
568		/**
569	<	* Forces decrement of encoded workerCounts, awaiting nonzero if
570	<	* (rarely) necessary when other count updates lag.
571	<	*
572	<	* @param dr -- either zero or ONE_RUNNING
573	<	* @param dt == either zero or ONE_TOTAL
569	>	* SeqLock and index masking for for updates to workers array.
570	>	* Locked when SG_UNIT is set. Unlocking clears bit by adding
571	>	* SG_UNIT. Staleness of read-only operations can be checked by
572	>	* comparing scanGuard to value before the reads. The low 16 bits
573	>	* (i.e, anding with SMASK) hold (the smallest power of two
574	>	* covering all worker indices, minus one, and is used to avoid
575	>	* dealing with large numbers of null slots when the workers array
576	>	* is overallocated.
577		*/
578	<	private void decrementWorkerCounts(int dr, int dt) {
579	<	for (;;) {
580	<	int wc = workerCounts;
612	<	if ((wc & RUNNING_COUNT_MASK)  - dr < 0 ||
613	<	(wc >>> TOTAL_COUNT_SHIFT) - dt < 0) {
614	<	if ((runState & TERMINATED) != 0)
615	<	return; // lagging termination on a backout
616	<	Thread.yield();
617	<	}
618	<	if (UNSAFE.compareAndSwapInt(this, workerCountsOffset,
619	<	wc, wc - (dr + dt)))
620	<	return;
621	<	}
622	<	}
578	>	volatile int scanGuard;
579	>
580	>	private static final int SG_UNIT = 1 << 16;
581
582		/**
583	<	* Tries decrementing active count; fails on contention.
584	<	* Called when workers cannot find tasks to run.
583	>	* The wakeup interval (in nanoseconds) for a worker waiting for a
584	>	* task when the pool is quiescent to instead try to shrink the
585	>	* number of workers.  The exact value does not matter too
586	>	* much. It must be short enough to release resources during
587	>	* sustained periods of idleness, but not so short that threads
588	>	* are continually re-created.
589		*/
590	<	final boolean tryDecrementActiveCount() {
591	<	int c;
630	<	return UNSAFE.compareAndSwapInt(this, runStateOffset,
631	<	c = runState, c - 1);
632	<	}
590	>	private static final long SHRINK_RATE =
591	>	4L * 1000L * 1000L * 1000L; // 4 seconds
592
593		/**
594	<	* Advances to at least the given level. Returns true if not
595	<	* already in at least the given level.
594	>	* Top-level loop for worker threads: On each step: if the
595	>	* previous step swept through all queues and found no tasks, or
596	>	* there are excess threads, then possibly blocks. Otherwise,
597	>	* scans for and, if found, executes a task. Returns when pool
598	>	* and/or worker terminate.
599	>	*
600	>	* @param w the worker
601		*/
602	<	private boolean advanceRunLevel(int level) {
603	<	for (;;) {
604	<	int s = runState;
605	<	if ((s & level) != 0)
606	<	return false;
607	<	if (UNSAFE.compareAndSwapInt(this, runStateOffset, s, s | level))
608	<	return true;
602	>	final void work(ForkJoinWorkerThread w) {
603	>	boolean swept = false;                // true on empty scans
604	>	long c;
605	>	while (!w.terminate && (int)(c = ctl) >= 0) {
606	>	int a;                            // active count
607	>	if (!swept && (a = (int)(c >> AC_SHIFT)) <= 0)
608	>	swept = scan(w, a);
609	>	else if (tryAwaitWork(w, c))
610	>	swept = false;
611		}
612		}
613
614	<	// workers array maintenance
614	>	// Signalling
615
616		/**
617	<	* Records and returns a workers array index for new worker.
617	>	* Wakes up or creates a worker.
618		*/
619	<	private int recordWorker(ForkJoinWorkerThread w) {
620	<	// Try using slot totalCount-1. If not available, scan and/or resize
621	<	int k = (workerCounts >>> TOTAL_COUNT_SHIFT) - 1;
622	<	final ReentrantLock lock = this.workerLock;
623	<	lock.lock();
624	<	try {
625	<	ForkJoinWorkerThread[] ws = workers;
626	<	int n = ws.length;
627	<	if (k < 0 || k >= n || ws[k] != null) {
628	<	for (k = 0; k < n && ws[k] != null; ++k)
629	<	;
630	<	if (k == n)
631	<	ws = Arrays.copyOf(ws, n << 1);
619	>	final void signalWork() {
620	>	/*
621	>	* The while condition is true if: (there is are too few total
622	>	* workers OR there is at least one waiter) AND (there are too
623	>	* few active workers OR the pool is terminating).  The value
624	>	* of e distinguishes the remaining cases: zero (no waiters)
625	>	* for create, negative if terminating (in which case do
626	>	* nothing), else release a waiter. The secondary checks for
627	>	* release (non-null array etc) can fail if the pool begins
628	>	* terminating after the test, and don't impose any added cost
629	>	* because JVMs must perform null and bounds checks anyway.
630	>	*/
631	>	long c; int e, u;
632	>	while ((((e = (int)(c = ctl)) | (u = (int)(c >>> 32))) &
633	>	(INT_SIGN|SHORT_SIGN)) == (INT_SIGN|SHORT_SIGN) && e >= 0) {
634	>	if (e > 0) {                         // release a waiting worker
635	>	int i; ForkJoinWorkerThread w; ForkJoinWorkerThread[] ws;
636	>	if ((ws = workers) == null ||
637	>	(i = ~e & SMASK) >= ws.length ||
638	>	(w = ws[i]) == null)
639	>	break;
640	>	long nc = (((long)(w.nextWait & E_MASK)) |
641	>	((long)(u + UAC_UNIT) << 32));
642	>	if (w.eventCount == e &&
643	>	UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc)) {
644	>	w.eventCount = (e + EC_UNIT) & E_MASK;
645	>	if (w.parked)
646	>	UNSAFE.unpark(w);
647	>	break;
648	>	}
649	>	}
650	>	else if (UNSAFE.compareAndSwapLong
651	>	(this, ctlOffset, c,
652	>	(long)(((u + UTC_UNIT) & UTC_MASK) |
653	>	((u + UAC_UNIT) & UAC_MASK)) << 32)) {
654	>	addWorker();
655	>	break;
656		}
667	–	ws[k] = w;
668	–	workers = ws; // volatile array write ensures slot visibility
669	–	} finally {
670	–	lock.unlock();
657		}
672	–	return k;
658		}
659
660		/**
661	<	* Nulls out record of worker in workers array.
661	>	* Variant of signalWork to help release waiters on rescans.
662	>	* Tries once to release a waiter if active count < 0.
663	>	*
664	>	* @return false if failed due to contention, else true
665		*/
666	<	private void forgetWorker(ForkJoinWorkerThread w) {
667	<	int idx = w.poolIndex;
668	<	// Locking helps method recordWorker avoid unnecessary expansion
669	<	final ReentrantLock lock = this.workerLock;
670	<	lock.lock();
671	<	try {
672	<	ForkJoinWorkerThread[] ws = workers;
673	<	if (idx >= 0 && idx < ws.length && ws[idx] == w) // verify
674	<	ws[idx] = null;
675	<	} finally {
676	<	lock.unlock();
666	>	private boolean tryReleaseWaiter() {
667	>	long c; int e, i; ForkJoinWorkerThread w; ForkJoinWorkerThread[] ws;
668	>	if ((e = (int)(c = ctl)) > 0 &&
669	>	(int)(c >> AC_SHIFT) < 0 &&
670	>	(ws = workers) != null &&
671	>	(i = ~e & SMASK) < ws.length &&
672	>	(w = ws[i]) != null) {
673	>	long nc = ((long)(w.nextWait & E_MASK) |
674	>	((c + AC_UNIT) & (AC_MASK|TC_MASK)));
675	>	if (w.eventCount != e ||
676	>	!UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc))
677	>	return false;
678	>	w.eventCount = (e + EC_UNIT) & E_MASK;
679	>	if (w.parked)
680	>	UNSAFE.unpark(w);
681		}
682	+	return true;
683		}
684
685	+	// Scanning for tasks
686	+
687		/**
688	<	* Final callback from terminating worker.  Removes record of
689	<	* worker from array, and adjusts counts. If pool is shutting
690	<	* down, tries to complete termination.
688	>	* Scans for and, if found, executes one task. Scans start at a
689	>	* random index of workers array, and randomly select the first
690	>	* (2*#workers)-1 probes, and then, if all empty, resort to 2
691	>	* circular sweeps, which is necessary to check quiescence. and
692	>	* taking a submission only if no stealable tasks were found.  The
693	>	* steal code inside the loop is a specialized form of
694	>	* ForkJoinWorkerThread.deqTask, followed bookkeeping to support
695	>	* helpJoinTask and signal propagation. The code for submission
696	>	* queues is almost identical. On each steal, the worker completes
697	>	* not only the task, but also all local tasks that this task may
698	>	* have generated. On detecting staleness or contention when
699	>	* trying to take a task, this method returns without finishing
700	>	* sweep, which allows global state rechecks before retry.
701		*
702		* @param w the worker
703	+	* @param a the number of active workers
704	+	* @return true if swept all queues without finding a task
705		*/
706	<	final void workerTerminated(ForkJoinWorkerThread w) {
707	<	forgetWorker(w);
708	<	decrementWorkerCounts(w.isTrimmed()? 0 : ONE_RUNNING, ONE_TOTAL);
702	<	while (w.stealCount != 0) // collect final count
703	<	tryAccumulateStealCount(w);
704	<	tryTerminate(false);
705	<	}
706	<
707	<	// Waiting for and signalling events
708	<
709	<	/**
710	<	* Releases workers blocked on a count not equal to current count.
711	<	* Normally called after precheck that eventWaiters isn't zero to
712	<	* avoid wasted array checks. Gives up upon a change in count or
713	<	* upon releasing two workers, letting others take over.
714	<	*/
715	<	private void releaseEventWaiters() {
706	>	private boolean scan(ForkJoinWorkerThread w, int a) {
707	>	int g = scanGuard; // mask 0 avoids useless scans if only one active
708	>	int m = parallelism == 1 - a? 0 : g & SMASK;
709		ForkJoinWorkerThread[] ws = workers;
710	<	int n = ws.length;
711	<	long h = eventWaiters;
712	<	int ec = eventCount;
713	<	boolean releasedOne = false;
714	<	ForkJoinWorkerThread w; int id;
715	<	while ((id = ((int)(h & WAITER_ID_MASK)) - 1) >= 0 &&
716	<	(int)(h >>> EVENT_COUNT_SHIFT) != ec &&
717	<	id < n && (w = ws[id]) != null) {
718	<	if (UNSAFE.compareAndSwapLong(this, eventWaitersOffset,
719	<	h,  w.nextWaiter)) {
720	<	LockSupport.unpark(w);
721	<	if (releasedOne) // exit on second release
722	<	break;
723	<	releasedOne = true;
710	>	if (ws == null || ws.length <= m)         // staleness check
711	>	return false;
712	>	for (int r = w.seed, k = r, j = -(m + m); j <= m + m; ++j) {
713	>	ForkJoinTask<?> t; ForkJoinTask<?>[] q; int b, i;
714	>	ForkJoinWorkerThread v = ws[k & m];
715	>	if (v != null && (b = v.queueBase) != v.queueTop &&
716	>	(q = v.queue) != null && (i = (q.length - 1) & b) >= 0) {
717	>	long u = (i << ASHIFT) + ABASE;
718	>	if ((t = q[i]) != null && v.queueBase == b &&
719	>	UNSAFE.compareAndSwapObject(q, u, t, null)) {
720	>	int d = (v.queueBase = b + 1) - v.queueTop;
721	>	v.stealHint = w.poolIndex;
722	>	if (d != 0)
723	>	signalWork();             // propagate if nonempty
724	>	w.execTask(t);
725	>	}
726	>	r ^= r << 13; r ^= r >>> 17; w.seed = r ^ (r << 5);
727	>	return false;                     // store next seed
728		}
729	<	if (eventCount != ec)
730	<	break;
731	<	h = eventWaiters;
729	>	else if (j < 0) {                     // xorshift
730	>	r ^= r << 13; r ^= r >>> 17; k = r ^= r << 5;
731	>	}
732	>	else
733	>	++k;
734	>	}
735	>	if (scanGuard != g)                       // staleness check
736	>	return false;
737	>	else {                                    // try to take submission
738	>	ForkJoinTask<?> t; ForkJoinTask<?>[] q; int b, i;
739	>	if ((b = queueBase) != queueTop &&
740	>	(q = submissionQueue) != null &&
741	>	(i = (q.length - 1) & b) >= 0) {
742	>	long u = (i << ASHIFT) + ABASE;
743	>	if ((t = q[i]) != null && queueBase == b &&
744	>	UNSAFE.compareAndSwapObject(q, u, t, null)) {
745	>	queueBase = b + 1;
746	>	w.execTask(t);
747	>	}
748	>	return false;
749	>	}
750	>	return true;                         // all queues empty
751		}
752		}
753
754		/**
755	<	* Tries to advance eventCount and releases waiters. Called only
756	<	* from workers.
757	<	*/
758	<	final void signalWork() {
759	<	int c; // try to increment event count -- CAS failure OK
760	<	UNSAFE.compareAndSwapInt(this, eventCountOffset, c = eventCount, c+1);
761	<	if (eventWaiters != 0L)
762	<	releaseEventWaiters();
763	<	}
764	<
765	<	/**
766	<	* Adds the given worker to event queue and blocks until
767	<	* terminating or event count advances from the given value
768	<	*
769	<	* @param w the calling worker thread
770	<	* @param ec the count
771	<	*/
772	<	private void eventSync(ForkJoinWorkerThread w, int ec) {
773	<	long nh = (((long)ec) << EVENT_COUNT_SHIFT) | ((long)(w.poolIndex+1));
774	<	long h;
775	<	while ((runState < SHUTDOWN || !tryTerminate(false)) &&
776	<	(((int)((h = eventWaiters) & WAITER_ID_MASK)) == 0 ||
777	<	(int)(h >>> EVENT_COUNT_SHIFT) == ec) &&
778	<	eventCount == ec) {
779	<	if (UNSAFE.compareAndSwapLong(this, eventWaitersOffset,
780	<	w.nextWaiter = h, nh)) {
781	<	awaitEvent(w, ec);
782	<	break;
755	>	* Tries to enqueue worker w in wait queue and await change in
756	>	* worker's eventCount.  If the pool is quiescent, possibly
757	>	* terminates worker upon exit.  Otherwise, before blocking,
758	>	* rescans queues to avoid missed signals.  Upon finding work,
759	>	* releases at least one worker (which may be the current
760	>	* worker). Rescans restart upon detected staleness or failure to
761	>	* release due to contention.
762	>	*
763	>	* @param w the calling worker
764	>	* @param c the ctl value on entry
765	>	* @return true if waited or another thread was released upon enq
766	>	*/
767	>	private boolean tryAwaitWork(ForkJoinWorkerThread w, long c) {
768	>	int v = w.eventCount;
769	>	w.nextWait = (int)c;                      // w's successor record
770	>	long nc = (long)(v & E_MASK) | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
771	>	if (ctl != c || !UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc)) {
772	>	long d = ctl; // return true if lost to a deq, to force scan
773	>	return (int)d != (int)c && ((d - c) & AC_MASK) >= 0L;
774	>	}
775	>	for (int sc = w.stealCount; sc != 0;) {   // accumulate stealCount
776	>	long s = stealCount;
777	>	if (UNSAFE.compareAndSwapLong(this, stealCountOffset, s, s + sc))
778	>	sc = w.stealCount = 0;
779	>	else if (w.eventCount != v)
780	>	return true;                      // update next time
781	>	}
782	>	if (parallelism + (int)(nc >> AC_SHIFT) == 0 &&
783	>	blockedCount == 0 && quiescerCount == 0)
784	>	idleAwaitWork(w, nc, c, v);           // quiescent
785	>	for (boolean rescanned = false;;) {
786	>	if (w.eventCount != v)
787	>	return true;
788	>	if (!rescanned) {
789	>	int g = scanGuard, m = g & SMASK;
790	>	ForkJoinWorkerThread[] ws = workers;
791	>	if (ws != null && m < ws.length) {
792	>	rescanned = true;
793	>	for (int i = 0; i <= m; ++i) {
794	>	ForkJoinWorkerThread u = ws[i];
795	>	if (u != null) {
796	>	if (u.queueBase != u.queueTop &&
797	>	!tryReleaseWaiter())
798	>	rescanned = false; // contended
799	>	if (w.eventCount != v)
800	>	return true;
801	>	}
802	>	}
803	>	}
804	>	if (scanGuard != g ||              // stale
805	>	(queueBase != queueTop && !tryReleaseWaiter()))
806	>	rescanned = false;
807	>	if (!rescanned)
808	>	Thread.yield();                // reduce contention
809	>	else
810	>	Thread.interrupted();          // clear before park
811	>	}
812	>	else {
813	>	w.parked = true;                   // must recheck
814	>	if (w.eventCount != v) {
815	>	w.parked = false;
816	>	return true;
817	>	}
818	>	LockSupport.park(this);
819	>	rescanned = w.parked = false;
820		}
821		}
822		}
823
824		/**
825	<	* Blocks the given worker (that has already been entered as an
826	<	* event waiter) until terminating or event count advances from
827	<	* the given value. The oldest (first) waiter uses a timed wait to
828	<	* occasionally one-by-one shrink the number of workers (to a
829	<	* minimum of one) if the pool has not been used for extended
830	<	* periods.
831	<	*
832	<	* @param w the calling worker thread
833	<	* @param ec the count
834	<	*/
835	<	private void awaitEvent(ForkJoinWorkerThread w, int ec) {
836	<	while (eventCount == ec) {
837	<	if (tryAccumulateStealCount(w)) { // transfer while idle
838	<	boolean untimed = (w.nextWaiter != 0L ||
839	<	(workerCounts & RUNNING_COUNT_MASK) <= 1);
840	<	long startTime = untimed? 0 : System.nanoTime();
841	<	Thread.interrupted();         // clear/ignore interrupt
842	<	if (eventCount != ec || w.runState != 0 ||
843	<	runState >= TERMINATING)  // recheck after clear
825	>	* If inactivating worker w has caused pool to become
826	>	* quiescent, check for pool termination, and wait for event
827	>	* for up to SHRINK_RATE nanosecs (rescans are unnecessary in
828	>	* this case because quiescence reflects consensus about lack
829	>	* of work). On timeout, if ctl has not changed, terminate the
830	>	* worker. Upon its termination (see deregisterWorker), it may
831	>	* wake up another worker to possibly repeat this process.
832	>	*
833	>	* @param w the calling worker
834	>	* @param currentCtl the ctl value after enqueuing w
835	>	* @param prevCtl the ctl value if w terminated
836	>	* @param v the eventCount w awaits change
837	>	*/
838	>	private void idleAwaitWork(ForkJoinWorkerThread w, long currentCtl,
839	>	long prevCtl, int v) {
840	>	if (w.eventCount == v) {
841	>	if (shutdown)
842	>	tryTerminate(false);
843	>	ForkJoinTask.helpExpungeStaleExceptions(); // help clean weak refs
844	>	while (ctl == currentCtl) {
845	>	long startTime = System.nanoTime();
846	>	w.parked = true;
847	>	if (w.eventCount == v)             // must recheck
848	>	LockSupport.parkNanos(this, SHRINK_RATE);
849	>	w.parked = false;
850	>	if (w.eventCount != v)
851	>	break;
852	>	else if (System.nanoTime() - startTime < SHRINK_RATE)
853	>	Thread.interrupted();          // spurious wakeup
854	>	else if (UNSAFE.compareAndSwapLong(this, ctlOffset,
855	>	currentCtl, prevCtl)) {
856	>	w.terminate = true;            // restore previous
857	>	w.eventCount = ((int)currentCtl + EC_UNIT) & E_MASK;
858		break;
792	–	if (untimed)
793	–	LockSupport.park(w);
794	–	else {
795	–	LockSupport.parkNanos(w, SHRINK_RATE_NANOS);
796	–	if (eventCount != ec || w.runState != 0 ||
797	–	runState >= TERMINATING)
798	–	break;
799	–	if (System.nanoTime() - startTime >= SHRINK_RATE_NANOS)
800	–	tryShutdownUnusedWorker(ec);
859		}
860		}
861		}
862		}
863
864	<	// Maintaining parallelism
864	>	// Submissions
865
866		/**
867	<	* Pushes worker onto the spare stack.
867	>	* Enqueues the given task in the submissionQueue.  Same idea as
868	>	* ForkJoinWorkerThread.pushTask except for use of submissionLock.
869	>	*
870	>	* @param t the task
871		*/
872	<	final void pushSpare(ForkJoinWorkerThread w) {
873	<	int ns = (++w.spareCount << SPARE_COUNT_SHIFT) | (w.poolIndex + 1);
874	<	do {} while (!UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
875	<	w.nextSpare = spareWaiters,ns));
872	>	private void addSubmission(ForkJoinTask<?> t) {
873	>	final ReentrantLock lock = this.submissionLock;
874	>	lock.lock();
875	>	try {
876	>	ForkJoinTask<?>[] q; int s, m;
877	>	if ((q = submissionQueue) != null) {    // ignore if queue removed
878	>	long u = (((s = queueTop) & (m = q.length-1)) << ASHIFT)+ABASE;
879	>	UNSAFE.putOrderedObject(q, u, t);
880	>	queueTop = s + 1;
881	>	if (s - queueBase == m)
882	>	growSubmissionQueue();
883	>	}
884	>	} finally {
885	>	lock.unlock();
886	>	}
887	>	signalWork();
888		}
889
890	+	//  (pollSubmission is defined below with exported methods)
891	+
892		/**
893	<	* Tries (once) to resume a spare if the number of running
894	<	* threads is less than target.
893	>	* Creates or doubles submissionQueue array.
894	>	* Basically identical to ForkJoinWorkerThread version
895		*/
896	<	private void tryResumeSpare() {
897	<	int sw, id;
898	<	ForkJoinWorkerThread[] ws = workers;
899	<	int n = ws.length;
900	<	ForkJoinWorkerThread w;
901	<	if ((sw = spareWaiters) != 0 &&
902	<	(id = (sw & SPARE_ID_MASK) - 1) >= 0 &&
903	<	id < n && (w = ws[id]) != null &&
904	<	(workerCounts & RUNNING_COUNT_MASK) < parallelism &&
905	<	spareWaiters == sw &&
906	<	UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
907	<	sw, w.nextSpare)) {
908	<	int c; // increment running count before resume
909	<	do {} while (!UNSAFE.compareAndSwapInt
910	<	(this, workerCountsOffset,
911	<	c = workerCounts, c + ONE_RUNNING));
912	<	if (w.tryUnsuspend())
913	<	LockSupport.unpark(w);
914	<	else   // back out if w was shutdown
840	<	decrementWorkerCounts(ONE_RUNNING, 0);
896	>	private void growSubmissionQueue() {
897	>	ForkJoinTask<?>[] oldQ = submissionQueue;
898	>	int size = oldQ != null ? oldQ.length << 1 : INITIAL_QUEUE_CAPACITY;
899	>	if (size > MAXIMUM_QUEUE_CAPACITY)
900	>	throw new RejectedExecutionException("Queue capacity exceeded");
901	>	if (size < INITIAL_QUEUE_CAPACITY)
902	>	size = INITIAL_QUEUE_CAPACITY;
903	>	ForkJoinTask<?>[] q = submissionQueue = new ForkJoinTask<?>[size];
904	>	int mask = size - 1;
905	>	int top = queueTop;
906	>	int oldMask;
907	>	if (oldQ != null && (oldMask = oldQ.length - 1) >= 0) {
908	>	for (int b = queueBase; b != top; ++b) {
909	>	long u = ((b & oldMask) << ASHIFT) + ABASE;
910	>	Object x = UNSAFE.getObjectVolatile(oldQ, u);
911	>	if (x != null && UNSAFE.compareAndSwapObject(oldQ, u, x, null))
912	>	UNSAFE.putObjectVolatile
913	>	(q, ((b & mask) << ASHIFT) + ABASE, x);
914	>	}
915		}
916		}
917
918	+	// Blocking support
919	+
920		/**
921	<	* Tries to increase the number of running workers if below target
922	<	* parallelism: If a spare exists tries to resume it via
923	<	* tryResumeSpare.  Otherwise, if not enough total workers or all
924	<	* existing workers are busy, adds a new worker. In all cases also
925	<	* helps wake up releasable workers waiting for work.
926	<	*/
927	<	private void helpMaintainParallelism() {
928	<	int pc = parallelism;
929	<	int wc, rs, tc;
930	<	while (((wc = workerCounts) & RUNNING_COUNT_MASK) < pc &&
931	<	(rs = runState) < TERMINATING) {
932	<	if (spareWaiters != 0)
933	<	tryResumeSpare();
934	<	else if ((tc = wc >>> TOTAL_COUNT_SHIFT) >= MAX_WORKERS ||
935	<	(tc >= pc && (rs & ACTIVE_COUNT_MASK) != tc))
936	<	break;   // enough total
937	<	else if (runState == rs && workerCounts == wc &&
938	<	UNSAFE.compareAndSwapInt(this, workerCountsOffset, wc,
939	<	wc + (ONE_RUNNING|ONE_TOTAL))) {
940	<	ForkJoinWorkerThread w = null;
941	<	try {
942	<	w = factory.newThread(this);
943	<	} finally { // adjust on null or exceptional factory return
944	<	if (w == null) {
945	<	decrementWorkerCounts(ONE_RUNNING, ONE_TOTAL);
946	<	tryTerminate(false); // handle failure during shutdown
921	>	* Tries to increment blockedCount, decrement active count
922	>	* (sometimes implicitly) and possibly release or create a
923	>	* compensating worker in preparation for blocking. Fails
924	>	* on contention or termination.
925	>	*
926	>	* @return true if the caller can block, else should recheck and retry
927	>	*/
928	>	private boolean tryPreBlock() {
929	>	int b = blockedCount;
930	>	if (UNSAFE.compareAndSwapInt(this, blockedCountOffset, b, b + 1)) {
931	>	int pc = parallelism;
932	>	do {
933	>	ForkJoinWorkerThread[] ws; ForkJoinWorkerThread w;
934	>	int e, ac, tc, rc, i;
935	>	long c = ctl;
936	>	int u = (int)(c >>> 32);
937	>	if ((e = (int)c) < 0) {
938	>	// skip -- terminating
939	>	}
940	>	else if ((ac = (u >> UAC_SHIFT)) <= 0 && e != 0 &&
941	>	(ws = workers) != null &&
942	>	(i = ~e & SMASK) < ws.length &&
943	>	(w = ws[i]) != null) {
944	>	long nc = ((long)(w.nextWait & E_MASK) |
945	>	(c & (AC_MASK|TC_MASK)));
946	>	if (w.eventCount == e &&
947	>	UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc)) {
948	>	w.eventCount = (e + EC_UNIT) & E_MASK;
949	>	if (w.parked)
950	>	UNSAFE.unpark(w);
951	>	return true;             // release an idle worker
952		}
953		}
954	<	if (w == null)
955	<	break;
956	<	w.start(recordWorker(w), ueh);
957	<	if ((workerCounts >>> TOTAL_COUNT_SHIFT) >= pc) {
877	<	int c; // advance event count
878	<	UNSAFE.compareAndSwapInt(this, eventCountOffset,
879	<	c = eventCount, c+1);
880	<	break; // add at most one unless total below target
954	>	else if ((tc = (short)(u >>> UTC_SHIFT)) >= 0 && ac + pc > 1) {
955	>	long nc = ((c - AC_UNIT) & AC_MASK) | (c & ~AC_MASK);
956	>	if (UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc))
957	>	return true;             // no compensation needed
958		}
959	<	}
959	>	else if (tc + pc < MAX_ID) {
960	>	long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK);
961	>	if (UNSAFE.compareAndSwapLong(this, ctlOffset, c, nc)) {
962	>	addWorker();
963	>	return true;            // create a replacement
964	>	}
965	>	}
966	>	// try to back out on any failure and let caller retry
967	>	} while (!UNSAFE.compareAndSwapInt(this, blockedCountOffset,
968	>	b = blockedCount, b - 1));
969		}
970	<	if (eventWaiters != 0L)
885	<	releaseEventWaiters();
970	>	return false;
971		}
972
973		/**
974	<	* Callback from the oldest waiter in awaitEvent waking up after a
890	<	* period of non-use. If all workers are idle, tries (once) to
891	<	* shutdown an event waiter or a spare, if one exists. Note that
892	<	* we don't need CAS or locks here because the method is called
893	<	* only from one thread occasionally waking (and even misfires are
894	<	* OK). Note that until the shutdown worker fully terminates,
895	<	* workerCounts will overestimate total count, which is tolerable.
896	<	*
897	<	* @param ec the event count waited on by caller (to abort
898	<	* attempt if count has since changed).
974	>	* Decrements blockedCount and increments active count
975		*/
976	<	private void tryShutdownUnusedWorker(int ec) {
977	<	if (runState == 0 && eventCount == ec) { // only trigger if all idle
978	<	ForkJoinWorkerThread[] ws = workers;
979	<	int n = ws.length;
980	<	ForkJoinWorkerThread w = null;
981	<	boolean shutdown = false;
982	<	int sw;
907	<	long h;
908	<	if ((sw = spareWaiters) != 0) { // prefer killing spares
909	<	int id = (sw & SPARE_ID_MASK) - 1;
910	<	if (id >= 0 && id < n && (w = ws[id]) != null &&
911	<	UNSAFE.compareAndSwapInt(this, spareWaitersOffset,
912	<	sw, w.nextSpare))
913	<	shutdown = true;
914	<	}
915	<	else if ((h = eventWaiters) != 0L) {
916	<	long nh;
917	<	int id = ((int)(h & WAITER_ID_MASK)) - 1;
918	<	if (id >= 0 && id < n && (w = ws[id]) != null &&
919	<	(nh = w.nextWaiter) != 0L && // keep at least one worker
920	<	UNSAFE.compareAndSwapLong(this, eventWaitersOffset, h, nh))
921	<	shutdown = true;
922	<	}
923	<	if (w != null && shutdown) {
924	<	w.shutdown();
925	<	LockSupport.unpark(w);
926	<	}
927	<	}
928	<	releaseEventWaiters(); // in case of interference
976	>	private void postBlock() {
977	>	long c;
978	>	do {} while (!UNSAFE.compareAndSwapLong(this, ctlOffset,  // no mask
979	>	c = ctl, c + AC_UNIT));
980	>	int b;
981	>	do {} while(!UNSAFE.compareAndSwapInt(this, blockedCountOffset,
982	>	b = blockedCount, b - 1));
983		}
984
985		/**
986	<	* Callback from workers invoked upon each top-level action (i.e.,
987	<	* stealing a task or taking a submission and running it).
934	<	* Performs one or more of the following:
935	<	*
936	<	* 1. If the worker is active and either did not run a task
937	<	*    or there are too many workers, try to set its active status
938	<	*    to inactive and update activeCount. On contention, we may
939	<	*    try again in this or a subsequent call.
940	<	*
941	<	* 2. If not enough total workers, help create some.
942	<	*
943	<	* 3. If there are too many running workers, suspend this worker
944	<	*    (first forcing inactive if necessary).  If it is not needed,
945	<	*    it may be shutdown while suspended (via
946	<	*    tryShutdownUnusedWorker).  Otherwise, upon resume it
947	<	*    rechecks running thread count and need for event sync.
948	<	*
949	<	* 4. If worker did not run a task, await the next task event via
950	<	*    eventSync if necessary (first forcing inactivation), upon
951	<	*    which the worker may be shutdown via
952	<	*    tryShutdownUnusedWorker.  Otherwise, help release any
953	<	*    existing event waiters that are now releasable,
986	>	* Possibly blocks waiting for the given task to complete, or
987	>	* cancels the task if terminating.  Fails to wait if contended.
988		*
989	<	* @param w the worker
956	<	* @param ran true if worker ran a task since last call to this method
989	>	* @param joinMe the task
990		*/
991	<	final void preStep(ForkJoinWorkerThread w, boolean ran) {
992	<	int wec = w.lastEventCount;
993	<	boolean active = w.active;
994	<	boolean inactivate = false;
995	<	int pc = parallelism;
996	<	int rs;
997	<	while (w.runState == 0 && (rs = runState) < TERMINATING) {
965	<	if ((inactivate || (active && (rs & ACTIVE_COUNT_MASK) >= pc)) &&
966	<	UNSAFE.compareAndSwapInt(this, runStateOffset, rs, rs - 1))
967	<	inactivate = active = w.active = false;
968	<	int wc = workerCounts;
969	<	if ((wc & RUNNING_COUNT_MASK) > pc) {
970	<	if (!(inactivate |= active) && // must inactivate to suspend
971	<	workerCounts == wc &&      // try to suspend as spare
972	<	UNSAFE.compareAndSwapInt(this, workerCountsOffset,
973	<	wc, wc - ONE_RUNNING))
974	<	w.suspendAsSpare();
975	<	}
976	<	else if ((wc >>> TOTAL_COUNT_SHIFT) < pc)
977	<	helpMaintainParallelism();     // not enough workers
978	<	else if (!ran) {
979	<	long h = eventWaiters;
980	<	int ec = eventCount;
981	<	if (h != 0L && (int)(h >>> EVENT_COUNT_SHIFT) != ec)
982	<	releaseEventWaiters();     // release others before waiting
983	<	else if (ec != wec) {
984	<	w.lastEventCount = ec;     // no need to wait
985	<	break;
986	<	}
987	<	else if (!(inactivate |= active))
988	<	eventSync(w, wec);         // must inactivate before sync
991	>	final void tryAwaitJoin(ForkJoinTask<?> joinMe) {
992	>	int s;
993	>	Thread.interrupted(); // clear interrupts before checking termination
994	>	if (joinMe.status >= 0) {
995	>	if (tryPreBlock()) {
996	>	joinMe.tryAwaitDone(0L);
997	>	postBlock();
998		}
999	<	else
1000	<	break;
999	>	if ((ctl & STOP_BIT) != 0L)
1000	>	joinMe.cancelIgnoringExceptions();
1001		}
1002		}
1003
1004		/**
1005	<	* Helps and/or blocks awaiting join of the given task.
1006	<	* See above for explanation.
1005	>	* Possibly blocks the given worker waiting for joinMe to
1006	>	* complete or timeout
1007		*
1008	<	* @param joinMe the task to join
1009	<	* @param worker the current worker thread
1008	>	* @param joinMe the task
1009	>	* @param millis the wait time for underlying Object.wait
1010		*/
1011	<	final void awaitJoin(ForkJoinTask<?> joinMe, ForkJoinWorkerThread worker) {
1003	<	int retries = 2 + (parallelism >> 2); // #helpJoins before blocking
1011	>	final void timedAwaitJoin(ForkJoinTask<?> joinMe, long nanos) {
1012		while (joinMe.status >= 0) {
1013	<	int wc;
1014	<	worker.helpJoinTask(joinMe);
1015	<	if (joinMe.status < 0)
1013	>	Thread.interrupted();
1014	>	if ((ctl & STOP_BIT) != 0L) {
1015	>	joinMe.cancelIgnoringExceptions();
1016	>	break;
1017	>	}
1018	>	if (tryPreBlock()) {
1019	>	long last = System.nanoTime();
1020	>	while (joinMe.status >= 0) {
1021	>	long millis = TimeUnit.NANOSECONDS.toMillis(nanos);
1022	>	if (millis <= 0)
1023	>	break;
1024	>	joinMe.tryAwaitDone(millis);
1025	>	if (joinMe.status < 0)
1026	>	break;
1027	>	if ((ctl & STOP_BIT) != 0L) {
1028	>	joinMe.cancelIgnoringExceptions();
1029	>	break;
1030	>	}
1031	>	long now = System.nanoTime();
1032	>	nanos -= now - last;
1033	>	last = now;
1034	>	}
1035	>	postBlock();
1036		break;
1009	–	else if (retries > 0)
1010	–	--retries;
1011	–	else if (((wc = workerCounts) & RUNNING_COUNT_MASK) != 0 &&
1012	–	UNSAFE.compareAndSwapInt(this, workerCountsOffset,
1013	–	wc, wc - ONE_RUNNING)) {
1014	–	int stat, c; long h;
1015	–	while ((stat = joinMe.status) >= 0 &&
1016	–	(h = eventWaiters) != 0L && // help release others
1017	–	(int)(h >>> EVENT_COUNT_SHIFT) != eventCount)
1018	–	releaseEventWaiters();
1019	–	if (stat >= 0 &&
1020	–	((workerCounts & RUNNING_COUNT_MASK) == 0 ||
1021	–	(stat =
1022	–	joinMe.internalAwaitDone(JOIN_TIMEOUT_MILLIS)) >= 0))
1023	–	helpMaintainParallelism(); // timeout or no running workers
1024	–	do {} while (!UNSAFE.compareAndSwapInt
1025	–	(this, workerCountsOffset,
1026	–	c = workerCounts, c + ONE_RUNNING));
1027	–	if (stat < 0)
1028	–	break;   // else restart
1037		}
1038		}
1039		}
1040
1041		/**
1042	<	* Same idea as awaitJoin, but no helping, retries, or timeouts.
1042	>	* If necessary, compensates for blocker, and blocks
1043		*/
1044	<	final void awaitBlocker(ManagedBlocker blocker)
1044	>	private void awaitBlocker(ManagedBlocker blocker)
1045		throws InterruptedException {
1046		while (!blocker.isReleasable()) {
1047	<	int wc = workerCounts;
1040	<	if ((wc & RUNNING_COUNT_MASK) != 0 &&
1041	<	UNSAFE.compareAndSwapInt(this, workerCountsOffset,
1042	<	wc, wc - ONE_RUNNING)) {
1047	>	if (tryPreBlock()) {
1048		try {
1049	<	while (!blocker.isReleasable()) {
1045	<	long h = eventWaiters;
1046	<	if (h != 0L &&
1047	<	(int)(h >>> EVENT_COUNT_SHIFT) != eventCount)
1048	<	releaseEventWaiters();
1049	<	else if ((workerCounts & RUNNING_COUNT_MASK) == 0 &&
1050	<	runState < TERMINATING)
1051	<	helpMaintainParallelism();
1052	<	else if (blocker.block())
1053	<	break;
1054	<	}
1049	>	do {} while (!blocker.isReleasable() && !blocker.block());
1050		} finally {
1051	<	int c;
1057	<	do {} while (!UNSAFE.compareAndSwapInt
1058	<	(this, workerCountsOffset,
1059	<	c = workerCounts, c + ONE_RUNNING));
1051	>	postBlock();
1052		}
1053		break;
1054		}
1055		}
1056		}
1057
1058	+	// Creating, registering and deregistring workers
1059	+
1060	+	/**
1061	+	* Tries to create and start a worker; minimally rolls back counts
1062	+	* on failure.
1063	+	*/
1064	+	private void addWorker() {
1065	+	Throwable ex = null;
1066	+	ForkJoinWorkerThread t = null;
1067	+	try {
1068	+	t = factory.newThread(this);
1069	+	} catch (Throwable e) {
1070	+	ex = e;
1071	+	}
1072	+	if (t == null) {  // null or exceptional factory return
1073	+	long c;       // adjust counts
1074	+	do {} while (!UNSAFE.compareAndSwapLong
1075	+	(this, ctlOffset, c = ctl,
1076	+	(((c - AC_UNIT) & AC_MASK) |
1077	+	((c - TC_UNIT) & TC_MASK) |
1078	+	(c & ~(AC_MASK|TC_MASK)))));
1079	+	// Propagate exception if originating from an external caller
1080	+	if (!tryTerminate(false) && ex != null &&
1081	+	!(Thread.currentThread() instanceof ForkJoinWorkerThread))
1082	+	UNSAFE.throwException(ex);
1083	+	}
1084	+	else
1085	+	t.start();
1086	+	}
1087	+
1088	+	/**
1089	+	* Callback from ForkJoinWorkerThread constructor to assign a
1090	+	* public name
1091	+	*/
1092	+	final String nextWorkerName() {
1093	+	for (int n;;) {
1094	+	if (UNSAFE.compareAndSwapInt(this, nextWorkerNumberOffset,
1095	+	n = nextWorkerNumber, ++n))
1096	+	return workerNamePrefix + n;
1097	+	}
1098	+	}
1099	+
1100	+	/**
1101	+	* Callback from ForkJoinWorkerThread constructor to
1102	+	* determine its poolIndex and record in workers array.
1103	+	*
1104	+	* @param w the worker
1105	+	* @return the worker's pool index
1106	+	*/
1107	+	final int registerWorker(ForkJoinWorkerThread w) {
1108	+	/*
1109	+	* In the typical case, a new worker acquires the lock, uses
1110	+	* next available index and returns quickly.  Since we should
1111	+	* not block callers (ultimately from signalWork or
1112	+	* tryPreBlock) waiting for the lock needed to do this, we
1113	+	* instead help release other workers while waiting for the
1114	+	* lock.
1115	+	*/
1116	+	for (int g;;) {
1117	+	ForkJoinWorkerThread[] ws;
1118	+	if (((g = scanGuard) & SG_UNIT) == 0 &&
1119	+	UNSAFE.compareAndSwapInt(this, scanGuardOffset,
1120	+	g, g | SG_UNIT)) {
1121	+	int k = nextWorkerIndex;
1122	+	try {
1123	+	if ((ws = workers) != null) { // ignore on shutdown
1124	+	int n = ws.length;
1125	+	if (k < 0 || k >= n || ws[k] != null) {
1126	+	for (k = 0; k < n && ws[k] != null; ++k)
1127	+	;
1128	+	if (k == n)
1129	+	ws = workers = Arrays.copyOf(ws, n << 1);
1130	+	}
1131	+	ws[k] = w;
1132	+	nextWorkerIndex = k + 1;
1133	+	int m = g & SMASK;
1134	+	g = k >= m? ((m << 1) + 1) & SMASK : g + (SG_UNIT<<1);
1135	+	}
1136	+	} finally {
1137	+	scanGuard = g;
1138	+	}
1139	+	return k;
1140	+	}
1141	+	else if ((ws = workers) != null) { // help release others
1142	+	for (ForkJoinWorkerThread u : ws) {
1143	+	if (u != null && u.queueBase != u.queueTop) {
1144	+	if (tryReleaseWaiter())
1145	+	break;
1146	+	}
1147	+	}
1148	+	}
1149	+	}
1150	+	}
1151	+
1152	+	/**
1153	+	* Final callback from terminating worker.  Removes record of
1154	+	* worker from array, and adjusts counts. If pool is shutting
1155	+	* down, tries to complete termination.
1156	+	*
1157	+	* @param w the worker
1158	+	*/
1159	+	final void deregisterWorker(ForkJoinWorkerThread w, Throwable ex) {
1160	+	int idx = w.poolIndex;
1161	+	int sc = w.stealCount;
1162	+	int steps = 0;
1163	+	// Remove from array, adjust worker counts and collect steal count.
1164	+	// We can intermix failed removes or adjusts with steal updates
1165	+	do {
1166	+	long s, c;
1167	+	int g;
1168	+	if (steps == 0 && ((g = scanGuard) & SG_UNIT) == 0 &&
1169	+	UNSAFE.compareAndSwapInt(this, scanGuardOffset,
1170	+	g, g |= SG_UNIT)) {
1171	+	ForkJoinWorkerThread[] ws = workers;
1172	+	if (ws != null && idx >= 0 &&
1173	+	idx < ws.length && ws[idx] == w)
1174	+	ws[idx] = null;    // verify
1175	+	nextWorkerIndex = idx;
1176	+	scanGuard = g + SG_UNIT;
1177	+	steps = 1;
1178	+	}
1179	+	if (steps == 1 &&
1180	+	UNSAFE.compareAndSwapLong(this, ctlOffset, c = ctl,
1181	+	(((c - AC_UNIT) & AC_MASK) |
1182	+	((c - TC_UNIT) & TC_MASK) |
1183	+	(c & ~(AC_MASK|TC_MASK)))))
1184	+	steps = 2;
1185	+	if (sc != 0 &&
1186	+	UNSAFE.compareAndSwapLong(this, stealCountOffset,
1187	+	s = stealCount, s + sc))
1188	+	sc = 0;
1189	+	} while (steps != 2 || sc != 0);
1190	+	if (!tryTerminate(false)) {
1191	+	if (ex != null)   // possibly replace if died abnormally
1192	+	signalWork();
1193	+	else
1194	+	tryReleaseWaiter();
1195	+	}
1196	+	}
1197	+
1198	+	// Shutdown and termination
1199	+
1200		/**
1201		* Possibly initiates and/or completes termination.
1202		*
#	Line 1071 | Line 1205 | public class ForkJoinPool extends Abstra
1205		* @return true if now terminating or terminated
1206		*/
1207		private boolean tryTerminate(boolean now) {
1208	<	if (now)
1209	<	advanceRunLevel(SHUTDOWN); // ensure at least SHUTDOWN
1210	<	else if (runState < SHUTDOWN ||
1211	<	!submissionQueue.isEmpty() ||
1212	<	(runState & ACTIVE_COUNT_MASK) != 0)
1213	<	return false;
1214	<
1215	<	if (advanceRunLevel(TERMINATING))
1216	<	startTerminating();
1217	<
1218	<	// Finish now if all threads terminated; else in some subsequent call
1219	<	if ((workerCounts >>> TOTAL_COUNT_SHIFT) == 0) {
1220	<	advanceRunLevel(TERMINATED);
1221	<	termination.arrive();
1208	>	long c;
1209	>	while (((c = ctl) & STOP_BIT) == 0) {
1210	>	if (!now) {
1211	>	if ((int)(c >> AC_SHIFT) != -parallelism)
1212	>	return false;
1213	>	if (!shutdown || blockedCount != 0 || quiescerCount != 0 ||
1214	>	queueTop - queueBase > 0) {
1215	>	if (ctl == c) // staleness check
1216	>	return false;
1217	>	continue;
1218	>	}
1219	>	}
1220	>	if (UNSAFE.compareAndSwapLong(this, ctlOffset, c, c | STOP_BIT))
1221	>	startTerminating();
1222	>	}
1223	>	if ((short)(c >>> TC_SHIFT) == -parallelism) {
1224	>	submissionLock.lock();
1225	>	termination.signalAll();
1226	>	submissionLock.unlock();
1227		}
1228		return true;
1229		}
1230
1231		/**
1232	<	* Actions on transition to TERMINATING
1233	<	*
1234	<	* Runs up to four passes through workers: (0) shutting down each
1235	<	* (without waking up if parked) to quickly spread notifications
1236	<	* without unnecessary bouncing around event queues etc (1) wake
1237	<	* up and help cancel tasks (2) interrupt (3) mop up races with
1099	<	* interrupted workers
1232	>	* Runs up to three passes through workers: (0) Setting
1233	>	* termination status for each worker, followed by wakeups up
1234	>	* queued workers (1) helping cancel tasks (2) interrupting
1235	>	* lagging threads (likely in external tasks, but possibly also
1236	>	* blocked in joins).  Each pass repeats previous steps because of
1237	>	* potential lagging thread creation.
1238		*/
1239		private void startTerminating() {
1240		cancelSubmissions();
1241	<	for (int passes = 0; passes < 4 && workerCounts != 0; ++passes) {
1242	<	int c; // advance event count
1243	<	UNSAFE.compareAndSwapInt(this, eventCountOffset,
1244	<	c = eventCount, c+1);
1245	<	eventWaiters = 0L; // clobber lists
1246	<	spareWaiters = 0;
1247	<	for (ForkJoinWorkerThread w : workers) {
1248	<	if (w != null) {
1249	<	w.shutdown();
1250	<	if (passes > 0 && !w.isTerminated()) {
1251	<	w.cancelTasks();
1252	<	LockSupport.unpark(w);
1253	<	if (passes > 1) {
1116	<	try {
1117	<	w.interrupt();
1118	<	} catch (SecurityException ignore) {
1241	>	for (int pass = 0; pass < 3; ++pass) {
1242	>	ForkJoinWorkerThread[] ws = workers;
1243	>	if (ws != null) {
1244	>	for (ForkJoinWorkerThread w : ws) {
1245	>	if (w != null) {
1246	>	w.terminate = true;
1247	>	if (pass > 0) {
1248	>	w.cancelTasks();
1249	>	if (pass > 1 && !w.isInterrupted()) {
1250	>	try {
1251	>	w.interrupt();
1252	>	} catch (SecurityException ignore) {
1253	>	}
1254		}
1255		}
1256		}
1257		}
1258	+	terminateWaiters();
1259		}
1260		}
1261		}
1262
1263		/**
1264	<	* Clears out and cancels submissions, ignoring exceptions.
1264	>	* Polls and cancels all submissions. Called only during termination.
1265		*/
1266		private void cancelSubmissions() {
1267	<	ForkJoinTask<?> task;
1268	<	while ((task = submissionQueue.poll()) != null) {
1269	<	try {
1270	<	task.cancel(false);
1271	<	} catch (Throwable ignore) {
1267	>	while (queueBase != queueTop) {
1268	>	ForkJoinTask<?> task = pollSubmission();
1269	>	if (task != null) {
1270	>	try {
1271	>	task.cancel(false);
1272	>	} catch (Throwable ignore) {
1273	>	}
1274		}
1275		}
1276		}
1277
1278	<	// misc support for ForkJoinWorkerThread
1278	>	/**
1279	>	* Tries to set the termination status of waiting workers, and
1280	>	* then wake them up (after which they will terminate).
1281	>	*/
1282	>	private void terminateWaiters() {
1283	>	ForkJoinWorkerThread[] ws = workers;
1284	>	if (ws != null) {
1285	>	ForkJoinWorkerThread w; long c; int i, e;
1286	>	int n = ws.length;
1287	>	while ((i = ~(e = (int)(c = ctl)) & SMASK) < n &&
1288	>	(w = ws[i]) != null && w.eventCount == (e & E_MASK)) {
1289	>	if (UNSAFE.compareAndSwapLong(this, ctlOffset, c,
1290	>	(long)(w.nextWait & E_MASK) |
1291	>	((c + AC_UNIT) & AC_MASK) |
1292	>	(c & (TC_MASK|STOP_BIT)))) {
1293	>	w.terminate = true;
1294	>	w.eventCount = e + EC_UNIT;
1295	>	if (w.parked)
1296	>	UNSAFE.unpark(w);
1297	>	}
1298	>	}
1299	>	}
1300	>	}
1301	>
1302	>	// misc ForkJoinWorkerThread support
1303
1304		/**
1305	<	* Returns pool number.
1305	>	* Increment or decrement quiescerCount. Needed only to prevent
1306	>	* triggering shutdown if a worker is transiently inactive while
1307	>	* checking quiescence.
1308	>	*
1309	>	* @param delta 1 for increment, -1 for decrement
1310		*/
1311	<	final int getPoolNumber() {
1312	<	return poolNumber;
1311	>	final void addQuiescerCount(int delta) {
1312	>	int c;
1313	>	do {} while(!UNSAFE.compareAndSwapInt(this, quiescerCountOffset,
1314	>	c = quiescerCount, c + delta));
1315		}
1316
1317		/**
1318	<	* Tries to accumulate steal count from a worker, clearing
1319	<	* the worker's value if successful.
1318	>	* Directly increment or decrement active count without
1319	>	* queuing. This method is used to transiently assert inactivation
1320	>	* while checking quiescence.
1321		*
1322	<	* @return true if worker steal count now zero
1322	>	* @param delta 1 for increment, -1 for decrement
1323		*/
1324	<	final boolean tryAccumulateStealCount(ForkJoinWorkerThread w) {
1325	<	int sc = w.stealCount;
1326	<	long c = stealCount;
1327	<	// CAS even if zero, for fence effects
1328	<	if (UNSAFE.compareAndSwapLong(this, stealCountOffset, c, c + sc)) {
1329	<	if (sc != 0)
1161	<	w.stealCount = 0;
1162	<	return true;
1163	<	}
1164	<	return sc == 0;
1324	>	final void addActiveCount(int delta) {
1325	>	long d = delta < 0 ? -AC_UNIT : AC_UNIT;
1326	>	long c;
1327	>	do {} while (!UNSAFE.compareAndSwapLong(this, ctlOffset, c = ctl,
1328	>	((c + d) & AC_MASK) |
1329	>	(c & ~AC_MASK)));
1330		}
1331
1332		/**
#	Line 1169 | Line 1334 | public class ForkJoinPool extends Abstra
1334		* active thread.
1335		*/
1336		final int idlePerActive() {
1337	<	int pc = parallelism; // use parallelism, not rc
1338	<	int ac = runState;    // no mask -- artificially boosts during shutdown
1339	<	// Use exact results for small values, saturate past 4
1340	<	return ((pc <= ac) ? 0 :
1341	<	(pc >>> 1 <= ac) ? 1 :
1342	<	(pc >>> 2 <= ac) ? 3 :
1343	<	pc >>> 3);
1337	>	// Approximate at powers of two for small values, saturate past 4
1338	>	int p = parallelism;
1339	>	int a = p + (int)(ctl >> AC_SHIFT);
1340	>	return (a > (p >>>= 1) ? 0 :
1341	>	a > (p >>>= 1) ? 1 :
1342	>	a > (p >>>= 1) ? 2 :
1343	>	a > (p >>>= 1) ? 4 :
1344	>	8);
1345		}
1346
1347	<	// Public and protected methods
1347	>	// Exported methods
1348
1349		// Constructors
1350
#	Line 1247 | Line 1413 | public class ForkJoinPool extends Abstra
1413		checkPermission();
1414		if (factory == null)
1415		throw new NullPointerException();
1416	<	if (parallelism <= 0 || parallelism > MAX_WORKERS)
1416	>	if (parallelism <= 0 || parallelism > MAX_ID)
1417		throw new IllegalArgumentException();
1418		this.parallelism = parallelism;
1419		this.factory = factory;
1420		this.ueh = handler;
1421		this.locallyFifo = asyncMode;
1422	<	int arraySize = initialArraySizeFor(parallelism);
1423	<	this.workers = new ForkJoinWorkerThread[arraySize];
1424	<	this.submissionQueue = new LinkedTransferQueue<ForkJoinTask<?>>();
1425	<	this.workerLock = new ReentrantLock();
1426	<	this.termination = new Phaser(1);
1427	<	this.poolNumber = poolNumberGenerator.incrementAndGet();
1428	<	}
1429	<
1430	<	/**
1431	<	* Returns initial power of two size for workers array.
1432	<	* @param pc the initial parallelism level
1433	<	*/
1434	<	private static int initialArraySizeFor(int pc) {
1435	<	// If possible, initially allocate enough space for one spare
1436	<	int size = pc < MAX_WORKERS ? pc + 1 : MAX_WORKERS;
1437	<	// See Hackers Delight, sec 3.2. We know MAX_WORKERS < (1 >>> 16)
1438	<	size |= size >>> 1;
1273	<	size |= size >>> 2;
1274	<	size |= size >>> 4;
1275	<	size |= size >>> 8;
1276	<	return size + 1;
1422	>	long np = (long)(-parallelism); // offset ctl counts
1423	>	this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
1424	>	this.submissionQueue = new ForkJoinTask<?>[INITIAL_QUEUE_CAPACITY];
1425	>	// initialize workers array with room for 2*parallelism if possible
1426	>	int n = parallelism << 1;
1427	>	if (n >= MAX_ID)
1428	>	n = MAX_ID;
1429	>	else { // See Hackers Delight, sec 3.2, where n < (1 << 16)
1430	>	n |= n >>> 1; n |= n >>> 2; n |= n >>> 4; n |= n >>> 8;
1431	>	}
1432	>	workers = new ForkJoinWorkerThread[n + 1];
1433	>	this.submissionLock = new ReentrantLock();
1434	>	this.termination = submissionLock.newCondition();
1435	>	StringBuilder sb = new StringBuilder("ForkJoinPool-");
1436	>	sb.append(poolNumberGenerator.incrementAndGet());
1437	>	sb.append("-worker-");
1438	>	this.workerNamePrefix = sb.toString();
1439		}
1440
1441		// Execution methods
1442
1443		/**
1282	–	* Common code for execute, invoke and submit
1283	–	*/
1284	–	private <T> void doSubmit(ForkJoinTask<T> task) {
1285	–	if (task == null)
1286	–	throw new NullPointerException();
1287	–	if (runState >= SHUTDOWN)
1288	–	throw new RejectedExecutionException();
1289	–	submissionQueue.offer(task);
1290	–	int c; // try to increment event count -- CAS failure OK
1291	–	UNSAFE.compareAndSwapInt(this, eventCountOffset, c = eventCount, c+1);
1292	–	helpMaintainParallelism(); // create, start, or resume some workers
1293	–	}
1294	–
1295	–	/**
1444		* Performs the given task, returning its result upon completion.
1445	+	* If the computation encounters an unchecked Exception or Error,
1446	+	* it is rethrown as the outcome of this invocation.  Rethrown
1447	+	* exceptions behave in the same way as regular exceptions, but,
1448	+	* when possible, contain stack traces (as displayed for example
1449	+	* using {@code ex.printStackTrace()}) of both the current thread
1450	+	* as well as the thread actually encountering the exception;
1451	+	* minimally only the latter.
1452		*
1453		* @param task the task
1454		* @return the task's result
#	Line 1302 | Line 1457 | public class ForkJoinPool extends Abstra
1457		*         scheduled for execution
1458		*/
1459		public <T> T invoke(ForkJoinTask<T> task) {
1460	<	doSubmit(task);
1461	<	return task.join();
1460	>	Thread t = Thread.currentThread();
1461	>	if (task == null)
1462	>	throw new NullPointerException();
1463	>	if (shutdown)
1464	>	throw new RejectedExecutionException();
1465	>	if ((t instanceof ForkJoinWorkerThread) &&
1466	>	((ForkJoinWorkerThread)t).pool == this)
1467	>	return task.invoke();  // bypass submit if in same pool
1468	>	else {
1469	>	addSubmission(task);
1470	>	return task.join();
1471	>	}
1472	>	}
1473	>
1474	>	/**
1475	>	* Unless terminating, forks task if within an ongoing FJ
1476	>	* computation in the current pool, else submits as external task.
1477	>	*/
1478	>	private <T> void forkOrSubmit(ForkJoinTask<T> task) {
1479	>	ForkJoinWorkerThread w;
1480	>	Thread t = Thread.currentThread();
1481	>	if (shutdown)
1482	>	throw new RejectedExecutionException();
1483	>	if ((t instanceof ForkJoinWorkerThread) &&
1484	>	(w = (ForkJoinWorkerThread)t).pool == this)
1485	>	w.pushTask(task);
1486	>	else
1487	>	addSubmission(task);
1488		}
1489
1490		/**
#	Line 1315 | Line 1496 | public class ForkJoinPool extends Abstra
1496		*         scheduled for execution
1497		*/
1498		public void execute(ForkJoinTask<?> task) {
1499	<	doSubmit(task);
1499	>	if (task == null)
1500	>	throw new NullPointerException();
1501	>	forkOrSubmit(task);
1502		}
1503
1504		// AbstractExecutorService methods
#	Line 1326 | Line 1509 | public class ForkJoinPool extends Abstra
1509		*         scheduled for execution
1510		*/
1511		public void execute(Runnable task) {
1512	+	if (task == null)
1513	+	throw new NullPointerException();
1514		ForkJoinTask<?> job;
1515		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
1516		job = (ForkJoinTask<?>) task;
1517		else
1518		job = ForkJoinTask.adapt(task, null);
1519	<	doSubmit(job);
1519	>	forkOrSubmit(job);
1520		}
1521
1522		/**
#	Line 1344 | Line 1529 | public class ForkJoinPool extends Abstra
1529		*         scheduled for execution
1530		*/
1531		public <T> ForkJoinTask<T> submit(ForkJoinTask<T> task) {
1532	<	doSubmit(task);
1532	>	if (task == null)
1533	>	throw new NullPointerException();
1534	>	forkOrSubmit(task);
1535		return task;
1536		}
1537
#	Line 1354 | Line 1541 | public class ForkJoinPool extends Abstra
1541		*         scheduled for execution
1542		*/
1543		public <T> ForkJoinTask<T> submit(Callable<T> task) {
1544	+	if (task == null)
1545	+	throw new NullPointerException();
1546		ForkJoinTask<T> job = ForkJoinTask.adapt(task);
1547	<	doSubmit(job);
1547	>	forkOrSubmit(job);
1548		return job;
1549		}
1550
#	Line 1365 | Line 1554 | public class ForkJoinPool extends Abstra
1554		*         scheduled for execution
1555		*/
1556		public <T> ForkJoinTask<T> submit(Runnable task, T result) {
1557	+	if (task == null)
1558	+	throw new NullPointerException();
1559		ForkJoinTask<T> job = ForkJoinTask.adapt(task, result);
1560	<	doSubmit(job);
1560	>	forkOrSubmit(job);
1561		return job;
1562		}
1563
#	Line 1376 | Line 1567 | public class ForkJoinPool extends Abstra
1567		*         scheduled for execution
1568		*/
1569		public ForkJoinTask<?> submit(Runnable task) {
1570	+	if (task == null)
1571	+	throw new NullPointerException();
1572		ForkJoinTask<?> job;
1573		if (task instanceof ForkJoinTask<?>) // avoid re-wrap
1574		job = (ForkJoinTask<?>) task;
1575		else
1576		job = ForkJoinTask.adapt(task, null);
1577	<	doSubmit(job);
1577	>	forkOrSubmit(job);
1578		return job;
1579		}
1580
#	Line 1441 | Line 1634 | public class ForkJoinPool extends Abstra
1634
1635		/**
1636		* Returns the number of worker threads that have started but not
1637	<	* yet terminated.  This result returned by this method may differ
1637	>	* yet terminated.  The result returned by this method may differ
1638		* from {@link #getParallelism} when threads are created to
1639		* maintain parallelism when others are cooperatively blocked.
1640		*
1641		* @return the number of worker threads
1642		*/
1643		public int getPoolSize() {
1644	<	return workerCounts >>> TOTAL_COUNT_SHIFT;
1644	>	return parallelism + (short)(ctl >>> TC_SHIFT);
1645		}
1646
1647		/**
#	Line 1470 | Line 1663 | public class ForkJoinPool extends Abstra
1663		* @return the number of worker threads
1664		*/
1665		public int getRunningThreadCount() {
1666	<	return workerCounts & RUNNING_COUNT_MASK;
1666	>	int r = parallelism + (int)(ctl >> AC_SHIFT);
1667	>	return r <= 0? 0 : r; // suppress momentarily negative values
1668		}
1669
1670		/**
#	Line 1481 | Line 1675 | public class ForkJoinPool extends Abstra
1675		* @return the number of active threads
1676		*/
1677		public int getActiveThreadCount() {
1678	<	return runState & ACTIVE_COUNT_MASK;
1678	>	int r = parallelism + (int)(ctl >> AC_SHIFT) + blockedCount;
1679	>	return r <= 0? 0 : r; // suppress momentarily negative values
1680		}
1681
1682		/**
#	Line 1496 | Line 1691 | public class ForkJoinPool extends Abstra
1691		* @return {@code true} if all threads are currently idle
1692		*/
1693		public boolean isQuiescent() {
1694	<	return (runState & ACTIVE_COUNT_MASK) == 0;
1694	>	return parallelism + (int)(ctl >> AC_SHIFT) + blockedCount == 0;
1695		}
1696
1697		/**
#	Line 1526 | Line 1721 | public class ForkJoinPool extends Abstra
1721		*/
1722		public long getQueuedTaskCount() {
1723		long count = 0;
1724	<	for (ForkJoinWorkerThread w : workers)
1725	<	if (w != null)
1726	<	count += w.getQueueSize();
1724	>	ForkJoinWorkerThread[] ws;
1725	>	if ((short)(ctl >>> TC_SHIFT) > -parallelism &&
1726	>	(ws = workers) != null) {
1727	>	for (ForkJoinWorkerThread w : ws)
1728	>	if (w != null)
1729	>	count -= w.queueBase - w.queueTop; // must read base first
1730	>	}
1731		return count;
1732		}
1733
1734		/**
1735		* Returns an estimate of the number of tasks submitted to this
1736	<	* pool that have not yet begun executing.  This method takes time
1737	<	* proportional to the number of submissions.
1736	>	* pool that have not yet begun executing.  This method may take
1737	>	* time proportional to the number of submissions.
1738		*
1739		* @return the number of queued submissions
1740		*/
1741		public int getQueuedSubmissionCount() {
1742	<	return submissionQueue.size();
1742	>	return -queueBase + queueTop;
1743		}
1744
1745		/**
#	Line 1550 | Line 1749 | public class ForkJoinPool extends Abstra
1749		* @return {@code true} if there are any queued submissions
1750		*/
1751		public boolean hasQueuedSubmissions() {
1752	<	return !submissionQueue.isEmpty();
1752	>	return queueBase != queueTop;
1753		}
1754
1755		/**
#	Line 1561 | Line 1760 | public class ForkJoinPool extends Abstra
1760		* @return the next submission, or {@code null} if none
1761		*/
1762		protected ForkJoinTask<?> pollSubmission() {
1763	<	return submissionQueue.poll();
1763	>	ForkJoinTask<?> t; ForkJoinTask<?>[] q; int b, i;
1764	>	while ((b = queueBase) != queueTop &&
1765	>	(q = submissionQueue) != null &&
1766	>	(i = (q.length - 1) & b) >= 0) {
1767	>	long u = (i << ASHIFT) + ABASE;
1768	>	if ((t = q[i]) != null &&
1769	>	queueBase == b &&
1770	>	UNSAFE.compareAndSwapObject(q, u, t, null)) {
1771	>	queueBase = b + 1;
1772	>	return t;
1773	>	}
1774	>	}
1775	>	return null;
1776		}
1777
1778		/**
#	Line 1582 | Line 1793 | public class ForkJoinPool extends Abstra
1793		* @return the number of elements transferred
1794		*/
1795		protected int drainTasksTo(Collection<? super ForkJoinTask<?>> c) {
1796	<	int count = submissionQueue.drainTo(c);
1797	<	for (ForkJoinWorkerThread w : workers)
1798	<	if (w != null)
1799	<	count += w.drainTasksTo(c);
1796	>	int count = 0;
1797	>	while (queueBase != queueTop) {
1798	>	ForkJoinTask<?> t = pollSubmission();
1799	>	if (t != null) {
1800	>	c.add(t);
1801	>	++count;
1802	>	}
1803	>	}
1804	>	ForkJoinWorkerThread[] ws;
1805	>	if ((short)(ctl >>> TC_SHIFT) > -parallelism &&
1806	>	(ws = workers) != null) {
1807	>	for (ForkJoinWorkerThread w : ws)
1808	>	if (w != null)
1809	>	count += w.drainTasksTo(c);
1810	>	}
1811		return count;
1812		}
1813
#	Line 1600 | Line 1822 | public class ForkJoinPool extends Abstra
1822		long st = getStealCount();
1823		long qt = getQueuedTaskCount();
1824		long qs = getQueuedSubmissionCount();
1603	–	int wc = workerCounts;
1604	–	int tc = wc >>> TOTAL_COUNT_SHIFT;
1605	–	int rc = wc & RUNNING_COUNT_MASK;
1825		int pc = parallelism;
1826	<	int rs = runState;
1827	<	int ac = rs & ACTIVE_COUNT_MASK;
1826	>	long c = ctl;
1827	>	int tc = pc + (short)(c >>> TC_SHIFT);
1828	>	int rc = pc + (int)(c >> AC_SHIFT);
1829	>	if (rc < 0) // ignore transient negative
1830	>	rc = 0;
1831	>	int ac = rc + blockedCount;
1832	>	String level;
1833	>	if ((c & STOP_BIT) != 0)
1834	>	level = (tc == 0)? "Terminated" : "Terminating";
1835	>	else
1836	>	level = shutdown? "Shutting down" : "Running";
1837		return super.toString() +
1838	<	"[" + runLevelToString(rs) +
1838	>	"[" + level +
1839		", parallelism = " + pc +
1840		", size = " + tc +
1841		", active = " + ac +
#	Line 1618 | Line 1846 | public class ForkJoinPool extends Abstra
1846		"]";
1847		}
1848
1621	–	private static String runLevelToString(int s) {
1622	–	return ((s & TERMINATED) != 0 ? "Terminated" :
1623	–	((s & TERMINATING) != 0 ? "Terminating" :
1624	–	((s & SHUTDOWN) != 0 ? "Shutting down" :
1625	–	"Running")));
1626	–	}
1627	–
1849		/**
1850		* Initiates an orderly shutdown in which previously submitted
1851		* tasks are executed, but no new tasks will be accepted.
#	Line 1639 | Line 1860 | public class ForkJoinPool extends Abstra
1860		*/
1861		public void shutdown() {
1862		checkPermission();
1863	<	advanceRunLevel(SHUTDOWN);
1863	>	shutdown = true;
1864		tryTerminate(false);
1865		}
1866
#	Line 1661 | Line 1882 | public class ForkJoinPool extends Abstra
1882		*/
1883		public List<Runnable> shutdownNow() {
1884		checkPermission();
1885	+	shutdown = true;
1886		tryTerminate(true);
1887		return Collections.emptyList();
1888		}
#	Line 1671 | Line 1893 | public class ForkJoinPool extends Abstra
1893		* @return {@code true} if all tasks have completed following shut down
1894		*/
1895		public boolean isTerminated() {
1896	<	return runState >= TERMINATED;
1896	>	long c = ctl;
1897	>	return ((c & STOP_BIT) != 0L &&
1898	>	(short)(c >>> TC_SHIFT) == -parallelism);
1899		}
1900
1901		/**
#	Line 1679 | Line 1903 | public class ForkJoinPool extends Abstra
1903		* commenced but not yet completed.  This method may be useful for
1904		* debugging. A return of {@code true} reported a sufficient
1905		* period after shutdown may indicate that submitted tasks have
1906	<	* ignored or suppressed interruption, causing this executor not
1907	<	* to properly terminate.
1906	>	* ignored or suppressed interruption, or are waiting for IO,
1907	>	* causing this executor not to properly terminate. (See the
1908	>	* advisory notes for class {@link ForkJoinTask} stating that
1909	>	* tasks should not normally entail blocking operations.  But if
1910	>	* they do, they must abort them on interrupt.)
1911		*
1912		* @return {@code true} if terminating but not yet terminated
1913		*/
1914		public boolean isTerminating() {
1915	<	return (runState & (TERMINATING|TERMINATED)) == TERMINATING;
1915	>	long c = ctl;
1916	>	return ((c & STOP_BIT) != 0L &&
1917	>	(short)(c >>> TC_SHIFT) != -parallelism);
1918	>	}
1919	>
1920	>	/**
1921	>	* Returns true if terminating or terminated. Used by ForkJoinWorkerThread.
1922	>	*/
1923	>	final boolean isAtLeastTerminating() {
1924	>	return (ctl & STOP_BIT) != 0L;
1925		}
1926
1927		/**
#	Line 1694 | Line 1930 | public class ForkJoinPool extends Abstra
1930		* @return {@code true} if this pool has been shut down
1931		*/
1932		public boolean isShutdown() {
1933	<	return runState >= SHUTDOWN;
1933	>	return shutdown;
1934		}
1935
1936		/**
#	Line 1710 | Line 1946 | public class ForkJoinPool extends Abstra
1946		*/
1947		public boolean awaitTermination(long timeout, TimeUnit unit)
1948		throws InterruptedException {
1949	+	long nanos = unit.toNanos(timeout);
1950	+	final ReentrantLock lock = this.submissionLock;
1951	+	lock.lock();
1952		try {
1953	<	return termination.awaitAdvanceInterruptibly(0, timeout, unit) > 0;
1954	<	} catch (TimeoutException ex) {
1955	<	return false;
1953	>	for (;;) {
1954	>	if (isTerminated())
1955	>	return true;
1956	>	if (nanos <= 0)
1957	>	return false;
1958	>	nanos = termination.awaitNanos(nanos);
1959	>	}
1960	>	} finally {
1961	>	lock.unlock();
1962		}
1963		}
1964
#	Line 1725 | Line 1970 | public class ForkJoinPool extends Abstra
1970		* {@code isReleasable} must return {@code true} if blocking is
1971		* not necessary. Method {@code block} blocks the current thread
1972		* if necessary (perhaps internally invoking {@code isReleasable}
1973	<	* before actually blocking). The unusual methods in this API
1974	<	* accommodate synchronizers that may, but don't usually, block
1975	<	* for long periods. Similarly, they allow more efficient internal
1976	<	* handling of cases in which additional workers may be, but
1977	<	* usually are not, needed to ensure sufficient parallelism.
1978	<	* Toward this end, implementations of method {@code isReleasable}
1979	<	* must be amenable to repeated invocation.
1973	>	* before actually blocking). These actions are performed by any
1974	>	* thread invoking {@link ForkJoinPool#managedBlock}.  The
1975	>	* unusual methods in this API accommodate synchronizers that may,
1976	>	* but don't usually, block for long periods. Similarly, they
1977	>	* allow more efficient internal handling of cases in which
1978	>	* additional workers may be, but usually are not, needed to
1979	>	* ensure sufficient parallelism.  Toward this end,
1980	>	* implementations of method {@code isReleasable} must be amenable
1981	>	* to repeated invocation.
1982		*
1983		* <p>For example, here is a ManagedBlocker based on a
1984		* ReentrantLock:
#	Line 1833 | Line 2080 | public class ForkJoinPool extends Abstra
2080		}
2081
2082		// Unsafe mechanics
2083	<
2084	<	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
2085	<	private static final long workerCountsOffset =
2086	<	objectFieldOffset("workerCounts", ForkJoinPool.class);
2087	<	private static final long runStateOffset =
2088	<	objectFieldOffset("runState", ForkJoinPool.class);
2089	<	private static final long eventCountOffset =
2090	<	objectFieldOffset("eventCount", ForkJoinPool.class);
2091	<	private static final long eventWaitersOffset =
2092	<	objectFieldOffset("eventWaiters",ForkJoinPool.class);
2093	<	private static final long stealCountOffset =
2094	<	objectFieldOffset("stealCount",ForkJoinPool.class);
2095	<	private static final long spareWaitersOffset =
2096	<	objectFieldOffset("spareWaiters",ForkJoinPool.class);
2097	<
2098	<	private static long objectFieldOffset(String field, Class<?> klazz) {
2083	>	private static final sun.misc.Unsafe UNSAFE;
2084	>	private static final long ctlOffset;
2085	>	private static final long stealCountOffset;
2086	>	private static final long blockedCountOffset;
2087	>	private static final long quiescerCountOffset;
2088	>	private static final long scanGuardOffset;
2089	>	private static final long nextWorkerNumberOffset;
2090	>	private static final long ABASE;
2091	>	private static final int ASHIFT;
2092	>
2093	>	static {
2094	>	poolNumberGenerator = new AtomicInteger();
2095	>	workerSeedGenerator = new Random();
2096	>	modifyThreadPermission = new RuntimePermission("modifyThread");
2097	>	defaultForkJoinWorkerThreadFactory =
2098	>	new DefaultForkJoinWorkerThreadFactory();
2099	>	int s;
2100		try {
2101	<	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
2102	<	} catch (NoSuchFieldException e) {
2103	<	// Convert Exception to corresponding Error
2104	<	NoSuchFieldError error = new NoSuchFieldError(field);
2105	<	error.initCause(e);
2106	<	throw error;
2107	<	}
2101	>	UNSAFE = getUnsafe();
2102	>	Class k = ForkJoinPool.class;
2103	>	ctlOffset = UNSAFE.objectFieldOffset
2104	>	(k.getDeclaredField("ctl"));
2105	>	stealCountOffset = UNSAFE.objectFieldOffset
2106	>	(k.getDeclaredField("stealCount"));
2107	>	blockedCountOffset = UNSAFE.objectFieldOffset
2108	>	(k.getDeclaredField("blockedCount"));
2109	>	quiescerCountOffset = UNSAFE.objectFieldOffset
2110	>	(k.getDeclaredField("quiescerCount"));
2111	>	scanGuardOffset = UNSAFE.objectFieldOffset
2112	>	(k.getDeclaredField("scanGuard"));
2113	>	nextWorkerNumberOffset = UNSAFE.objectFieldOffset
2114	>	(k.getDeclaredField("nextWorkerNumber"));
2115	>	Class a = ForkJoinTask[].class;
2116	>	ABASE = UNSAFE.arrayBaseOffset(a);
2117	>	s = UNSAFE.arrayIndexScale(a);
2118	>	} catch (Exception e) {
2119	>	throw new Error(e);
2120	>	}
2121	>	if ((s & (s-1)) != 0)
2122	>	throw new Error("data type scale not a power of two");
2123	>	ASHIFT = 31 - Integer.numberOfLeadingZeros(s);
2124		}
2125
2126		/**

Diff Legend

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