ViewVC Help

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

root/jsr166/jsr166/src/jsr166y/Phaser.java

Comparing jsr166/src/jsr166y/Phaser.java (file contents):
Revision 1.55 by dl, Mon Nov 15 12:51:54 2010 UTC vs.
Revision 1.73 by jsr166, Wed May 25 16:08:03 2011 UTC

#	Line 1 | Line 1
1		/*
2		* Written by Doug Lea with assistance from members of JCP JSR-166
3		* Expert Group and released to the public domain, as explained at
4	<	* http://creativecommons.org/licenses/publicdomain
4	>	* http://creativecommons.org/publicdomain/zero/1.0/
5		*/
6
7		package jsr166y;
#	Line 34 | Line 34 | import java.util.concurrent.locks.LockSu
34		* Phaser} may be repeatedly awaited.  Method {@link
35		* #arriveAndAwaitAdvance} has effect analogous to {@link
36		* java.util.concurrent.CyclicBarrier#await CyclicBarrier.await}. Each
37	<	* generation of a {@code Phaser} has an associated phase number. The
38	<	* phase number starts at zero, and advances when all parties arrive
39	<	* at the barrier, wrapping around to zero after reaching {@code
37	>	* generation of a phaser has an associated phase number. The phase
38	>	* number starts at zero, and advances when all parties arrive at the
39	>	* phaser, wrapping around to zero after reaching {@code
40		* Integer.MAX_VALUE}. The use of phase numbers enables independent
41	<	* control of actions upon arrival at a barrier and upon awaiting
41	>	* control of actions upon arrival at a phaser and upon awaiting
42		* others, via two kinds of methods that may be invoked by any
43		* registered party:
44		*
45		* <ul>
46		*
47		*   <li> <b>Arrival.</b> Methods {@link #arrive} and
48	<	*       {@link #arriveAndDeregister} record arrival at a
49	<	*       barrier. These methods do not block, but return an associated
50	<	*       <em>arrival phase number</em>; that is, the phase number of
51	<	*       the barrier to which the arrival applied. When the final
52	<	*       party for a given phase arrives, an optional barrier action
53	<	*       is performed and the phase advances.  Barrier actions,
54	<	*       performed by the party triggering a phase advance, are
55	<	*       arranged by overriding method {@link #onAdvance(int, int)},
56	<	*       which also controls termination. Overriding this method is
57	<	*       similar to, but more flexible than, providing a barrier
58	<	*       action to a {@code CyclicBarrier}.
48	>	*       {@link #arriveAndDeregister} record arrival.  These methods
49	>	*       do not block, but return an associated <em>arrival phase
50	>	*       number</em>; that is, the phase number of the phaser to which
51	>	*       the arrival applied. When the final party for a given phase
52	>	*       arrives, an optional action is performed and the phase
53	>	*       advances.  These actions are performed by the party
54	>	*       triggering a phase advance, and are arranged by overriding
55	>	*       method {@link #onAdvance(int, int)}, which also controls
56	>	*       termination. Overriding this method is similar to, but more
57	>	*       flexible than, providing a barrier action to a {@code
58	>	*       CyclicBarrier}.
59		*
60		*   <li> <b>Waiting.</b> Method {@link #awaitAdvance} requires an
61		*       argument indicating an arrival phase number, and returns when
62	<	*       the barrier advances to (or is already at) a different phase.
62	>	*       the phaser advances to (or is already at) a different phase.
63		*       Unlike similar constructions using {@code CyclicBarrier},
64		*       method {@code awaitAdvance} continues to wait even if the
65		*       waiting thread is interrupted. Interruptible and timeout
66		*       versions are also available, but exceptions encountered while
67		*       tasks wait interruptibly or with timeout do not change the
68	<	*       state of the barrier. If necessary, you can perform any
68	>	*       state of the phaser. If necessary, you can perform any
69		*       associated recovery within handlers of those exceptions,
70		*       often after invoking {@code forceTermination}.  Phasers may
71		*       also be used by tasks executing in a {@link ForkJoinPool},
#	Line 74 | Line 74 | import java.util.concurrent.locks.LockSu
74		*
75		* </ul>
76		*
77	<	* <p> <b>Termination.</b> A {@code Phaser} may enter a
78	<	* <em>termination</em> state in which all synchronization methods
79	<	* immediately return without updating phaser state or waiting for
80	<	* advance, and indicating (via a negative phase value) that execution
81	<	* is complete.  Termination is triggered when an invocation of {@code
82	<	* onAdvance} returns {@code true}.  As illustrated below, when
83	<	* phasers control actions with a fixed number of iterations, it is
84	<	* often convenient to override this method to cause termination when
85	<	* the current phase number reaches a threshold. Method {@link
86	<	* #forceTermination} is also available to abruptly release waiting
87	<	* threads and allow them to terminate.
88	<	*
89	<	* <p> <b>Tiering.</b> Phasers may be <em>tiered</em> (i.e., arranged
90	<	* in tree structures) to reduce contention. Phasers with large
91	<	* numbers of parties that would otherwise experience heavy
77	>	* <p> <b>Termination.</b> A phaser may enter a <em>termination</em>
78	>	* state, that may be checked using method {@link #isTerminated}. Upon
79	>	* termination, all synchronization methods immediately return without
80	>	* waiting for advance, as indicated by a negative return value.
81	>	* Similarly, attempts to register upon termination have no effect.
82	>	* Termination is triggered when an invocation of {@code onAdvance}
83	>	* returns {@code true}. The default implementation returns {@code
84	>	* true} if a deregistration has caused the number of registered
85	>	* parties to become zero.  As illustrated below, when phasers control
86	>	* actions with a fixed number of iterations, it is often convenient
87	>	* to override this method to cause termination when the current phase
88	>	* number reaches a threshold. Method {@link #forceTermination} is
89	>	* also available to abruptly release waiting threads and allow them
90	>	* to terminate.
91	>	*
92	>	* <p> <b>Tiering.</b> Phasers may be <em>tiered</em> (i.e.,
93	>	* constructed in tree structures) to reduce contention. Phasers with
94	>	* large numbers of parties that would otherwise experience heavy
95		* synchronization contention costs may instead be set up so that
96		* groups of sub-phasers share a common parent.  This may greatly
97		* increase throughput even though it incurs greater per-operation
98		* overhead.
99		*
100	+	* <p>In a tree of tiered phasers, registration and deregistration of
101	+	* child phasers with their parent are managed automatically.
102	+	* Whenever the number of registered parties of a child phaser becomes
103	+	* non-zero (as established in the {@link #Phaser(Phaser,int)}
104	+	* constructor, {@link #register}, or {@link #bulkRegister}), the
105	+	* child phaser is registered with its parent.  Whenever the number of
106	+	* registered parties becomes zero as the result of an invocation of
107	+	* {@link #arriveAndDeregister}, the child phaser is deregistered
108	+	* from its parent.
109	+	*
110		* <p><b>Monitoring.</b> While synchronization methods may be invoked
111		* only by registered parties, the current state of a phaser may be
112		* monitored by any caller.  At any given moment there are {@link
#	Line 117 | Line 130 | import java.util.concurrent.locks.LockSu
130		* void runTasks(List<Runnable> tasks) {
131		*   final Phaser phaser = new Phaser(1); // "1" to register self
132		*   // create and start threads
133	<	*   for (Runnable task : tasks) {
133	>	*   for (final Runnable task : tasks) {
134		*     phaser.register();
135		*     new Thread() {
136		*       public void run() {
#	Line 181 | Line 194 | import java.util.concurrent.locks.LockSu
194		* }}</pre>
195		*
196		*
197	<	* <p>To create a set of tasks using a tree of phasers,
198	<	* you could use code of the following form, assuming a
199	<	* Task class with a constructor accepting a phaser that
200	<	* it registers with upon construction:
197	>	* <p>To create a set of {@code n} tasks using a tree of phasers, you
198	>	* could use code of the following form, assuming a Task class with a
199	>	* constructor accepting a {@code Phaser} that it registers with upon
200	>	* construction. After invocation of {@code build(new Task[n], 0, n,
201	>	* new Phaser())}, these tasks could then be started, for example by
202	>	* submitting to a pool:
203		*
204		*  <pre> {@code
205	<	* void build(Task[] actions, int lo, int hi, Phaser ph) {
205	>	* void build(Task[] tasks, int lo, int hi, Phaser ph) {
206		*   if (hi - lo > TASKS_PER_PHASER) {
207		*     for (int i = lo; i < hi; i += TASKS_PER_PHASER) {
208		*       int j = Math.min(i + TASKS_PER_PHASER, hi);
209	<	*       build(actions, i, j, new Phaser(ph));
209	>	*       build(tasks, i, j, new Phaser(ph));
210		*     }
211		*   } else {
212		*     for (int i = lo; i < hi; ++i)
213	<	*       actions[i] = new Task(ph);
213	>	*       tasks[i] = new Task(ph);
214		*       // assumes new Task(ph) performs ph.register()
215		*   }
216	<	* }
202	<	* // .. initially called, for n tasks via
203	<	* build(new Task[n], 0, n, new Phaser());}</pre>
216	>	* }}</pre>
217		*
218		* The best value of {@code TASKS_PER_PHASER} depends mainly on
219	<	* expected barrier synchronization rates. A value as low as four may
220	<	* be appropriate for extremely small per-barrier task bodies (thus
219	>	* expected synchronization rates. A value as low as four may
220	>	* be appropriate for extremely small per-phase task bodies (thus
221		* high rates), or up to hundreds for extremely large ones.
222		*
223		* <p><b>Implementation notes</b>: This implementation restricts the
#	Line 224 | Line 237 | public class Phaser {
237		*/
238
239		/**
240	<	* Barrier state representation. Conceptually, a barrier contains
228	<	* four values:
240	>	* Primary state representation, holding four bit-fields:
241		*
242	<	* * unarrived -- the number of parties yet to hit barrier (bits  0-15)
243	<	* * parties -- the number of parties to wait              (bits 16-31)
244	<	* * phase -- the generation of the barrier                (bits 32-62)
245	<	* * terminated -- set if barrier is terminated            (bit  63 / sign)
246	<	*
247	<	* However, to efficiently maintain atomicity, these values are
248	<	* packed into a single (atomic) long. Termination uses the sign
249	<	* bit of 32 bit representation of phase, so phase is set to -1 on
250	<	* termination. Good performance relies on keeping state decoding
251	<	* and encoding simple, and keeping race windows short.
242	>	* unarrived  -- the number of parties yet to hit barrier (bits  0-15)
243	>	* parties    -- the number of parties to wait            (bits 16-31)
244	>	* phase      -- the generation of the barrier            (bits 32-62)
245	>	* terminated -- set if barrier is terminated             (bit  63 / sign)
246	>	*
247	>	* Except that a phaser with no registered parties is
248	>	* distinguished by the otherwise illegal state of having zero
249	>	* parties and one unarrived parties (encoded as EMPTY below).
250	>	*
251	>	* To efficiently maintain atomicity, these values are packed into
252	>	* a single (atomic) long. Good performance relies on keeping
253	>	* state decoding and encoding simple, and keeping race windows
254	>	* short.
255	>	*
256	>	* All state updates are performed via CAS except initial
257	>	* registration of a sub-phaser (i.e., one with a non-null
258	>	* parent).  In this (relatively rare) case, we use built-in
259	>	* synchronization to lock while first registering with its
260	>	* parent.
261	>	*
262	>	* The phase of a subphaser is allowed to lag that of its
263	>	* ancestors until it is actually accessed -- see method
264	>	* reconcileState.
265		*/
266		private volatile long state;
267
268	<	private static final int  MAX_PARTIES    = 0xffff;
269	<	private static final int  MAX_PHASE      = 0x7fffffff;
270	<	private static final int  PARTIES_SHIFT  = 16;
271	<	private static final int  PHASE_SHIFT    = 32;
272	<	private static final int  UNARRIVED_MASK = 0xffff;
273	<	private static final int  PARTIES_MASK   = 0xffff0000;
274	<	private static final long LPARTIES_MASK  = 0xffff0000L; // long version
275	<	private static final long ONE_ARRIVAL    = 1L;
276	<	private static final long ONE_PARTY      = 1L << PARTIES_SHIFT;
277	<	private static final long TERMINATION_PHASE  = -1L << PHASE_SHIFT;
268	>	private static final int  MAX_PARTIES     = 0xffff;
269	>	private static final int  MAX_PHASE       = Integer.MAX_VALUE;
270	>	private static final int  PARTIES_SHIFT   = 16;
271	>	private static final int  PHASE_SHIFT     = 32;
272	>	private static final int  UNARRIVED_MASK  = 0xffff;      // to mask ints
273	>	private static final long PARTIES_MASK    = 0xffff0000L; // to mask longs
274	>	private static final long TERMINATION_BIT = 1L << 63;
275	>
276	>	// some special values
277	>	private static final int  ONE_ARRIVAL     = 1;
278	>	private static final int  ONE_PARTY       = 1 << PARTIES_SHIFT;
279	>	private static final int  EMPTY           = 1;
280
281		// The following unpacking methods are usually manually inlined
282
283		private static int unarrivedOf(long s) {
284	<	return ((int) s) & UNARRIVED_MASK;
284	>	int counts = (int)s;
285	>	return (counts == EMPTY) ? 0 : counts & UNARRIVED_MASK;
286		}
287
288		private static int partiesOf(long s) {
289	<	return (((int) s) & PARTIES_MASK) >>> PARTIES_SHIFT;
289	>	return (int)s >>> PARTIES_SHIFT;
290		}
291
292		private static int phaseOf(long s) {
293	<	return (int) (s >>> PHASE_SHIFT);
293	>	return (int)(s >>> PHASE_SHIFT);
294		}
295
296		private static int arrivedOf(long s) {
297	<	return partiesOf(s) - unarrivedOf(s);
297	>	int counts = (int)s;
298	>	return (counts == EMPTY) ? 0 :
299	>	(counts >>> PARTIES_SHIFT) - (counts & UNARRIVED_MASK);
300		}
301
302		/**
#	Line 275 | Line 305 | public class Phaser {
305		private final Phaser parent;
306
307		/**
308	<	* The root of phaser tree. Equals this if not in a tree.  Used to
279	<	* support faster state push-down.
308	>	* The root of phaser tree. Equals this if not in a tree.
309		*/
310		private final Phaser root;
311
#	Line 294 | Line 323 | public class Phaser {
323		}
324
325		/**
326	+	* Returns message string for bounds exceptions on arrival.
327	+	*/
328	+	private String badArrive(long s) {
329	+	return "Attempted arrival of unregistered party for " +
330	+	stateToString(s);
331	+	}
332	+
333	+	/**
334	+	* Returns message string for bounds exceptions on registration.
335	+	*/
336	+	private String badRegister(long s) {
337	+	return "Attempt to register more than " +
338	+	MAX_PARTIES + " parties for " + stateToString(s);
339	+	}
340	+
341	+	/**
342		* Main implementation for methods arrive and arriveAndDeregister.
343		* Manually tuned to speed up and minimize race windows for the
344		* common case of just decrementing unarrived field.
345		*
346	<	* @param adj - adjustment to apply to state -- either
302	<	* ONE_ARRIVAL (for arrive) or
303	<	* ONE_ARRIVAL|ONE_PARTY (for arriveAndDeregister)
346	>	* @param deregister false for arrive, true for arriveAndDeregister
347		*/
348	<	private int doArrive(long adj) {
348	>	private int doArrive(boolean deregister) {
349	>	int adj = deregister ? ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL;
350	>	final Phaser root = this.root;
351		for (;;) {
352	<	long s;
353	<	int phase, unarrived;
354	<	if ((phase = (int)((s = state) >>> PHASE_SHIFT)) < 0)
352	>	long s = (root == this) ? state : reconcileState();
353	>	int phase = (int)(s >>> PHASE_SHIFT);
354	>	int counts = (int)s;
355	>	int unarrived = (counts & UNARRIVED_MASK) - 1;
356	>	if (phase < 0)
357		return phase;
358	<	else if ((unarrived = ((int)s) & UNARRIVED_MASK) == 0)
359	<	checkBadArrive(s);
358	>	else if (counts == EMPTY || unarrived < 0) {
359	>	if (root == this || reconcileState() == s)
360	>	throw new IllegalStateException(badArrive(s));
361	>	}
362		else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adj)) {
363	<	if (unarrived == 1) {
364	<	Phaser par;
365	<	long p = s & LPARTIES_MASK; // unshifted parties field
366	<	long lu = p >>> PARTIES_SHIFT;
367	<	int u = (int)lu;
368	<	int nextPhase = (phase + 1) & MAX_PHASE;
369	<	long next = ((long)nextPhase << PHASE_SHIFT) | p | lu;
370	<	if ((par = parent) == null) {
371	<	if (onAdvance(phase, u))
372	<	next |= TERMINATION_PHASE; // obliterate phase
373	<	UNSAFE.compareAndSwapLong(this, stateOffset, s, next);
374	<	releaseWaiters(phase);
375	<	}
376	<	else {
328	<	par.doArrive(u == 0?
329	<	ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL);
330	<	if ((int)(par.state >>> PHASE_SHIFT) != nextPhase ||
331	<	((int)(state >>> PHASE_SHIFT) != nextPhase &&
332	<	!UNSAFE.compareAndSwapLong(this, stateOffset,
333	<	s, next)))
334	<	reconcileState();
335	<	}
363	>	if (unarrived == 0) {
364	>	long n = s & PARTIES_MASK;  // base of next state
365	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
366	>	if (root != this)
367	>	return parent.doArrive(nextUnarrived == 0);
368	>	if (onAdvance(phase, nextUnarrived))
369	>	n |= TERMINATION_BIT;
370	>	else if (nextUnarrived == 0)
371	>	n |= EMPTY;
372	>	else
373	>	n |= nextUnarrived;
374	>	n |= (long)((phase + 1) & MAX_PHASE) << PHASE_SHIFT;
375	>	UNSAFE.compareAndSwapLong(this, stateOffset, s, n);
376	>	releaseWaiters(phase);
377		}
378		return phase;
379		}
#	Line 340 | Line 381 | public class Phaser {
381		}
382
383		/**
343	–	* Rechecks state and throws bounds exceptions on arrival -- called
344	–	* only if unarrived is apparently zero.
345	–	*/
346	–	private void checkBadArrive(long s) {
347	–	if (reconcileState() == s)
348	–	throw new IllegalStateException
349	–	("Attempted arrival of unregistered party for " +
350	–	stateToString(s));
351	–	}
352	–
353	–	/**
384		* Implementation of register, bulkRegister
385		*
386	<	* @param registrations number to add to both parties and unarrived fields
386	>	* @param registrations number to add to both parties and
387	>	* unarrived fields. Must be greater than zero.
388		*/
389		private int doRegister(int registrations) {
390	<	long adj = (long)registrations; // adjustment to state
391	<	adj |= adj << PARTIES_SHIFT;
392	<	Phaser par = parent;
390	>	// adjustment to state
391	>	long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
392	>	final Phaser parent = this.parent;
393	>	int phase;
394		for (;;) {
395	<	int phase, parties;
396	<	long s = par == null? state : reconcileState();
397	<	if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
398	<	return phase;
399	<	if ((parties = (((int)s) & PARTIES_MASK) >>> PARTIES_SHIFT) != 0 &&
368	<	(((int)s) & UNARRIVED_MASK) == 0)
369	<	internalAwaitAdvance(phase, null); // wait for onAdvance
370	<	else if (parties + registrations > MAX_PARTIES)
395	>	long s = state;
396	>	int counts = (int)s;
397	>	int parties = counts >>> PARTIES_SHIFT;
398	>	int unarrived = counts & UNARRIVED_MASK;
399	>	if (registrations > MAX_PARTIES - parties)
400		throw new IllegalStateException(badRegister(s));
401	<	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s + adj))
402	<	return phase;
401	>	else if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
402	>	break;
403	>	else if (counts != EMPTY) {             // not 1st registration
404	>	if (parent == null || reconcileState() == s) {
405	>	if (unarrived == 0)             // wait out advance
406	>	root.internalAwaitAdvance(phase, null);
407	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset,
408	>	s, s + adj))
409	>	break;
410	>	}
411	>	}
412	>	else if (parent == null) {              // 1st root registration
413	>	long next = ((long)phase << PHASE_SHIFT) | adj;
414	>	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
415	>	break;
416	>	}
417	>	else {
418	>	synchronized (this) {               // 1st sub registration
419	>	if (state == s) {               // recheck under lock
420	>	parent.doRegister(1);
421	>	do {                        // force current phase
422	>	phase = (int)(root.state >>> PHASE_SHIFT);
423	>	// assert phase < 0 || (int)state == EMPTY;
424	>	} while (!UNSAFE.compareAndSwapLong
425	>	(this, stateOffset, state,
426	>	((long)phase << PHASE_SHIFT) | adj));
427	>	break;
428	>	}
429	>	}
430	>	}
431		}
432	+	return phase;
433		}
434
435		/**
436	<	* Returns message string for out of bounds exceptions on registration.
437	<	*/
438	<	private String badRegister(long s) {
439	<	return "Attempt to register more than " +
440	<	MAX_PARTIES + " parties for " + stateToString(s);
441	<	}
442	<
443	<	/**
444	<	* Recursively resolves lagged phase propagation from root if necessary.
436	>	* Resolves lagged phase propagation from root if necessary.
437	>	* Reconciliation normally occurs when root has advanced but
438	>	* subphasers have not yet done so, in which case they must finish
439	>	* their own advance by setting unarrived to parties (or if
440	>	* parties is zero, resetting to unregistered EMPTY state).
441	>	* However, this method may also be called when "floating"
442	>	* subphasers with possibly some unarrived parties are merely
443	>	* catching up to current phase, in which case counts are
444	>	* unaffected.
445	>	*
446	>	* @return reconciled state
447		*/
448		private long reconcileState() {
449	<	Phaser par = parent;
450	<	if (par == null)
451	<	return state;
452	<	Phaser rt = root;
453	<	for (;;) {
454	<	long s, u;
455	<	int phase, rPhase, pPhase;
456	<	if ((phase = (int)((s = state)>>> PHASE_SHIFT)) < 0 ||
457	<	(rPhase = (int)(rt.state >>> PHASE_SHIFT)) == phase)
458	<	return s;
459	<	long pState = par.parent == null? par.state : par.reconcileState();
460	<	if (state == s) {
461	<	if ((rPhase < 0 || (((int)s) & UNARRIVED_MASK) == 0) &&
462	<	((pPhase = (int)(pState >>> PHASE_SHIFT)) < 0 ||
403	<	pPhase == ((phase + 1) & MAX_PHASE)))
404	<	UNSAFE.compareAndSwapLong
405	<	(this, stateOffset, s,
406	<	(((long) pPhase) << PHASE_SHIFT) |
407	<	(u = s & LPARTIES_MASK) |
408	<	(u >>> PARTIES_SHIFT)); // reset unarrived to parties
409	<	else
410	<	releaseWaiters(phase); // help release others
411	<	}
449	>	final Phaser root = this.root;
450	>	long s = state;
451	>	if (root != this) {
452	>	int phase, u, p;
453	>	// CAS root phase with current parties; possibly trip unarrived
454	>	while ((phase = (int)(root.state >>> PHASE_SHIFT)) !=
455	>	(int)(s >>> PHASE_SHIFT) &&
456	>	!UNSAFE.compareAndSwapLong
457	>	(this, stateOffset, s,
458	>	s = (((long)phase << PHASE_SHIFT) |
459	>	(s & PARTIES_MASK) |
460	>	((p = (int)s >>> PARTIES_SHIFT) == 0 ? EMPTY :
461	>	(u = (int)s & UNARRIVED_MASK) == 0 ? p : u))))
462	>	s = state;
463		}
464	+	return s;
465		}
466
467		/**
468	<	* Creates a new phaser without any initially registered parties,
469	<	* initial phase number 0, and no parent. Any thread using this
468	>	* Creates a new phaser with no initially registered parties, no
469	>	* parent, and initial phase number 0. Any thread using this
470		* phaser will need to first register for it.
471		*/
472		public Phaser() {
#	Line 423 | Line 475 | public class Phaser {
475
476		/**
477		* Creates a new phaser with the given number of registered
478	<	* unarrived parties, initial phase number 0, and no parent.
478	>	* unarrived parties, no parent, and initial phase number 0.
479		*
480	<	* @param parties the number of parties required to trip barrier
480	>	* @param parties the number of parties required to advance to the
481	>	* next phase
482		* @throws IllegalArgumentException if parties less than zero
483		* or greater than the maximum number of parties supported
484		*/
#	Line 434 | Line 487 | public class Phaser {
487		}
488
489		/**
490	<	* Creates a new phaser with the given parent, without any
438	<	* initially registered parties. If parent is non-null this phaser
439	<	* is registered with the parent and its initial phase number is
440	<	* the same as that of parent phaser.
490	>	* Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
491		*
492		* @param parent the parent phaser
493		*/
#	Line 447 | Line 497 | public class Phaser {
497
498		/**
499		* Creates a new phaser with the given parent and number of
500	<	* registered unarrived parties. If parent is non-null, this phaser
501	<	* is registered with the parent and its initial phase number is
502	<	* the same as that of parent phaser.
500	>	* registered unarrived parties.  When the given parent is non-null
501	>	* and the given number of parties is greater than zero, this
502	>	* child phaser is registered with its parent.
503		*
504		* @param parent the parent phaser
505	<	* @param parties the number of parties required to trip barrier
505	>	* @param parties the number of parties required to advance to the
506	>	* next phase
507		* @throws IllegalArgumentException if parties less than zero
508		* or greater than the maximum number of parties supported
509		*/
510		public Phaser(Phaser parent, int parties) {
511		if (parties >>> PARTIES_SHIFT != 0)
512		throw new IllegalArgumentException("Illegal number of parties");
513	<	int phase;
513	>	int phase = 0;
514		this.parent = parent;
515		if (parent != null) {
516	<	Phaser r = parent.root;
517	<	this.root = r;
518	<	this.evenQ = r.evenQ;
519	<	this.oddQ = r.oddQ;
520	<	phase = parent.register();
516	>	final Phaser root = parent.root;
517	>	this.root = root;
518	>	this.evenQ = root.evenQ;
519	>	this.oddQ = root.oddQ;
520	>	if (parties != 0)
521	>	phase = parent.doRegister(1);
522		}
523		else {
524		this.root = this;
525		this.evenQ = new AtomicReference<QNode>();
526		this.oddQ = new AtomicReference<QNode>();
475	–	phase = 0;
527		}
528	<	long p = (long)parties;
529	<	this.state = (((long)phase) << PHASE_SHIFT) | p | (p << PARTIES_SHIFT);
528	>	this.state = (parties == 0) ? (long)EMPTY :
529	>	((long)phase << PHASE_SHIFT) |
530	>	((long)parties << PARTIES_SHIFT) |
531	>	((long)parties);
532		}
533
534		/**
535	<	* Adds a new unarrived party to this phaser.
536	<	* If an ongoing invocation of {@link #onAdvance} is in progress,
537	<	* this method may wait until its completion before registering.
538	<	*
539	<	* @return the arrival phase number to which this registration applied
535	>	* Adds a new unarrived party to this phaser.  If an ongoing
536	>	* invocation of {@link #onAdvance} is in progress, this method
537	>	* may await its completion before returning.  If this phaser has
538	>	* a parent, and this phaser previously had no registered parties,
539	>	* this child phaser is also registered with its parent. If
540	>	* this phaser is terminated, the attempt to register has
541	>	* no effect, and a negative value is returned.
542	>	*
543	>	* @return the arrival phase number to which this registration
544	>	* applied.  If this value is negative, then this phaser has
545	>	* terminated, in which case registration has no effect.
546		* @throws IllegalStateException if attempting to register more
547		* than the maximum supported number of parties
548		*/
#	Line 494 | Line 553 | public class Phaser {
553		/**
554		* Adds the given number of new unarrived parties to this phaser.
555		* If an ongoing invocation of {@link #onAdvance} is in progress,
556	<	* this method may wait until its completion before registering.
557	<	*
558	<	* @param parties the number of additional parties required to trip barrier
559	<	* @return the arrival phase number to which this registration applied
556	>	* this method may await its completion before returning.  If this
557	>	* phaser has a parent, and the given number of parties is greater
558	>	* than zero, and this phaser previously had no registered
559	>	* parties, this child phaser is also registered with its parent.
560	>	* If this phaser is terminated, the attempt to register has no
561	>	* effect, and a negative value is returned.
562	>	*
563	>	* @param parties the number of additional parties required to
564	>	* advance to the next phase
565	>	* @return the arrival phase number to which this registration
566	>	* applied.  If this value is negative, then this phaser has
567	>	* terminated, in which case registration has no effect.
568		* @throws IllegalStateException if attempting to register more
569		* than the maximum supported number of parties
570		* @throws IllegalArgumentException if {@code parties < 0}
#	Line 505 | Line 572 | public class Phaser {
572		public int bulkRegister(int parties) {
573		if (parties < 0)
574		throw new IllegalArgumentException();
508	–	if (parties > MAX_PARTIES)
509	–	throw new IllegalStateException(badRegister(state));
575		if (parties == 0)
576		return getPhase();
577		return doRegister(parties);
578		}
579
580		/**
581	<	* Arrives at the barrier, but does not wait for others.  (You can
582	<	* in turn wait for others via {@link #awaitAdvance}).  It is an
583	<	* unenforced usage error for an unregistered party to invoke this
584	<	* method.
581	>	* Arrives at this phaser, without waiting for others to arrive.
582	>	*
583	>	* <p>It is a usage error for an unregistered party to invoke this
584	>	* method.  However, this error may result in an {@code
585	>	* IllegalStateException} only upon some subsequent operation on
586	>	* this phaser, if ever.
587		*
588		* @return the arrival phase number, or a negative value if terminated
589		* @throws IllegalStateException if not terminated and the number
590		* of unarrived parties would become negative
591		*/
592		public int arrive() {
593	<	return doArrive(ONE_ARRIVAL);
593	>	return doArrive(false);
594		}
595
596		/**
597	<	* Arrives at the barrier and deregisters from it without waiting
598	<	* for others. Deregistration reduces the number of parties
599	<	* required to trip the barrier in future phases.  If this phaser
597	>	* Arrives at this phaser and deregisters from it without waiting
598	>	* for others to arrive. Deregistration reduces the number of
599	>	* parties required to advance in future phases.  If this phaser
600		* has a parent, and deregistration causes this phaser to have
601	<	* zero parties, this phaser also arrives at and is deregistered
602	<	* from its parent.  It is an unenforced usage error for an
603	<	* unregistered party to invoke this method.
601	>	* zero parties, this phaser is also deregistered from its parent.
602	>	*
603	>	* <p>It is a usage error for an unregistered party to invoke this
604	>	* method.  However, this error may result in an {@code
605	>	* IllegalStateException} only upon some subsequent operation on
606	>	* this phaser, if ever.
607		*
608		* @return the arrival phase number, or a negative value if terminated
609		* @throws IllegalStateException if not terminated and the number
610		* of registered or unarrived parties would become negative
611		*/
612		public int arriveAndDeregister() {
613	<	return doArrive(ONE_ARRIVAL|ONE_PARTY);
613	>	return doArrive(true);
614		}
615
616		/**
617	<	* Arrives at the barrier and awaits others. Equivalent in effect
617	>	* Arrives at this phaser and awaits others. Equivalent in effect
618		* to {@code awaitAdvance(arrive())}.  If you need to await with
619		* interruption or timeout, you can arrange this with an analogous
620		* construction using one of the other forms of the {@code
621		* awaitAdvance} method.  If instead you need to deregister upon
622	<	* arrival, use {@link #arriveAndDeregister}. It is an unenforced
623	<	* usage error for an unregistered party to invoke this method.
622	>	* arrival, use {@code awaitAdvance(arriveAndDeregister())}.
623	>	*
624	>	* <p>It is a usage error for an unregistered party to invoke this
625	>	* method.  However, this error may result in an {@code
626	>	* IllegalStateException} only upon some subsequent operation on
627	>	* this phaser, if ever.
628		*
629	<	* @return the arrival phase number, or a negative number if terminated
629	>	* @return the arrival phase number, or the (negative)
630	>	* {@linkplain #getPhase() current phase} if terminated
631		* @throws IllegalStateException if not terminated and the number
632		* of unarrived parties would become negative
633		*/
634		public int arriveAndAwaitAdvance() {
635	<	return awaitAdvance(arrive());
635	>	// Specialization of doArrive+awaitAdvance eliminating some reads/paths
636	>	final Phaser root = this.root;
637	>	for (;;) {
638	>	long s = (root == this) ? state : reconcileState();
639	>	int phase = (int)(s >>> PHASE_SHIFT);
640	>	int counts = (int)s;
641	>	int unarrived = (counts & UNARRIVED_MASK) - 1;
642	>	if (phase < 0)
643	>	return phase;
644	>	else if (counts == EMPTY || unarrived < 0) {
645	>	if (reconcileState() == s)
646	>	throw new IllegalStateException(badArrive(s));
647	>	}
648	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s,
649	>	s -= ONE_ARRIVAL)) {
650	>	if (unarrived != 0)
651	>	return root.internalAwaitAdvance(phase, null);
652	>	if (root != this)
653	>	return parent.arriveAndAwaitAdvance();
654	>	long n = s & PARTIES_MASK;  // base of next state
655	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
656	>	if (onAdvance(phase, nextUnarrived))
657	>	n |= TERMINATION_BIT;
658	>	else if (nextUnarrived == 0)
659	>	n |= EMPTY;
660	>	else
661	>	n |= nextUnarrived;
662	>	int nextPhase = (phase + 1) & MAX_PHASE;
663	>	n |= (long)nextPhase << PHASE_SHIFT;
664	>	if (!UNSAFE.compareAndSwapLong(this, stateOffset, s, n))
665	>	return (int)(state >>> PHASE_SHIFT); // terminated
666	>	releaseWaiters(phase);
667	>	return nextPhase;
668	>	}
669	>	}
670		}
671
672		/**
673	<	* Awaits the phase of the barrier to advance from the given phase
674	<	* value, returning immediately if the current phase of the
675	<	* barrier is not equal to the given phase value or this barrier
567	<	* is terminated.
673	>	* Awaits the phase of this phaser to advance from the given phase
674	>	* value, returning immediately if the current phase is not equal
675	>	* to the given phase value or this phaser is terminated.
676		*
677		* @param phase an arrival phase number, or negative value if
678		* terminated; this argument is normally the value returned by a
679	<	* previous call to {@code arrive} or its variants
680	<	* @return the next arrival phase number, or a negative value
681	<	* if terminated or argument is negative
679	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
680	>	* @return the next arrival phase number, or the argument if it is
681	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
682	>	* if terminated
683		*/
684		public int awaitAdvance(int phase) {
685	<	int p;
685	>	final Phaser root = this.root;
686	>	long s = (root == this) ? state : reconcileState();
687	>	int p = (int)(s >>> PHASE_SHIFT);
688		if (phase < 0)
689		return phase;
690	<	else if ((p = (int)((parent == null? state : reconcileState())
691	<	>>> PHASE_SHIFT)) == phase)
692	<	return internalAwaitAdvance(phase, null);
582	<	else
583	<	return p;
690	>	if (p == phase)
691	>	return root.internalAwaitAdvance(phase, null);
692	>	return p;
693		}
694
695		/**
696	<	* Awaits the phase of the barrier to advance from the given phase
696	>	* Awaits the phase of this phaser to advance from the given phase
697		* value, throwing {@code InterruptedException} if interrupted
698	<	* while waiting, or returning immediately if the current phase of
699	<	* the barrier is not equal to the given phase value or this
700	<	* barrier is terminated.
698	>	* while waiting, or returning immediately if the current phase is
699	>	* not equal to the given phase value or this phaser is
700	>	* terminated.
701		*
702		* @param phase an arrival phase number, or negative value if
703		* terminated; this argument is normally the value returned by a
704	<	* previous call to {@code arrive} or its variants
705	<	* @return the next arrival phase number, or a negative value
706	<	* if terminated or argument is negative
704	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
705	>	* @return the next arrival phase number, or the argument if it is
706	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
707	>	* if terminated
708		* @throws InterruptedException if thread interrupted while waiting
709		*/
710		public int awaitAdvanceInterruptibly(int phase)
711		throws InterruptedException {
712	<	int p;
712	>	final Phaser root = this.root;
713	>	long s = (root == this) ? state : reconcileState();
714	>	int p = (int)(s >>> PHASE_SHIFT);
715		if (phase < 0)
716		return phase;
717	<	if ((p = (int)((parent == null? state : reconcileState())
606	<	>>> PHASE_SHIFT)) == phase) {
717	>	if (p == phase) {
718		QNode node = new QNode(this, phase, true, false, 0L);
719	<	p = internalAwaitAdvance(phase, node);
719	>	p = root.internalAwaitAdvance(phase, node);
720		if (node.wasInterrupted)
721		throw new InterruptedException();
722		}
#	Line 613 | Line 724 | public class Phaser {
724		}
725
726		/**
727	<	* Awaits the phase of the barrier to advance from the given phase
727	>	* Awaits the phase of this phaser to advance from the given phase
728		* value or the given timeout to elapse, throwing {@code
729		* InterruptedException} if interrupted while waiting, or
730	<	* returning immediately if the current phase of the barrier is
731	<	* not equal to the given phase value or this barrier is
621	<	* terminated.
730	>	* returning immediately if the current phase is not equal to the
731	>	* given phase value or this phaser is terminated.
732		*
733		* @param phase an arrival phase number, or negative value if
734		* terminated; this argument is normally the value returned by a
735	<	* previous call to {@code arrive} or its variants
735	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
736		* @param timeout how long to wait before giving up, in units of
737		*        {@code unit}
738		* @param unit a {@code TimeUnit} determining how to interpret the
739		*        {@code timeout} parameter
740	<	* @return the next arrival phase number, or a negative value
741	<	* if terminated or argument is negative
740	>	* @return the next arrival phase number, or the argument if it is
741	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
742	>	* if terminated
743		* @throws InterruptedException if thread interrupted while waiting
744		* @throws TimeoutException if timed out while waiting
745		*/
#	Line 636 | Line 747 | public class Phaser {
747		long timeout, TimeUnit unit)
748		throws InterruptedException, TimeoutException {
749		long nanos = unit.toNanos(timeout);
750	<	int p;
750	>	final Phaser root = this.root;
751	>	long s = (root == this) ? state : reconcileState();
752	>	int p = (int)(s >>> PHASE_SHIFT);
753		if (phase < 0)
754		return phase;
755	<	if ((p = (int)((parent == null? state : reconcileState())
643	<	>>> PHASE_SHIFT)) == phase) {
755	>	if (p == phase) {
756		QNode node = new QNode(this, phase, true, true, nanos);
757	<	p = internalAwaitAdvance(phase, node);
757	>	p = root.internalAwaitAdvance(phase, node);
758		if (node.wasInterrupted)
759		throw new InterruptedException();
760		else if (p == phase)
#	Line 652 | Line 764 | public class Phaser {
764		}
765
766		/**
767	<	* Forces this barrier to enter termination state.  Counts of
768	<	* arrived and registered parties are unaffected.  If this phaser
769	<	* is a member of a tiered set of phasers, then all of the phasers
770	<	* in the set are terminated.  If this phaser is already
771	<	* terminated, this method has no effect.  This method may be
772	<	* useful for coordinating recovery after one or more tasks
773	<	* encounter unexpected exceptions.
767	>	* Forces this phaser to enter termination state.  Counts of
768	>	* registered parties are unaffected.  If this phaser is a member
769	>	* of a tiered set of phasers, then all of the phasers in the set
770	>	* are terminated.  If this phaser is already terminated, this
771	>	* method has no effect.  This method may be useful for
772	>	* coordinating recovery after one or more tasks encounter
773	>	* unexpected exceptions.
774		*/
775		public void forceTermination() {
776		// Only need to change root state
#	Line 666 | Line 778 | public class Phaser {
778		long s;
779		while ((s = root.state) >= 0) {
780		if (UNSAFE.compareAndSwapLong(root, stateOffset,
781	<	s, s | TERMINATION_PHASE)) {
782	<	releaseWaiters(0); // signal all threads
781	>	s, s | TERMINATION_BIT)) {
782	>	// signal all threads
783	>	releaseWaiters(0);
784		releaseWaiters(1);
785		return;
786		}
#	Line 677 | Line 790 | public class Phaser {
790		/**
791		* Returns the current phase number. The maximum phase number is
792		* {@code Integer.MAX_VALUE}, after which it restarts at
793	<	* zero. Upon termination, the phase number is negative.
793	>	* zero. Upon termination, the phase number is negative,
794	>	* in which case the prevailing phase prior to termination
795	>	* may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
796		*
797		* @return the phase number, or a negative value if terminated
798		*/
799		public final int getPhase() {
800	<	return (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
800	>	return (int)(root.state >>> PHASE_SHIFT);
801		}
802
803		/**
804	<	* Returns the number of parties registered at this barrier.
804	>	* Returns the number of parties registered at this phaser.
805		*
806		* @return the number of parties
807		*/
808		public int getRegisteredParties() {
809	<	return partiesOf(parent==null? state : reconcileState());
809	>	return partiesOf(state);
810		}
811
812		/**
813		* Returns the number of registered parties that have arrived at
814	<	* the current phase of this barrier.
814	>	* the current phase of this phaser. If this phaser has terminated,
815	>	* the returned value is meaningless and arbitrary.
816		*
817		* @return the number of arrived parties
818		*/
819		public int getArrivedParties() {
820	<	return arrivedOf(parent==null? state : reconcileState());
820	>	return arrivedOf(reconcileState());
821		}
822
823		/**
824		* Returns the number of registered parties that have not yet
825	<	* arrived at the current phase of this barrier.
825	>	* arrived at the current phase of this phaser. If this phaser has
826	>	* terminated, the returned value is meaningless and arbitrary.
827		*
828		* @return the number of unarrived parties
829		*/
830		public int getUnarrivedParties() {
831	<	return unarrivedOf(parent==null? state : reconcileState());
831	>	return unarrivedOf(reconcileState());
832		}
833
834		/**
#	Line 734 | Line 851 | public class Phaser {
851		}
852
853		/**
854	<	* Returns {@code true} if this barrier has been terminated.
854	>	* Returns {@code true} if this phaser has been terminated.
855		*
856	<	* @return {@code true} if this barrier has been terminated
856	>	* @return {@code true} if this phaser has been terminated
857		*/
858		public boolean isTerminated() {
859	<	return (parent == null? state : reconcileState()) < 0;
859	>	return root.state < 0L;
860		}
861
862		/**
863		* Overridable method to perform an action upon impending phase
864		* advance, and to control termination. This method is invoked
865	<	* upon arrival of the party tripping the barrier (when all other
865	>	* upon arrival of the party advancing this phaser (when all other
866		* waiting parties are dormant).  If this method returns {@code
867	<	* true}, then, rather than advance the phase number, this barrier
868	<	* will be set to a final termination state, and subsequent calls
869	<	* to {@link #isTerminated} will return true. Any (unchecked)
870	<	* Exception or Error thrown by an invocation of this method is
871	<	* propagated to the party attempting to trip the barrier, in
872	<	* which case no advance occurs.
867	>	* true}, this phaser will be set to a final termination state
868	>	* upon advance, and subsequent calls to {@link #isTerminated}
869	>	* will return true. Any (unchecked) Exception or Error thrown by
870	>	* an invocation of this method is propagated to the party
871	>	* attempting to advance this phaser, in which case no advance
872	>	* occurs.
873		*
874		* <p>The arguments to this method provide the state of the phaser
875		* prevailing for the current transition.  The effects of invoking
876	<	* arrival, registration, and waiting methods on this Phaser from
876	>	* arrival, registration, and waiting methods on this phaser from
877		* within {@code onAdvance} are unspecified and should not be
878		* relied on.
879		*
880	<	* <p>If this Phaser is a member of a tiered set of Phasers, then
881	<	* {@code onAdvance} is invoked only for its root Phaser on each
880	>	* <p>If this phaser is a member of a tiered set of phasers, then
881	>	* {@code onAdvance} is invoked only for its root phaser on each
882		* advance.
883		*
884	<	* <p>The default version returns {@code true} when the number of
885	<	* registered parties is zero. Normally, overrides that arrange
886	<	* termination for other reasons should also preserve this
887	<	* property.
884	>	* <p>To support the most common use cases, the default
885	>	* implementation of this method returns {@code true} when the
886	>	* number of registered parties has become zero as the result of a
887	>	* party invoking {@code arriveAndDeregister}.  You can disable
888	>	* this behavior, thus enabling continuation upon future
889	>	* registrations, by overriding this method to always return
890	>	* {@code false}:
891	>	*
892	>	* <pre> {@code
893	>	* Phaser phaser = new Phaser() {
894	>	*   protected boolean onAdvance(int phase, int parties) { return false; }
895	>	* }}</pre>
896		*
897	<	* @param phase the phase number on entering the barrier
897	>	* @param phase the current phase number on entry to this method,
898	>	* before this phaser is advanced
899		* @param registeredParties the current number of registered parties
900	<	* @return {@code true} if this barrier should terminate
900	>	* @return {@code true} if this phaser should terminate
901		*/
902		protected boolean onAdvance(int phase, int registeredParties) {
903	<	return registeredParties <= 0;
903	>	return registeredParties == 0;
904		}
905
906		/**
#	Line 784 | Line 910 | public class Phaser {
910		* followed by the number of registered parties, and {@code
911		* "arrived = "} followed by the number of arrived parties.
912		*
913	<	* @return a string identifying this barrier, as well as its state
913	>	* @return a string identifying this phaser, as well as its state
914		*/
915		public String toString() {
916		return stateToString(reconcileState());
#	Line 803 | Line 929 | public class Phaser {
929		// Waiting mechanics
930
931		/**
932	<	* Removes and signals threads from queue for phase
932	>	* Removes and signals threads from queue for phase.
933		*/
934		private void releaseWaiters(int phase) {
935	<	AtomicReference<QNode> head = queueFor(phase);
936	<	QNode q;
937	<	int p;
935	>	QNode q;   // first element of queue
936	>	Thread t;  // its thread
937	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
938		while ((q = head.get()) != null &&
939	<	((p = q.phase) == phase ||
940	<	(int)(root.state >>> PHASE_SHIFT) != p)) {
941	<	if (head.compareAndSet(q, q.next))
942	<	q.signal();
939	>	q.phase != (int)(root.state >>> PHASE_SHIFT)) {
940	>	if (head.compareAndSet(q, q.next) &&
941	>	(t = q.thread) != null) {
942	>	q.thread = null;
943	>	LockSupport.unpark(t);
944	>	}
945		}
946		}
947
948		/**
949	<	* Tries to enqueue given node in the appropriate wait queue.
949	>	* Variant of releaseWaiters that additionally tries to remove any
950	>	* nodes no longer waiting for advance due to timeout or
951	>	* interrupt. Currently, nodes are removed only if they are at
952	>	* head of queue, which suffices to reduce memory footprint in
953	>	* most usages.
954		*
955	<	* @return true if successful
955	>	* @return current phase on exit
956		*/
957	<	private boolean tryEnqueue(int phase, QNode node) {
958	<	releaseWaiters(phase-1); // ensure old queue clean
959	<	AtomicReference<QNode> head = queueFor(phase);
960	<	QNode q = head.get();
961	<	return ((q == null || q.phase == phase) &&
962	<	(int)(root.state >>> PHASE_SHIFT) == phase &&
963	<	head.compareAndSet(node.next = q, node));
957	>	private int abortWait(int phase) {
958	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
959	>	for (;;) {
960	>	Thread t;
961	>	QNode q = head.get();
962	>	int p = (int)(root.state >>> PHASE_SHIFT);
963	>	if (q == null || ((t = q.thread) != null && q.phase == p))
964	>	return p;
965	>	if (head.compareAndSet(q, q.next) && t != null) {
966	>	q.thread = null;
967	>	LockSupport.unpark(t);
968	>	}
969	>	}
970		}
971
972		/** The number of CPUs, for spin control */
#	Line 842 | Line 980 | public class Phaser {
980		* avoid it when threads regularly arrive: When a thread in
981		* internalAwaitAdvance notices another arrival before blocking,
982		* and there appear to be enough CPUs available, it spins
983	<	* SPINS_PER_ARRIVAL more times before blocking. Plus, even on
984	<	* uniprocessors, there is at least one intervening Thread.yield
847	<	* before blocking. The value trades off good-citizenship vs big
848	<	* unnecessary slowdowns.
983	>	* SPINS_PER_ARRIVAL more times before blocking. The value trades
984	>	* off good-citizenship vs big unnecessary slowdowns.
985		*/
986		static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
987
988		/**
989		* Possibly blocks and waits for phase to advance unless aborted.
990	+	* Call only from root node.
991		*
992		* @param phase current phase
993		* @param node if non-null, the wait node to track interrupt and timeout;
#	Line 858 | Line 995 | public class Phaser {
995		* @return current phase
996		*/
997		private int internalAwaitAdvance(int phase, QNode node) {
998	<	Phaser current = this;       // to eventually wait at root if tiered
999	<	boolean queued = false;      // true when node is enqueued
1000	<	int lastUnarrived = -1;      // to increase spins upon change
998	>	releaseWaiters(phase-1);          // ensure old queue clean
999	>	boolean queued = false;           // true when node is enqueued
1000	>	int lastUnarrived = 0;            // to increase spins upon change
1001		int spins = SPINS_PER_ARRIVAL;
1002	<	for (;;) {
1003	<	int p, unarrived;
1004	<	Phaser par;
1005	<	long s = current.state;
1006	<	if ((p = (int)(s >>> PHASE_SHIFT)) != phase) {
1007	<	if (node != null)
1008	<	node.onRelease();
872	<	releaseWaiters(phase);
873	<	return p;
874	<	}
875	<	else if ((unarrived = ((int)s) & UNARRIVED_MASK) == 0 &&
876	<	(par = current.parent) != null) {
877	<	current = par;       // if all arrived, use parent
878	<	par = par.parent;
879	<	lastUnarrived = -1;
880	<	}
881	<	else if (unarrived != lastUnarrived) {
882	<	if ((lastUnarrived = unarrived) < NCPU)
1002	>	long s;
1003	>	int p;
1004	>	while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
1005	>	if (node == null) {           // spinning in noninterruptible mode
1006	>	int unarrived = (int)s & UNARRIVED_MASK;
1007	>	if (unarrived != lastUnarrived &&
1008	>	(lastUnarrived = unarrived) < NCPU)
1009		spins += SPINS_PER_ARRIVAL;
1010	+	boolean interrupted = Thread.interrupted();
1011	+	if (interrupted || --spins < 0) { // need node to record intr
1012	+	node = new QNode(this, phase, false, false, 0L);
1013	+	node.wasInterrupted = interrupted;
1014	+	}
1015		}
1016	<	else if (spins > 0) {
1017	<	if (--spins == (SPINS_PER_ARRIVAL >>> 1))
1018	<	Thread.yield();  // yield midway through spin
1019	<	}
1020	<	else if (node == null)   // must be noninterruptible
1021	<	node = new QNode(this, phase, false, false, 0L);
1022	<	else if (node.isReleasable()) {
1023	<	if ((int)(reconcileState() >>> PHASE_SHIFT) == phase)
893	<	return phase;    // aborted
1016	>	else if (node.isReleasable()) // done or aborted
1017	>	break;
1018	>	else if (!queued) {           // push onto queue
1019	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
1020	>	QNode q = node.next = head.get();
1021	>	if ((q == null || q.phase == phase) &&
1022	>	(int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
1023	>	queued = head.compareAndSet(q, node);
1024		}
895	–	else if (!queued)
896	–	queued = tryEnqueue(phase, node);
1025		else {
1026		try {
1027		ForkJoinPool.managedBlock(node);
#	Line 902 | Line 1030 | public class Phaser {
1030		}
1031		}
1032		}
1033	+
1034	+	if (node != null) {
1035	+	if (node.thread != null)
1036	+	node.thread = null;       // avoid need for unpark()
1037	+	if (node.wasInterrupted && !node.interruptible)
1038	+	Thread.currentThread().interrupt();
1039	+	if (p == phase && (p = (int)(state >>> PHASE_SHIFT)) == phase)
1040	+	return abortWait(phase); // possibly clean up on abort
1041	+	}
1042	+	releaseWaiters(phase);
1043	+	return p;
1044		}
1045
1046		/**
#	Line 925 | Line 1064 | public class Phaser {
1064		this.interruptible = interruptible;
1065		this.nanos = nanos;
1066		this.timed = timed;
1067	<	this.lastTime = timed? System.nanoTime() : 0L;
1067	>	this.lastTime = timed ? System.nanoTime() : 0L;
1068		thread = Thread.currentThread();
1069		}
1070
1071		public boolean isReleasable() {
1072	<	Thread t = thread;
1073	<	if (t != null) {
1074	<	if (phaser.getPhase() != phase)
936	<	t = null;
937	<	else {
938	<	if (Thread.interrupted())
939	<	wasInterrupted = true;
940	<	if (interruptible && wasInterrupted)
941	<	t = null;
942	<	else if (timed) {
943	<	if (nanos > 0) {
944	<	long now = System.nanoTime();
945	<	nanos -= now - lastTime;
946	<	lastTime = now;
947	<	}
948	<	if (nanos <= 0)
949	<	t = null;
950	<	}
951	<	}
952	<	if (t != null)
953	<	return false;
1072	>	if (thread == null)
1073	>	return true;
1074	>	if (phaser.getPhase() != phase) {
1075		thread = null;
1076	+	return true;
1077		}
1078	<	return true;
1078	>	if (Thread.interrupted())
1079	>	wasInterrupted = true;
1080	>	if (wasInterrupted && interruptible) {
1081	>	thread = null;
1082	>	return true;
1083	>	}
1084	>	if (timed) {
1085	>	if (nanos > 0L) {
1086	>	long now = System.nanoTime();
1087	>	nanos -= now - lastTime;
1088	>	lastTime = now;
1089	>	}
1090	>	if (nanos <= 0L) {
1091	>	thread = null;
1092	>	return true;
1093	>	}
1094	>	}
1095	>	return false;
1096		}
1097
1098		public boolean block() {
#	Line 965 | Line 1104 | public class Phaser {
1104		LockSupport.parkNanos(this, nanos);
1105		return isReleasable();
1106		}
968	–
969	–	void signal() {
970	–	Thread t = thread;
971	–	if (t != null) {
972	–	thread = null;
973	–	LockSupport.unpark(t);
974	–	}
975	–	}
976	–
977	–	void onRelease() { // actions upon return from internalAwaitAdvance
978	–	if (!interruptible && wasInterrupted)
979	–	Thread.currentThread().interrupt();
980	–	if (thread != null)
981	–	thread = null;
982	–	}
983	–
1107		}
1108
1109		// Unsafe mechanics
1110
1111	<	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1112	<	private static final long stateOffset =
1113	<	objectFieldOffset("state", Phaser.class);
991	<
992	<	private static long objectFieldOffset(String field, Class<?> klazz) {
1111	>	private static final sun.misc.Unsafe UNSAFE;
1112	>	private static final long stateOffset;
1113	>	static {
1114		try {
1115	<	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1116	<	} catch (NoSuchFieldException e) {
1117	<	// Convert Exception to corresponding Error
1118	<	NoSuchFieldError error = new NoSuchFieldError(field);
1119	<	error.initCause(e);
1120	<	throw error;
1115	>	UNSAFE = getUnsafe();
1116	>	Class k = Phaser.class;
1117	>	stateOffset = UNSAFE.objectFieldOffset
1118	>	(k.getDeclaredField("state"));
1119	>	} catch (Exception e) {
1120	>	throw new Error(e);
1121		}
1122		}
1123

Diff Legend

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