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 1.54 by dl, Sat Nov 13 13:10:04 2010 UTC vs.
Revision 1.76 by jsr166, Sat Oct 15 21:46:25 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_COUNT      = 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 long UNARRIVED_MASK = 0xffffL;
273	<	private static final long PARTIES_MASK   = 0xffff0000L;
274	<	private static final long ONE_ARRIVAL    = 1L;
275	<	private static final long ONE_PARTY      = 1L << PARTIES_SHIFT;
276	<	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  ONE_DEREGISTER  = ONE_ARRIVAL|ONE_PARTY;
280	>	private static final int  EMPTY           = 1;
281
282		// The following unpacking methods are usually manually inlined
283
284		private static int unarrivedOf(long s) {
285	<	return (int) (s & UNARRIVED_MASK);
285	>	int counts = (int)s;
286	>	return (counts == EMPTY) ? 0 : (counts & UNARRIVED_MASK);
287		}
288
289		private static int partiesOf(long s) {
290	<	return ((int) (s & PARTIES_MASK)) >>> PARTIES_SHIFT;
290	>	return (int)s >>> PARTIES_SHIFT;
291		}
292
293		private static int phaseOf(long s) {
294	<	return (int) (s >>> PHASE_SHIFT);
294	>	return (int)(s >>> PHASE_SHIFT);
295		}
296
297		private static int arrivedOf(long s) {
298	<	return partiesOf(s) - unarrivedOf(s);
298	>	int counts = (int)s;
299	>	return (counts == EMPTY) ? 0 :
300	>	(counts >>> PARTIES_SHIFT) - (counts & UNARRIVED_MASK);
301		}
302
303		/**
#	Line 274 | Line 306 | public class Phaser {
306		private final Phaser parent;
307
308		/**
309	<	* The root of phaser tree. Equals this if not in a tree.  Used to
278	<	* support faster state push-down.
309	>	* The root of phaser tree. Equals this if not in a tree.
310		*/
311		private final Phaser root;
312
#	Line 293 | Line 324 | public class Phaser {
324		}
325
326		/**
327	+	* Returns message string for bounds exceptions on arrival.
328	+	*/
329	+	private String badArrive(long s) {
330	+	return "Attempted arrival of unregistered party for " +
331	+	stateToString(s);
332	+	}
333	+
334	+	/**
335	+	* Returns message string for bounds exceptions on registration.
336	+	*/
337	+	private String badRegister(long s) {
338	+	return "Attempt to register more than " +
339	+	MAX_PARTIES + " parties for " + stateToString(s);
340	+	}
341	+
342	+	/**
343		* Main implementation for methods arrive and arriveAndDeregister.
344		* Manually tuned to speed up and minimize race windows for the
345		* common case of just decrementing unarrived field.
346		*
347	<	* @param adj - adjustment to apply to state -- either
348	<	* ONE_ARRIVAL (for arrive) or
349	<	* ONE_ARRIVAL|ONE_PARTY (for arriveAndDeregister)
347	>	* @param adjust value to subtract from state;
348	>	*               ONE_ARRIVAL for arrive,
349	>	*               ONE_DEREGISTER for arriveAndDeregister
350		*/
351	<	private int doArrive(long adj) {
351	>	private int doArrive(int adjust) {
352	>	final Phaser root = this.root;
353		for (;;) {
354	<	long s;
355	<	int phase, unarrived;
356	<	if ((phase = (int)((s = state) >>> PHASE_SHIFT)) < 0)
354	>	long s = (root == this) ? state : reconcileState();
355	>	int phase = (int)(s >>> PHASE_SHIFT);
356	>	if (phase < 0)
357		return phase;
358	<	else if ((unarrived = (int)(s & UNARRIVED_MASK)) == 0)
359	<	checkBadArrive(s);
360	<	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s -= adj)){
358	>	int counts = (int)s;
359	>	int unarrived = (counts == EMPTY) ? 0 : (counts & UNARRIVED_MASK);
360	>	if (unarrived <= 0)
361	>	throw new IllegalStateException(badArrive(s));
362	>	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adjust)) {
363		if (unarrived == 1) {
364	<	Phaser par;
365	<	long p = s & PARTIES_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	<	UNSAFE.compareAndSwapLong
372	<	(this, stateOffset, s, onAdvance(phase, u)?
373	<	next | TERMINATION_PHASE : next);
364	>	long n = s & PARTIES_MASK;  // base of next state
365	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
366	>	if (root == this) {
367	>	if (onAdvance(phase, nextUnarrived))
368	>	n |= TERMINATION_BIT;
369	>	else if (nextUnarrived == 0)
370	>	n |= EMPTY;
371	>	else
372	>	n |= nextUnarrived;
373	>	int nextPhase = (phase + 1) & MAX_PHASE;
374	>	n |= (long)nextPhase << PHASE_SHIFT;
375	>	UNSAFE.compareAndSwapLong(this, stateOffset, s, n);
376		releaseWaiters(phase);
377		}
378	<	else {
379	<	par.doArrive(u == 0?
380	<	ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL);
381	<	if ((int)(par.state >>> PHASE_SHIFT) != nextPhase ||
330	<	((int)(state >>> PHASE_SHIFT) != nextPhase &&
331	<	!UNSAFE.compareAndSwapLong(this, stateOffset,
332	<	s, next)))
333	<	reconcileState();
378	>	else if (nextUnarrived == 0) { // propagate deregistration
379	>	phase = parent.doArrive(ONE_DEREGISTER);
380	>	UNSAFE.compareAndSwapLong(this, stateOffset,
381	>	s, s | EMPTY);
382		}
383	+	else
384	+	phase = parent.doArrive(ONE_ARRIVAL);
385		}
386		return phase;
387		}
#	Line 339 | Line 389 | public class Phaser {
389		}
390
391		/**
342	–	* Rechecks state and throws bounds exceptions on arrival -- called
343	–	* only if unarrived is apparently zero.
344	–	*/
345	–	private void checkBadArrive(long s) {
346	–	if (reconcileState() == s)
347	–	throw new IllegalStateException
348	–	("Attempted arrival of unregistered party for " +
349	–	stateToString(s));
350	–	}
351	–
352	–	/**
392		* Implementation of register, bulkRegister
393		*
394	<	* @param registrations number to add to both parties and unarrived fields
394	>	* @param registrations number to add to both parties and
395	>	* unarrived fields. Must be greater than zero.
396		*/
397		private int doRegister(int registrations) {
398	<	long adj = (long)registrations; // adjustment to state
399	<	adj |= adj << PARTIES_SHIFT;
400	<	Phaser par = parent;
398	>	// adjustment to state
399	>	long adjust = ((long)registrations << PARTIES_SHIFT) | registrations;
400	>	final Phaser parent = this.parent;
401	>	int phase;
402		for (;;) {
403	<	int phase, parties;
404	<	long s = par == null? state : reconcileState();
405	<	if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
406	<	return phase;
407	<	if ((parties = ((int)(s & PARTIES_MASK)) >>> PARTIES_SHIFT) != 0 &&
367	<	(s & UNARRIVED_MASK) == 0)
368	<	internalAwaitAdvance(phase, null); // wait for onAdvance
369	<	else if (parties + registrations > MAX_COUNT)
403	>	long s = (parent == null) ? state : reconcileState();
404	>	int counts = (int)s;
405	>	int parties = counts >>> PARTIES_SHIFT;
406	>	int unarrived = counts & UNARRIVED_MASK;
407	>	if (registrations > MAX_PARTIES - parties)
408		throw new IllegalStateException(badRegister(s));
409	<	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s + adj))
410	<	return phase;
409	>	phase = (int)(s >>> PHASE_SHIFT);
410	>	if (phase < 0)
411	>	break;
412	>	if (counts != EMPTY) {                  // not 1st registration
413	>	if (parent == null || reconcileState() == s) {
414	>	if (unarrived == 0)             // wait out advance
415	>	root.internalAwaitAdvance(phase, null);
416	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset,
417	>	s, s + adjust))
418	>	break;
419	>	}
420	>	}
421	>	else if (parent == null) {              // 1st root registration
422	>	long next = ((long)phase << PHASE_SHIFT) | adjust;
423	>	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
424	>	break;
425	>	}
426	>	else {
427	>	synchronized (this) {               // 1st sub registration
428	>	if (state == s) {               // recheck under lock
429	>	phase = parent.doRegister(1);
430	>	if (phase < 0)
431	>	break;
432	>	// finish registration whenever parent registration
433	>	// succeeded, even when racing with termination,
434	>	// since these are part of the same "transaction".
435	>	while (!UNSAFE.compareAndSwapLong
436	>	(this, stateOffset, s,
437	>	((long)phase << PHASE_SHIFT) | adjust)) {
438	>	s = state;
439	>	phase = (int)(root.state >>> PHASE_SHIFT);
440	>	// assert (int)s == EMPTY;
441	>	}
442	>	break;
443	>	}
444	>	}
445	>	}
446		}
447	+	return phase;
448		}
449
450		/**
451	<	* Returns message string for bounds exceptions on registration
452	<	*/
453	<	private String badRegister(long s) {
454	<	return "Attempt to register more than " +
455	<	MAX_COUNT + " parties for " + stateToString(s);
456	<	}
457	<
458	<	/**
459	<	* Recursively resolves lagged phase propagation from root if
460	<	* necessary.
451	>	* Resolves lagged phase propagation from root if necessary.
452	>	* Reconciliation normally occurs when root has advanced but
453	>	* subphasers have not yet done so, in which case they must finish
454	>	* their own advance by setting unarrived to parties (or if
455	>	* parties is zero, resetting to unregistered EMPTY state).
456	>	* However, this method may also be called when "floating"
457	>	* subphasers with possibly some unarrived parties are merely
458	>	* catching up to current phase, in which case counts are
459	>	* unaffected.
460	>	*
461	>	* @return reconciled state
462		*/
463		private long reconcileState() {
464	<	Phaser par = parent;
465	<	if (par == null)
466	<	return state;
467	<	Phaser rt = root;
468	<	for (;;) {
469	<	long s, u;
470	<	int phase, rPhase, pPhase;
471	<	if ((phase = (int)((s = state)>>> PHASE_SHIFT)) < 0 ||
472	<	(rPhase = (int)(rt.state >>> PHASE_SHIFT)) == phase)
473	<	return s;
474	<	long pState = par.parent == null? par.state : par.reconcileState();
475	<	if (state == s) {
476	<	if ((rPhase < 0 || (s & UNARRIVED_MASK) == 0) &&
477	<	((pPhase = (int)(pState >>> PHASE_SHIFT)) < 0 ||
478	<	pPhase == ((phase + 1) & MAX_PHASE)))
404	<	UNSAFE.compareAndSwapLong
405	<	(this, stateOffset, s,
406	<	(((long) pPhase) << PHASE_SHIFT) |
407	<	(u = s & PARTIES_MASK) |
408	<	(u >>> PARTIES_SHIFT)); // reset unarrived to parties
409	<	else
410	<	releaseWaiters(phase); // help release others
411	<	}
464	>	final Phaser root = this.root;
465	>	long s = state;
466	>	if (root != this) {
467	>	int phase, u, p;
468	>	// CAS root phase with current parties; possibly trip unarrived
469	>	while ((phase = (int)(root.state >>> PHASE_SHIFT)) !=
470	>	(int)(s >>> PHASE_SHIFT) &&
471	>	!UNSAFE.compareAndSwapLong
472	>	(this, stateOffset, s,
473	>	s = (((long)phase << PHASE_SHIFT) |
474	>	(s & PARTIES_MASK) |
475	>	((p = (int)s >>> PARTIES_SHIFT) == 0 ? EMPTY :
476	>	((u = (int)s & UNARRIVED_MASK) == 0 && phase >= 0) ?
477	>	p : u))))
478	>	s = state;
479		}
480	+	return s;
481		}
482
483		/**
484	<	* Creates a new phaser without any initially registered parties,
485	<	* initial phase number 0, and no parent. Any thread using this
484	>	* Creates a new phaser with no initially registered parties, no
485	>	* parent, and initial phase number 0. Any thread using this
486		* phaser will need to first register for it.
487		*/
488		public Phaser() {
#	Line 423 | Line 491 | public class Phaser {
491
492		/**
493		* Creates a new phaser with the given number of registered
494	<	* unarrived parties, initial phase number 0, and no parent.
494	>	* unarrived parties, no parent, and initial phase number 0.
495		*
496	<	* @param parties the number of parties required to trip barrier
496	>	* @param parties the number of parties required to advance to the
497	>	* next phase
498		* @throws IllegalArgumentException if parties less than zero
499		* or greater than the maximum number of parties supported
500		*/
#	Line 434 | Line 503 | public class Phaser {
503		}
504
505		/**
506	<	* 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.
506	>	* Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
507		*
508		* @param parent the parent phaser
509		*/
#	Line 447 | Line 513 | public class Phaser {
513
514		/**
515		* Creates a new phaser with the given parent and number of
516	<	* registered unarrived parties. If parent is non-null, this phaser
517	<	* is registered with the parent and its initial phase number is
518	<	* the same as that of parent phaser.
516	>	* registered unarrived parties.  When the given parent is non-null
517	>	* and the given number of parties is greater than zero, this
518	>	* child phaser is registered with its parent.
519		*
520		* @param parent the parent phaser
521	<	* @param parties the number of parties required to trip barrier
521	>	* @param parties the number of parties required to advance to the
522	>	* next phase
523		* @throws IllegalArgumentException if parties less than zero
524		* or greater than the maximum number of parties supported
525		*/
526		public Phaser(Phaser parent, int parties) {
527	<	if (parties < 0 || parties > MAX_COUNT)
527	>	if (parties >>> PARTIES_SHIFT != 0)
528		throw new IllegalArgumentException("Illegal number of parties");
529	<	int phase;
529	>	int phase = 0;
530		this.parent = parent;
531		if (parent != null) {
532	<	Phaser r = parent.root;
533	<	this.root = r;
534	<	this.evenQ = r.evenQ;
535	<	this.oddQ = r.oddQ;
536	<	phase = parent.register();
532	>	final Phaser root = parent.root;
533	>	this.root = root;
534	>	this.evenQ = root.evenQ;
535	>	this.oddQ = root.oddQ;
536	>	if (parties != 0)
537	>	phase = parent.doRegister(1);
538		}
539		else {
540		this.root = this;
541		this.evenQ = new AtomicReference<QNode>();
542		this.oddQ = new AtomicReference<QNode>();
475	–	phase = 0;
543		}
544	<	long p = (long)parties;
545	<	this.state = (((long) phase) << PHASE_SHIFT) | p | (p << PARTIES_SHIFT);
544	>	this.state = (parties == 0) ? (long)EMPTY :
545	>	((long)phase << PHASE_SHIFT) |
546	>	((long)parties << PARTIES_SHIFT) |
547	>	((long)parties);
548		}
549
550		/**
551	<	* Adds a new unarrived party to this phaser.
552	<	* If an ongoing invocation of {@link #onAdvance} is in progress,
553	<	* this method may wait until its completion before registering.
554	<	*
555	<	* @return the arrival phase number to which this registration applied
551	>	* Adds a new unarrived party to this phaser.  If an ongoing
552	>	* invocation of {@link #onAdvance} is in progress, this method
553	>	* may await its completion before returning.  If this phaser has
554	>	* a parent, and this phaser previously had no registered parties,
555	>	* this child phaser is also registered with its parent. If
556	>	* this phaser is terminated, the attempt to register has
557	>	* no effect, and a negative value is returned.
558	>	*
559	>	* @return the arrival phase number to which this registration
560	>	* applied.  If this value is negative, then this phaser has
561	>	* terminated, in which case registration has no effect.
562		* @throws IllegalStateException if attempting to register more
563		* than the maximum supported number of parties
564		*/
#	Line 494 | Line 569 | public class Phaser {
569		/**
570		* Adds the given number of new unarrived parties to this phaser.
571		* If an ongoing invocation of {@link #onAdvance} is in progress,
572	<	* this method may wait until its completion before registering.
573	<	*
574	<	* @param parties the number of additional parties required to trip barrier
575	<	* @return the arrival phase number to which this registration applied
572	>	* this method may await its completion before returning.  If this
573	>	* phaser has a parent, and the given number of parties is greater
574	>	* than zero, and this phaser previously had no registered
575	>	* parties, this child phaser is also registered with its parent.
576	>	* If this phaser is terminated, the attempt to register has no
577	>	* effect, and a negative value is returned.
578	>	*
579	>	* @param parties the number of additional parties required to
580	>	* advance to the next phase
581	>	* @return the arrival phase number to which this registration
582	>	* applied.  If this value is negative, then this phaser has
583	>	* terminated, in which case registration has no effect.
584		* @throws IllegalStateException if attempting to register more
585		* than the maximum supported number of parties
586		* @throws IllegalArgumentException if {@code parties < 0}
#	Line 505 | Line 588 | public class Phaser {
588		public int bulkRegister(int parties) {
589		if (parties < 0)
590		throw new IllegalArgumentException();
508	–	if (parties > MAX_COUNT)
509	–	throw new IllegalStateException(badRegister(state));
591		if (parties == 0)
592		return getPhase();
593		return doRegister(parties);
594		}
595
596		/**
597	<	* Arrives at the barrier, but does not wait for others.  (You can
598	<	* in turn wait for others via {@link #awaitAdvance}).  It is an
599	<	* unenforced usage error for an unregistered party to invoke this
600	<	* method.
597	>	* Arrives at this phaser, without waiting for others to arrive.
598	>	*
599	>	* <p>It is a usage error for an unregistered party to invoke this
600	>	* method.  However, this error may result in an {@code
601	>	* IllegalStateException} only upon some subsequent operation on
602	>	* this phaser, if ever.
603		*
604		* @return the arrival phase number, or a negative value if terminated
605		* @throws IllegalStateException if not terminated and the number
#	Line 527 | Line 610 | public class Phaser {
610		}
611
612		/**
613	<	* Arrives at the barrier and deregisters from it without waiting
614	<	* for others. Deregistration reduces the number of parties
615	<	* required to trip the barrier in future phases.  If this phaser
613	>	* Arrives at this phaser and deregisters from it without waiting
614	>	* for others to arrive. Deregistration reduces the number of
615	>	* parties required to advance in future phases.  If this phaser
616		* has a parent, and deregistration causes this phaser to have
617	<	* zero parties, this phaser also arrives at and is deregistered
618	<	* from its parent.  It is an unenforced usage error for an
619	<	* unregistered party to invoke this method.
617	>	* zero parties, this phaser is also deregistered from its parent.
618	>	*
619	>	* <p>It is a usage error for an unregistered party to invoke this
620	>	* method.  However, this error may result in an {@code
621	>	* IllegalStateException} only upon some subsequent operation on
622	>	* this phaser, if ever.
623		*
624		* @return the arrival phase number, or a negative value if terminated
625		* @throws IllegalStateException if not terminated and the number
626		* of registered or unarrived parties would become negative
627		*/
628		public int arriveAndDeregister() {
629	<	return doArrive(ONE_ARRIVAL|ONE_PARTY);
629	>	return doArrive(ONE_DEREGISTER);
630		}
631
632		/**
633	<	* Arrives at the barrier and awaits others. Equivalent in effect
633	>	* Arrives at this phaser and awaits others. Equivalent in effect
634		* to {@code awaitAdvance(arrive())}.  If you need to await with
635		* interruption or timeout, you can arrange this with an analogous
636		* construction using one of the other forms of the {@code
637		* awaitAdvance} method.  If instead you need to deregister upon
638	<	* arrival, use {@link #arriveAndDeregister}. It is an unenforced
639	<	* usage error for an unregistered party to invoke this method.
638	>	* arrival, use {@code awaitAdvance(arriveAndDeregister())}.
639	>	*
640	>	* <p>It is a usage error for an unregistered party to invoke this
641	>	* method.  However, this error may result in an {@code
642	>	* IllegalStateException} only upon some subsequent operation on
643	>	* this phaser, if ever.
644		*
645	<	* @return the arrival phase number, or a negative number if terminated
645	>	* @return the arrival phase number, or the (negative)
646	>	* {@linkplain #getPhase() current phase} if terminated
647		* @throws IllegalStateException if not terminated and the number
648		* of unarrived parties would become negative
649		*/
650		public int arriveAndAwaitAdvance() {
651	<	return awaitAdvance(arrive());
651	>	// Specialization of doArrive+awaitAdvance eliminating some reads/paths
652	>	final Phaser root = this.root;
653	>	for (;;) {
654	>	long s = (root == this) ? state : reconcileState();
655	>	int phase = (int)(s >>> PHASE_SHIFT);
656	>	if (phase < 0)
657	>	return phase;
658	>	int counts = (int)s;
659	>	int unarrived = (counts == EMPTY) ? 0 : (counts & UNARRIVED_MASK);
660	>	if (unarrived <= 0)
661	>	throw new IllegalStateException(badArrive(s));
662	>	if (UNSAFE.compareAndSwapLong(this, stateOffset, s,
663	>	s -= ONE_ARRIVAL)) {
664	>	if (unarrived > 1)
665	>	return root.internalAwaitAdvance(phase, null);
666	>	if (root != this)
667	>	return parent.arriveAndAwaitAdvance();
668	>	long n = s & PARTIES_MASK;  // base of next state
669	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
670	>	if (onAdvance(phase, nextUnarrived))
671	>	n |= TERMINATION_BIT;
672	>	else if (nextUnarrived == 0)
673	>	n |= EMPTY;
674	>	else
675	>	n |= nextUnarrived;
676	>	int nextPhase = (phase + 1) & MAX_PHASE;
677	>	n |= (long)nextPhase << PHASE_SHIFT;
678	>	if (!UNSAFE.compareAndSwapLong(this, stateOffset, s, n))
679	>	return (int)(state >>> PHASE_SHIFT); // terminated
680	>	releaseWaiters(phase);
681	>	return nextPhase;
682	>	}
683	>	}
684		}
685
686		/**
687	<	* Awaits the phase of the barrier to advance from the given phase
688	<	* value, returning immediately if the current phase of the
689	<	* barrier is not equal to the given phase value or this barrier
567	<	* is terminated.
687	>	* Awaits the phase of this phaser to advance from the given phase
688	>	* value, returning immediately if the current phase is not equal
689	>	* to the given phase value or this phaser is terminated.
690		*
691		* @param phase an arrival phase number, or negative value if
692		* terminated; this argument is normally the value returned by a
693	<	* previous call to {@code arrive} or its variants
694	<	* @return the next arrival phase number, or a negative value
695	<	* if terminated or argument is negative
693	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
694	>	* @return the next arrival phase number, or the argument if it is
695	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
696	>	* if terminated
697		*/
698		public int awaitAdvance(int phase) {
699	+	final Phaser root = this.root;
700	+	long s = (root == this) ? state : reconcileState();
701	+	int p = (int)(s >>> PHASE_SHIFT);
702		if (phase < 0)
703		return phase;
704	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
705	<	if (p != phase)
706	<	return p;
581	<	return internalAwaitAdvance(phase, null);
704	>	if (p == phase)
705	>	return root.internalAwaitAdvance(phase, null);
706	>	return p;
707		}
708
709		/**
710	<	* Awaits the phase of the barrier to advance from the given phase
710	>	* Awaits the phase of this phaser to advance from the given phase
711		* value, throwing {@code InterruptedException} if interrupted
712	<	* while waiting, or returning immediately if the current phase of
713	<	* the barrier is not equal to the given phase value or this
714	<	* barrier is terminated.
712	>	* while waiting, or returning immediately if the current phase is
713	>	* not equal to the given phase value or this phaser is
714	>	* terminated.
715		*
716		* @param phase an arrival phase number, or negative value if
717		* terminated; this argument is normally the value returned by a
718	<	* previous call to {@code arrive} or its variants
719	<	* @return the next arrival phase number, or a negative value
720	<	* if terminated or argument is negative
718	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
719	>	* @return the next arrival phase number, or the argument if it is
720	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
721	>	* if terminated
722		* @throws InterruptedException if thread interrupted while waiting
723		*/
724		public int awaitAdvanceInterruptibly(int phase)
725		throws InterruptedException {
726	+	final Phaser root = this.root;
727	+	long s = (root == this) ? state : reconcileState();
728	+	int p = (int)(s >>> PHASE_SHIFT);
729		if (phase < 0)
730		return phase;
731	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
732	<	if (p != phase)
733	<	return p;
734	<	QNode node = new QNode(this, phase, true, false, 0L);
735	<	p = internalAwaitAdvance(phase, node);
736	<	if (node.wasInterrupted)
737	<	throw new InterruptedException();
609	<	else
610	<	return p;
731	>	if (p == phase) {
732	>	QNode node = new QNode(this, phase, true, false, 0L);
733	>	p = root.internalAwaitAdvance(phase, node);
734	>	if (node.wasInterrupted)
735	>	throw new InterruptedException();
736	>	}
737	>	return p;
738		}
739
740		/**
741	<	* Awaits the phase of the barrier to advance from the given phase
741	>	* Awaits the phase of this phaser to advance from the given phase
742		* value or the given timeout to elapse, throwing {@code
743		* InterruptedException} if interrupted while waiting, or
744	<	* returning immediately if the current phase of the barrier is
745	<	* not equal to the given phase value or this barrier is
619	<	* terminated.
744	>	* returning immediately if the current phase is not equal to the
745	>	* given phase value or this phaser is terminated.
746		*
747		* @param phase an arrival phase number, or negative value if
748		* terminated; this argument is normally the value returned by a
749	<	* previous call to {@code arrive} or its variants
749	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
750		* @param timeout how long to wait before giving up, in units of
751		*        {@code unit}
752		* @param unit a {@code TimeUnit} determining how to interpret the
753		*        {@code timeout} parameter
754	<	* @return the next arrival phase number, or a negative value
755	<	* if terminated or argument is negative
754	>	* @return the next arrival phase number, or the argument if it is
755	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
756	>	* if terminated
757		* @throws InterruptedException if thread interrupted while waiting
758		* @throws TimeoutException if timed out while waiting
759		*/
#	Line 634 | Line 761 | public class Phaser {
761		long timeout, TimeUnit unit)
762		throws InterruptedException, TimeoutException {
763		long nanos = unit.toNanos(timeout);
764	+	final Phaser root = this.root;
765	+	long s = (root == this) ? state : reconcileState();
766	+	int p = (int)(s >>> PHASE_SHIFT);
767		if (phase < 0)
768		return phase;
769	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
770	<	if (p != phase)
771	<	return p;
772	<	QNode node = new QNode(this, phase, true, true, nanos);
773	<	p = internalAwaitAdvance(phase, node);
774	<	if (node.wasInterrupted)
775	<	throw new InterruptedException();
776	<	else if (p == phase)
777	<	throw new TimeoutException();
648	<	else
649	<	return p;
769	>	if (p == phase) {
770	>	QNode node = new QNode(this, phase, true, true, nanos);
771	>	p = root.internalAwaitAdvance(phase, node);
772	>	if (node.wasInterrupted)
773	>	throw new InterruptedException();
774	>	else if (p == phase)
775	>	throw new TimeoutException();
776	>	}
777	>	return p;
778		}
779
780		/**
781	<	* Forces this barrier to enter termination state. Counts of
782	<	* arrived and registered parties are unaffected. If this phaser
783	<	* has a parent, it too is terminated. This method may be useful
784	<	* for coordinating recovery after one or more tasks encounter
781	>	* Forces this phaser to enter termination state.  Counts of
782	>	* registered parties are unaffected.  If this phaser is a member
783	>	* of a tiered set of phasers, then all of the phasers in the set
784	>	* are terminated.  If this phaser is already terminated, this
785	>	* method has no effect.  This method may be useful for
786	>	* coordinating recovery after one or more tasks encounter
787		* unexpected exceptions.
788		*/
789		public void forceTermination() {
790	<	Phaser r = root;    // force at root then reconcile
790	>	// Only need to change root state
791	>	final Phaser root = this.root;
792		long s;
793	<	while ((s = r.state) >= 0)
794	<	UNSAFE.compareAndSwapLong(r, stateOffset, s, s | TERMINATION_PHASE);
795	<	reconcileState();
796	<	releaseWaiters(0); // signal all threads
797	<	releaseWaiters(1);
793	>	while ((s = root.state) >= 0) {
794	>	if (UNSAFE.compareAndSwapLong(root, stateOffset,
795	>	s, s | TERMINATION_BIT)) {
796	>	// signal all threads
797	>	releaseWaiters(0); // Waiters on evenQ
798	>	releaseWaiters(1); // Waiters on oddQ
799	>	return;
800	>	}
801	>	}
802		}
803
804		/**
805		* Returns the current phase number. The maximum phase number is
806		* {@code Integer.MAX_VALUE}, after which it restarts at
807	<	* zero. Upon termination, the phase number is negative.
807	>	* zero. Upon termination, the phase number is negative,
808	>	* in which case the prevailing phase prior to termination
809	>	* may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
810		*
811		* @return the phase number, or a negative value if terminated
812		*/
813		public final int getPhase() {
814	<	return (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
814	>	return (int)(root.state >>> PHASE_SHIFT);
815		}
816
817		/**
818	<	* Returns the number of parties registered at this barrier.
818	>	* Returns the number of parties registered at this phaser.
819		*
820		* @return the number of parties
821		*/
822		public int getRegisteredParties() {
823	<	return partiesOf(parent==null? state : reconcileState());
823	>	return partiesOf(state);
824		}
825
826		/**
827		* Returns the number of registered parties that have arrived at
828	<	* the current phase of this barrier.
828	>	* the current phase of this phaser. If this phaser has terminated,
829	>	* the returned value is meaningless and arbitrary.
830		*
831		* @return the number of arrived parties
832		*/
833		public int getArrivedParties() {
834	<	return arrivedOf(parent==null? state : reconcileState());
834	>	return arrivedOf(reconcileState());
835		}
836
837		/**
838		* Returns the number of registered parties that have not yet
839	<	* arrived at the current phase of this barrier.
839	>	* arrived at the current phase of this phaser. If this phaser has
840	>	* terminated, the returned value is meaningless and arbitrary.
841		*
842		* @return the number of unarrived parties
843		*/
844		public int getUnarrivedParties() {
845	<	return unarrivedOf(parent==null? state : reconcileState());
845	>	return unarrivedOf(reconcileState());
846		}
847
848		/**
#	Line 726 | Line 865 | public class Phaser {
865		}
866
867		/**
868	<	* Returns {@code true} if this barrier has been terminated.
868	>	* Returns {@code true} if this phaser has been terminated.
869		*
870	<	* @return {@code true} if this barrier has been terminated
870	>	* @return {@code true} if this phaser has been terminated
871		*/
872		public boolean isTerminated() {
873	<	return (parent == null? state : reconcileState()) < 0;
873	>	return root.state < 0L;
874		}
875
876		/**
877		* Overridable method to perform an action upon impending phase
878		* advance, and to control termination. This method is invoked
879	<	* upon arrival of the party tripping the barrier (when all other
879	>	* upon arrival of the party advancing this phaser (when all other
880		* waiting parties are dormant).  If this method returns {@code
881	<	* true}, then, rather than advance the phase number, this barrier
882	<	* will be set to a final termination state, and subsequent calls
883	<	* to {@link #isTerminated} will return true. Any (unchecked)
884	<	* Exception or Error thrown by an invocation of this method is
885	<	* propagated to the party attempting to trip the barrier, in
886	<	* which case no advance occurs.
881	>	* true}, this phaser will be set to a final termination state
882	>	* upon advance, and subsequent calls to {@link #isTerminated}
883	>	* will return true. Any (unchecked) Exception or Error thrown by
884	>	* an invocation of this method is propagated to the party
885	>	* attempting to advance this phaser, in which case no advance
886	>	* occurs.
887		*
888		* <p>The arguments to this method provide the state of the phaser
889		* prevailing for the current transition.  The effects of invoking
890	<	* arrival, registration, and waiting methods on this Phaser from
890	>	* arrival, registration, and waiting methods on this phaser from
891		* within {@code onAdvance} are unspecified and should not be
892		* relied on.
893		*
894	<	* <p>If this Phaser is a member of a tiered set of Phasers, then
895	<	* {@code onAdvance} is invoked only for its root Phaser on each
894	>	* <p>If this phaser is a member of a tiered set of phasers, then
895	>	* {@code onAdvance} is invoked only for its root phaser on each
896		* advance.
897		*
898	<	* <p>The default version returns {@code true} when the number of
899	<	* registered parties is zero. Normally, overrides that arrange
900	<	* termination for other reasons should also preserve this
901	<	* property.
898	>	* <p>To support the most common use cases, the default
899	>	* implementation of this method returns {@code true} when the
900	>	* number of registered parties has become zero as the result of a
901	>	* party invoking {@code arriveAndDeregister}.  You can disable
902	>	* this behavior, thus enabling continuation upon future
903	>	* registrations, by overriding this method to always return
904	>	* {@code false}:
905	>	*
906	>	* <pre> {@code
907	>	* Phaser phaser = new Phaser() {
908	>	*   protected boolean onAdvance(int phase, int parties) { return false; }
909	>	* }}</pre>
910		*
911	<	* @param phase the phase number on entering the barrier
911	>	* @param phase the current phase number on entry to this method,
912	>	* before this phaser is advanced
913		* @param registeredParties the current number of registered parties
914	<	* @return {@code true} if this barrier should terminate
914	>	* @return {@code true} if this phaser should terminate
915		*/
916		protected boolean onAdvance(int phase, int registeredParties) {
917	<	return registeredParties <= 0;
917	>	return registeredParties == 0;
918		}
919
920		/**
#	Line 776 | Line 924 | public class Phaser {
924		* followed by the number of registered parties, and {@code
925		* "arrived = "} followed by the number of arrived parties.
926		*
927	<	* @return a string identifying this barrier, as well as its state
927	>	* @return a string identifying this phaser, as well as its state
928		*/
929		public String toString() {
930		return stateToString(reconcileState());
#	Line 795 | Line 943 | public class Phaser {
943		// Waiting mechanics
944
945		/**
946	<	* Removes and signals threads from queue for phase
946	>	* Removes and signals threads from queue for phase.
947		*/
948		private void releaseWaiters(int phase) {
949	<	AtomicReference<QNode> head = queueFor(phase);
950	<	QNode q;
951	<	int p;
949	>	QNode q;   // first element of queue
950	>	Thread t;  // its thread
951	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
952		while ((q = head.get()) != null &&
953	<	((p = q.phase) == phase ||
954	<	(int)(root.state >>> PHASE_SHIFT) != p)) {
955	<	if (head.compareAndSet(q, q.next))
956	<	q.signal();
953	>	q.phase != (int)(root.state >>> PHASE_SHIFT)) {
954	>	if (head.compareAndSet(q, q.next) &&
955	>	(t = q.thread) != null) {
956	>	q.thread = null;
957	>	LockSupport.unpark(t);
958	>	}
959		}
960		}
961
962		/**
963	<	* Tries to enqueue given node in the appropriate wait queue.
963	>	* Variant of releaseWaiters that additionally tries to remove any
964	>	* nodes no longer waiting for advance due to timeout or
965	>	* interrupt. Currently, nodes are removed only if they are at
966	>	* head of queue, which suffices to reduce memory footprint in
967	>	* most usages.
968		*
969	<	* @return true if successful
969	>	* @return current phase on exit
970		*/
971	<	private boolean tryEnqueue(int phase, QNode node) {
972	<	releaseWaiters(phase-1); // ensure old queue clean
973	<	AtomicReference<QNode> head = queueFor(phase);
974	<	QNode q = head.get();
975	<	return ((q == null || q.phase == phase) &&
976	<	(int)(root.state >>> PHASE_SHIFT) == phase &&
977	<	head.compareAndSet(node.next = q, node));
971	>	private int abortWait(int phase) {
972	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
973	>	for (;;) {
974	>	Thread t;
975	>	QNode q = head.get();
976	>	int p = (int)(root.state >>> PHASE_SHIFT);
977	>	if (q == null || ((t = q.thread) != null && q.phase == p))
978	>	return p;
979	>	if (head.compareAndSet(q, q.next) && t != null) {
980	>	q.thread = null;
981	>	LockSupport.unpark(t);
982	>	}
983	>	}
984		}
985
986		/** The number of CPUs, for spin control */
#	Line 834 | Line 994 | public class Phaser {
994		* avoid it when threads regularly arrive: When a thread in
995		* internalAwaitAdvance notices another arrival before blocking,
996		* and there appear to be enough CPUs available, it spins
997	<	* SPINS_PER_ARRIVAL more times before continuing to try to
998	<	* block. The value trades off good-citizenship vs big unnecessary
839	<	* slowdowns.
997	>	* SPINS_PER_ARRIVAL more times before blocking. The value trades
998	>	* off good-citizenship vs big unnecessary slowdowns.
999		*/
1000	<	static final int SPINS_PER_ARRIVAL = NCPU < 2? 1 : 1 << 8;
1000	>	static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
1001
1002		/**
1003		* Possibly blocks and waits for phase to advance unless aborted.
1004	+	* Call only on root phaser.
1005		*
1006		* @param phase current phase
1007		* @param node if non-null, the wait node to track interrupt and timeout;
#	Line 849 | Line 1009 | public class Phaser {
1009		* @return current phase
1010		*/
1011		private int internalAwaitAdvance(int phase, QNode node) {
1012	<	Phaser current = this;       // to eventually wait at root if tiered
1013	<	boolean queued = false;      // true when node is enqueued
1014	<	int lastUnarrived = -1;      // to increase spins upon change
1012	>	// assert root == this;
1013	>	releaseWaiters(phase-1);          // ensure old queue clean
1014	>	boolean queued = false;           // true when node is enqueued
1015	>	int lastUnarrived = 0;            // to increase spins upon change
1016		int spins = SPINS_PER_ARRIVAL;
1017	<	for (;;) {
1018	<	int p, unarrived;
1019	<	Phaser par;
1020	<	long s = current.state;
1021	<	if ((p = (int)(s >>> PHASE_SHIFT)) != phase) {
1022	<	if (node != null)
1023	<	node.onRelease();
863	<	releaseWaiters(phase);
864	<	return p;
865	<	}
866	<	else if ((unarrived = (int)(s & UNARRIVED_MASK)) != lastUnarrived) {
867	<	if ((lastUnarrived = unarrived) < NCPU)
1017	>	long s;
1018	>	int p;
1019	>	while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
1020	>	if (node == null) {           // spinning in noninterruptible mode
1021	>	int unarrived = (int)s & UNARRIVED_MASK;
1022	>	if (unarrived != lastUnarrived &&
1023	>	(lastUnarrived = unarrived) < NCPU)
1024		spins += SPINS_PER_ARRIVAL;
1025	+	boolean interrupted = Thread.interrupted();
1026	+	if (interrupted || --spins < 0) { // need node to record intr
1027	+	node = new QNode(this, phase, false, false, 0L);
1028	+	node.wasInterrupted = interrupted;
1029	+	}
1030		}
1031	<	else if (unarrived == 0 && (par = current.parent) != null) {
1032	<	current = par;       // if all arrived, use parent
1033	<	par = par.parent;
1034	<	lastUnarrived = -1;
1035	<	}
1036	<	else if (spins > 0)
1037	<	--spins;
1038	<	else if (node == null)   // must be noninterruptible
878	<	node = new QNode(this, phase, false, false, 0L);
879	<	else if (node.isReleasable()) {
880	<	if ((int)(reconcileState() >>> PHASE_SHIFT) == phase)
881	<	return phase;    // aborted
1031	>	else if (node.isReleasable()) // done or aborted
1032	>	break;
1033	>	else if (!queued) {           // push onto queue
1034	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
1035	>	QNode q = node.next = head.get();
1036	>	if ((q == null || q.phase == phase) &&
1037	>	(int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
1038	>	queued = head.compareAndSet(q, node);
1039		}
883	–	else if (!queued)
884	–	queued = tryEnqueue(phase, node);
1040		else {
1041		try {
1042		ForkJoinPool.managedBlock(node);
#	Line 890 | Line 1045 | public class Phaser {
1045		}
1046		}
1047		}
1048	+
1049	+	if (node != null) {
1050	+	if (node.thread != null)
1051	+	node.thread = null;       // avoid need for unpark()
1052	+	if (node.wasInterrupted && !node.interruptible)
1053	+	Thread.currentThread().interrupt();
1054	+	if (p == phase && (p = (int)(state >>> PHASE_SHIFT)) == phase)
1055	+	return abortWait(phase); // possibly clean up on abort
1056	+	}
1057	+	releaseWaiters(phase);
1058	+	return p;
1059		}
1060
1061		/**
#	Line 913 | Line 1079 | public class Phaser {
1079		this.interruptible = interruptible;
1080		this.nanos = nanos;
1081		this.timed = timed;
1082	<	this.lastTime = timed? System.nanoTime() : 0L;
1082	>	this.lastTime = timed ? System.nanoTime() : 0L;
1083		thread = Thread.currentThread();
1084		}
1085
1086		public boolean isReleasable() {
1087	<	Thread t = thread;
1088	<	if (t != null) {
1089	<	if (phaser.getPhase() != phase)
924	<	t = null;
925	<	else {
926	<	if (Thread.interrupted())
927	<	wasInterrupted = true;
928	<	if (interruptible && wasInterrupted)
929	<	t = null;
930	<	else if (timed) {
931	<	if (nanos > 0) {
932	<	long now = System.nanoTime();
933	<	nanos -= now - lastTime;
934	<	lastTime = now;
935	<	}
936	<	if (nanos <= 0)
937	<	t = null;
938	<	}
939	<	}
940	<	if (t != null)
941	<	return false;
1087	>	if (thread == null)
1088	>	return true;
1089	>	if (phaser.getPhase() != phase) {
1090		thread = null;
1091	+	return true;
1092	+	}
1093	+	if (Thread.interrupted())
1094	+	wasInterrupted = true;
1095	+	if (wasInterrupted && interruptible) {
1096	+	thread = null;
1097	+	return true;
1098	+	}
1099	+	if (timed) {
1100	+	if (nanos > 0L) {
1101	+	long now = System.nanoTime();
1102	+	nanos -= now - lastTime;
1103	+	lastTime = now;
1104	+	}
1105	+	if (nanos <= 0L) {
1106	+	thread = null;
1107	+	return true;
1108	+	}
1109		}
1110	<	return true;
1110	>	return false;
1111		}
1112
1113		public boolean block() {
#	Line 953 | Line 1119 | public class Phaser {
1119		LockSupport.parkNanos(this, nanos);
1120		return isReleasable();
1121		}
956	–
957	–	void signal() {
958	–	Thread t = thread;
959	–	if (t != null) {
960	–	thread = null;
961	–	LockSupport.unpark(t);
962	–	}
963	–	}
964	–
965	–	void onRelease() { // actions upon return from internalAwaitAdvance
966	–	if (!interruptible && wasInterrupted)
967	–	Thread.currentThread().interrupt();
968	–	if (thread != null)
969	–	thread = null;
970	–	}
971	–
1122		}
1123
1124		// Unsafe mechanics
1125
1126	<	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1127	<	private static final long stateOffset =
1128	<	objectFieldOffset("state", Phaser.class);
979	<
980	<	private static long objectFieldOffset(String field, Class<?> klazz) {
1126	>	private static final sun.misc.Unsafe UNSAFE;
1127	>	private static final long stateOffset;
1128	>	static {
1129		try {
1130	<	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1131	<	} catch (NoSuchFieldException e) {
1132	<	// Convert Exception to corresponding Error
1133	<	NoSuchFieldError error = new NoSuchFieldError(field);
1134	<	error.initCause(e);
1135	<	throw error;
1130	>	UNSAFE = getUnsafe();
1131	>	Class<?> k = Phaser.class;
1132	>	stateOffset = UNSAFE.objectFieldOffset
1133	>	(k.getDeclaredField("state"));
1134	>	} catch (Exception e) {
1135	>	throw new Error(e);
1136		}
1137		}
1138

Diff Legend

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