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 1.53 by jsr166, Sat Nov 13 05:59:25 2010 UTC vs.
Revision 1.75 by dl, Wed Sep 21 12:30:39 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  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 274 | 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
278	<	* 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 293 | 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
301	<	* ONE_ARRIVAL (for arrive) or
302	<	* ONE_ARRIVAL|ONE_PARTY (for arriveAndDeregister)
346	>	* @param deregister false for arrive, true for arriveAndDeregister
347		*/
348	<	private int doArrive(long adj) {
349	<	long s;
350	<	int phase, unarrived;
351	<	while ((phase = (int)((s = state) >>> PHASE_SHIFT)) >= 0) {
352	<	if ((unarrived = (int)(s & UNARRIVED_MASK)) != 0) {
353	<	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s -= adj)) {
354	<	if (unarrived == 1) {
355	<	Phaser par;
356	<	long p = s & PARTIES_MASK; // unshifted parties field
357	<	long lu = p >>> PARTIES_SHIFT;
358	<	int u = (int)lu;
359	<	int nextPhase = (phase + 1) & MAX_PHASE;
360	<	long next = ((long)nextPhase << PHASE_SHIFT) | p | lu;
361	<	if ((par = parent) == null) {
362	<	UNSAFE.compareAndSwapLong
363	<	(this, stateOffset, s, onAdvance(phase, u)?
364	<	next | TERMINATION_PHASE : next);
365	<	releaseWaiters(phase);
366	<	}
367	<	else {
368	<	par.doArrive(u == 0?
369	<	ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL);
370	<	if ((int)(par.state >>> PHASE_SHIFT) != nextPhase ||
371	<	((int)(state >>> PHASE_SHIFT) != nextPhase &&
372	<	!UNSAFE.compareAndSwapLong(this, stateOffset,
373	<	s, next)))
374	<	reconcileState();
331	<	}
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 = (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 (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	>	long n = s & PARTIES_MASK;  // base of next state
364	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
365	>	if (unarrived == 0) {
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	>	n |= (long)((phase + 1) & MAX_PHASE) << PHASE_SHIFT;
374	>	UNSAFE.compareAndSwapLong(this, stateOffset, s, n);
375		}
376	<	break;
376	>	else if (nextUnarrived == 0) { // propagate deregistration
377	>	phase = parent.doArrive(true);
378	>	UNSAFE.compareAndSwapLong(this, stateOffset,
379	>	s, s | EMPTY);
380	>	}
381	>	else
382	>	phase = parent.doArrive(false);
383	>	releaseWaiters(phase);
384		}
385	+	return phase;
386		}
336	–	else if (state == s && reconcileState() == s) // recheck
337	–	throw new IllegalStateException(badArrive());
387		}
339	–	return phase;
340	–	}
341	–
342	–	/**
343	–	* Returns message string for bounds exceptions on arrival.
344	–	* Declared out of-line from doArrive to reduce string op bulk.
345	–	*/
346	–	private String badArrive() {
347	–	return ("Attempted arrival of unregistered party for " +
348	–	this.toString());
388		}
389
390		/**
391		* Implementation of register, bulkRegister
392		*
393	<	* @param registrations number to add to both parties and unarrived fields
393	>	* @param registrations number to add to both parties and
394	>	* unarrived fields. Must be greater than zero.
395		*/
396		private int doRegister(int registrations) {
397	<	long adj = (long)registrations; // adjustment to state
398	<	adj |= adj << PARTIES_SHIFT;
399	<	Phaser par = parent;
360	<	long s;
397	>	// adjustment to state
398	>	long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
399	>	final Phaser parent = this.parent;
400		int phase;
401	<	while ((phase = (int)((s = (par == null? state : reconcileState()))
402	<	>>> PHASE_SHIFT)) >= 0) {
403	<	int parties = ((int)(s & PARTIES_MASK)) >>> PARTIES_SHIFT;
404	<	if (parties != 0 && (s & UNARRIVED_MASK) == 0)
405	<	internalAwaitAdvance(phase, null); // wait for onAdvance
406	<	else if (parties + registrations > MAX_COUNT)
407	<	throw new IllegalStateException(badRegister());
408	<	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s + adj))
401	>	for (;;) {
402	>	long s = (parent == null) ? state : reconcileState();
403	>	int counts = (int)s;
404	>	int parties = counts >>> PARTIES_SHIFT;
405	>	int unarrived = counts & UNARRIVED_MASK;
406	>	if (registrations > MAX_PARTIES - parties)
407	>	throw new IllegalStateException(badRegister(s));
408	>	else if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
409		break;
410	+	else if (counts != EMPTY) {             // not 1st registration
411	+	if (parent == null || reconcileState() == s) {
412	+	if (unarrived == 0)             // wait out advance
413	+	root.internalAwaitAdvance(phase, null);
414	+	else if (UNSAFE.compareAndSwapLong(this, stateOffset,
415	+	s, s + adj))
416	+	break;
417	+	}
418	+	}
419	+	else if (parent == null) {              // 1st root registration
420	+	long next = ((long)phase << PHASE_SHIFT) | adj;
421	+	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
422	+	break;
423	+	}
424	+	else {
425	+	synchronized (this) {               // 1st sub registration
426	+	if (state == s) {               // recheck under lock
427	+	parent.doRegister(1);
428	+	do {                        // force current phase
429	+	phase = (int)(root.state >>> PHASE_SHIFT);
430	+	// assert phase < 0 || (int)state == EMPTY;
431	+	} while (!UNSAFE.compareAndSwapLong
432	+	(this, stateOffset, state,
433	+	((long)phase << PHASE_SHIFT) | adj));
434	+	break;
435	+	}
436	+	}
437	+	}
438		}
439		return phase;
440		}
441
442		/**
443	<	* Returns message string for bounds exceptions on registration
444	<	*/
445	<	private String badRegister() {
446	<	return ("Attempt to register more than " + MAX_COUNT + " parties for "+
447	<	this.toString());
448	<	}
449	<
450	<	/**
451	<	* Recursively resolves lagged phase propagation from root if
452	<	* necessary.
443	>	* Resolves lagged phase propagation from root if necessary.
444	>	* Reconciliation normally occurs when root has advanced but
445	>	* subphasers have not yet done so, in which case they must finish
446	>	* their own advance by setting unarrived to parties (or if
447	>	* parties is zero, resetting to unregistered EMPTY state).
448	>	* However, this method may also be called when "floating"
449	>	* subphasers with possibly some unarrived parties are merely
450	>	* catching up to current phase, in which case counts are
451	>	* unaffected.
452	>	*
453	>	* @return reconciled state
454		*/
455		private long reconcileState() {
456	<	Phaser par = parent;
457	<	if (par == null)
458	<	return state;
459	<	Phaser rt = root;
460	<	long s;
461	<	int phase, rPhase;
462	<	while ((phase = (int)((s = state) >>> PHASE_SHIFT)) >= 0 &&
463	<	(rPhase = (int)(rt.state >>> PHASE_SHIFT)) != phase) {
464	<	if (rPhase < 0 || (s & UNARRIVED_MASK) == 0) {
465	<	long ps = par.parent == null? par.state : par.reconcileState();
466	<	int pPhase = (int)(ps >>> PHASE_SHIFT);
467	<	if (pPhase < 0 || pPhase == ((phase + 1) & MAX_PHASE)) {
468	<	if (state != s)
469	<	continue;
470	<	long p = s & PARTIES_MASK;
403	<	long next = ((((long) pPhase) << PHASE_SHIFT) |
404	<	(p >>> PARTIES_SHIFT) | p);
405	<	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
406	<	return next;
407	<	}
408	<	}
409	<	if (state == s)
410	<	releaseWaiters(phase); // help release others
456	>	final Phaser root = this.root;
457	>	long s = state;
458	>	if (root != this) {
459	>	int phase, u, p;
460	>	// CAS root phase with current parties; possibly trip unarrived
461	>	while ((phase = (int)(root.state >>> PHASE_SHIFT)) !=
462	>	(int)(s >>> PHASE_SHIFT) &&
463	>	!UNSAFE.compareAndSwapLong
464	>	(this, stateOffset, s,
465	>	s = (((long)phase << PHASE_SHIFT) |
466	>	(s & PARTIES_MASK) |
467	>	((p = (int)s >>> PARTIES_SHIFT) == 0 ? EMPTY :
468	>	((u = (int)s & UNARRIVED_MASK) == 0 && phase >= 0) ?
469	>	p : u))))
470	>	s = state;
471		}
472		return s;
473		}
474
475		/**
476	<	* Creates a new phaser without any initially registered parties,
477	<	* initial phase number 0, and no parent. Any thread using this
476	>	* Creates a new phaser with no initially registered parties, no
477	>	* parent, and initial phase number 0. Any thread using this
478		* phaser will need to first register for it.
479		*/
480		public Phaser() {
#	Line 423 | Line 483 | public class Phaser {
483
484		/**
485		* Creates a new phaser with the given number of registered
486	<	* unarrived parties, initial phase number 0, and no parent.
486	>	* unarrived parties, no parent, and initial phase number 0.
487		*
488	<	* @param parties the number of parties required to trip barrier
488	>	* @param parties the number of parties required to advance to the
489	>	* next phase
490		* @throws IllegalArgumentException if parties less than zero
491		* or greater than the maximum number of parties supported
492		*/
#	Line 434 | Line 495 | public class Phaser {
495		}
496
497		/**
498	<	* 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.
498	>	* Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
499		*
500		* @param parent the parent phaser
501		*/
#	Line 447 | Line 505 | public class Phaser {
505
506		/**
507		* Creates a new phaser with the given parent and number of
508	<	* registered unarrived parties. If parent is non-null, this phaser
509	<	* is registered with the parent and its initial phase number is
510	<	* the same as that of parent phaser.
508	>	* registered unarrived parties.  When the given parent is non-null
509	>	* and the given number of parties is greater than zero, this
510	>	* child phaser is registered with its parent.
511		*
512		* @param parent the parent phaser
513	<	* @param parties the number of parties required to trip barrier
513	>	* @param parties the number of parties required to advance to the
514	>	* next phase
515		* @throws IllegalArgumentException if parties less than zero
516		* or greater than the maximum number of parties supported
517		*/
518		public Phaser(Phaser parent, int parties) {
519	<	if (parties < 0 || parties > MAX_COUNT)
519	>	if (parties >>> PARTIES_SHIFT != 0)
520		throw new IllegalArgumentException("Illegal number of parties");
521	<	int phase;
521	>	int phase = 0;
522		this.parent = parent;
523		if (parent != null) {
524	<	Phaser r = parent.root;
525	<	this.root = r;
526	<	this.evenQ = r.evenQ;
527	<	this.oddQ = r.oddQ;
528	<	phase = parent.register();
524	>	final Phaser root = parent.root;
525	>	this.root = root;
526	>	this.evenQ = root.evenQ;
527	>	this.oddQ = root.oddQ;
528	>	if (parties != 0)
529	>	phase = parent.doRegister(1);
530		}
531		else {
532		this.root = this;
533		this.evenQ = new AtomicReference<QNode>();
534		this.oddQ = new AtomicReference<QNode>();
475	–	phase = 0;
535		}
536	<	long p = (long)parties;
537	<	this.state = (((long) phase) << PHASE_SHIFT) | p | (p << PARTIES_SHIFT);
536	>	this.state = (parties == 0) ? (long)EMPTY :
537	>	((long)phase << PHASE_SHIFT) |
538	>	((long)parties << PARTIES_SHIFT) |
539	>	((long)parties);
540		}
541
542		/**
543	<	* Adds a new unarrived party to this phaser.
544	<	* If an ongoing invocation of {@link #onAdvance} is in progress,
545	<	* this method may wait until its completion before registering.
546	<	*
547	<	* @return the arrival phase number to which this registration applied
543	>	* Adds a new unarrived party to this phaser.  If an ongoing
544	>	* invocation of {@link #onAdvance} is in progress, this method
545	>	* may await its completion before returning.  If this phaser has
546	>	* a parent, and this phaser previously had no registered parties,
547	>	* this child phaser is also registered with its parent. If
548	>	* this phaser is terminated, the attempt to register has
549	>	* no effect, and a negative value is returned.
550	>	*
551	>	* @return the arrival phase number to which this registration
552	>	* applied.  If this value is negative, then this phaser has
553	>	* terminated, in which case registration has no effect.
554		* @throws IllegalStateException if attempting to register more
555		* than the maximum supported number of parties
556		*/
#	Line 494 | Line 561 | public class Phaser {
561		/**
562		* Adds the given number of new unarrived parties to this phaser.
563		* If an ongoing invocation of {@link #onAdvance} is in progress,
564	<	* this method may wait until its completion before registering.
565	<	*
566	<	* @param parties the number of additional parties required to trip barrier
567	<	* @return the arrival phase number to which this registration applied
564	>	* this method may await its completion before returning.  If this
565	>	* phaser has a parent, and the given number of parties is greater
566	>	* than zero, and this phaser previously had no registered
567	>	* parties, this child phaser is also registered with its parent.
568	>	* If this phaser is terminated, the attempt to register has no
569	>	* effect, and a negative value is returned.
570	>	*
571	>	* @param parties the number of additional parties required to
572	>	* advance to the next phase
573	>	* @return the arrival phase number to which this registration
574	>	* applied.  If this value is negative, then this phaser has
575	>	* terminated, in which case registration has no effect.
576		* @throws IllegalStateException if attempting to register more
577		* than the maximum supported number of parties
578		* @throws IllegalArgumentException if {@code parties < 0}
#	Line 505 | Line 580 | public class Phaser {
580		public int bulkRegister(int parties) {
581		if (parties < 0)
582		throw new IllegalArgumentException();
508	–	if (parties > MAX_COUNT)
509	–	throw new IllegalStateException(badRegister());
583		if (parties == 0)
584		return getPhase();
585		return doRegister(parties);
586		}
587
588		/**
589	<	* Arrives at the barrier, but does not wait for others.  (You can
590	<	* in turn wait for others via {@link #awaitAdvance}).  It is an
591	<	* unenforced usage error for an unregistered party to invoke this
592	<	* method.
589	>	* Arrives at this phaser, without waiting for others to arrive.
590	>	*
591	>	* <p>It is a usage error for an unregistered party to invoke this
592	>	* method.  However, this error may result in an {@code
593	>	* IllegalStateException} only upon some subsequent operation on
594	>	* this phaser, if ever.
595		*
596		* @return the arrival phase number, or a negative value if terminated
597		* @throws IllegalStateException if not terminated and the number
598		* of unarrived parties would become negative
599		*/
600		public int arrive() {
601	<	return doArrive(ONE_ARRIVAL);
601	>	return doArrive(false);
602		}
603
604		/**
605	<	* Arrives at the barrier and deregisters from it without waiting
606	<	* for others. Deregistration reduces the number of parties
607	<	* required to trip the barrier in future phases.  If this phaser
605	>	* Arrives at this phaser and deregisters from it without waiting
606	>	* for others to arrive. Deregistration reduces the number of
607	>	* parties required to advance in future phases.  If this phaser
608		* has a parent, and deregistration causes this phaser to have
609	<	* zero parties, this phaser also arrives at and is deregistered
610	<	* from its parent.  It is an unenforced usage error for an
611	<	* unregistered party to invoke this method.
609	>	* zero parties, this phaser is also deregistered from its parent.
610	>	*
611	>	* <p>It is a usage error for an unregistered party to invoke this
612	>	* method.  However, this error may result in an {@code
613	>	* IllegalStateException} only upon some subsequent operation on
614	>	* this phaser, if ever.
615		*
616		* @return the arrival phase number, or a negative value if terminated
617		* @throws IllegalStateException if not terminated and the number
618		* of registered or unarrived parties would become negative
619		*/
620		public int arriveAndDeregister() {
621	<	return doArrive(ONE_ARRIVAL|ONE_PARTY);
621	>	return doArrive(true);
622		}
623
624		/**
625	<	* Arrives at the barrier and awaits others. Equivalent in effect
625	>	* Arrives at this phaser and awaits others. Equivalent in effect
626		* to {@code awaitAdvance(arrive())}.  If you need to await with
627		* interruption or timeout, you can arrange this with an analogous
628		* construction using one of the other forms of the {@code
629		* awaitAdvance} method.  If instead you need to deregister upon
630	<	* arrival, use {@link #arriveAndDeregister}. It is an unenforced
631	<	* usage error for an unregistered party to invoke this method.
630	>	* arrival, use {@code awaitAdvance(arriveAndDeregister())}.
631	>	*
632	>	* <p>It is a usage error for an unregistered party to invoke this
633	>	* method.  However, this error may result in an {@code
634	>	* IllegalStateException} only upon some subsequent operation on
635	>	* this phaser, if ever.
636		*
637	<	* @return the arrival phase number, or a negative number if terminated
637	>	* @return the arrival phase number, or the (negative)
638	>	* {@linkplain #getPhase() current phase} if terminated
639		* @throws IllegalStateException if not terminated and the number
640		* of unarrived parties would become negative
641		*/
642		public int arriveAndAwaitAdvance() {
643	<	return awaitAdvance(arrive());
643	>	// Specialization of doArrive+awaitAdvance eliminating some reads/paths
644	>	final Phaser root = this.root;
645	>	for (;;) {
646	>	long s = (root == this) ? state : reconcileState();
647	>	int phase = (int)(s >>> PHASE_SHIFT);
648	>	int counts = (int)s;
649	>	int unarrived = (counts & UNARRIVED_MASK) - 1;
650	>	if (phase < 0)
651	>	return phase;
652	>	else if (counts == EMPTY || unarrived < 0) {
653	>	if (reconcileState() == s)
654	>	throw new IllegalStateException(badArrive(s));
655	>	}
656	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s,
657	>	s -= ONE_ARRIVAL)) {
658	>	if (unarrived != 0)
659	>	return root.internalAwaitAdvance(phase, null);
660	>	if (root != this)
661	>	return parent.arriveAndAwaitAdvance();
662	>	long n = s & PARTIES_MASK;  // base of next state
663	>	int nextUnarrived = (int)n >>> PARTIES_SHIFT;
664	>	if (onAdvance(phase, nextUnarrived))
665	>	n |= TERMINATION_BIT;
666	>	else if (nextUnarrived == 0)
667	>	n |= EMPTY;
668	>	else
669	>	n |= nextUnarrived;
670	>	int nextPhase = (phase + 1) & MAX_PHASE;
671	>	n |= (long)nextPhase << PHASE_SHIFT;
672	>	if (!UNSAFE.compareAndSwapLong(this, stateOffset, s, n))
673	>	return (int)(state >>> PHASE_SHIFT); // terminated
674	>	releaseWaiters(phase);
675	>	return nextPhase;
676	>	}
677	>	}
678		}
679
680		/**
681	<	* Awaits the phase of the barrier to advance from the given phase
682	<	* value, returning immediately if the current phase of the
683	<	* barrier is not equal to the given phase value or this barrier
567	<	* is terminated.
681	>	* Awaits the phase of this phaser to advance from the given phase
682	>	* value, returning immediately if the current phase is not equal
683	>	* to the given phase value or this phaser is terminated.
684		*
685		* @param phase an arrival phase number, or negative value if
686		* terminated; this argument is normally the value returned by a
687	<	* previous call to {@code arrive} or its variants
688	<	* @return the next arrival phase number, or a negative value
689	<	* if terminated or argument is negative
687	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
688	>	* @return the next arrival phase number, or the argument if it is
689	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
690	>	* if terminated
691		*/
692		public int awaitAdvance(int phase) {
693	+	final Phaser root = this.root;
694	+	long s = (root == this) ? state : reconcileState();
695	+	int p = (int)(s >>> PHASE_SHIFT);
696		if (phase < 0)
697		return phase;
698	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
699	<	if (p != phase)
700	<	return p;
581	<	return internalAwaitAdvance(phase, null);
698	>	if (p == phase)
699	>	return root.internalAwaitAdvance(phase, null);
700	>	return p;
701		}
702
703		/**
704	<	* Awaits the phase of the barrier to advance from the given phase
704	>	* Awaits the phase of this phaser to advance from the given phase
705		* value, throwing {@code InterruptedException} if interrupted
706	<	* while waiting, or returning immediately if the current phase of
707	<	* the barrier is not equal to the given phase value or this
708	<	* barrier is terminated.
706	>	* while waiting, or returning immediately if the current phase is
707	>	* not equal to the given phase value or this phaser is
708	>	* terminated.
709		*
710		* @param phase an arrival phase number, or negative value if
711		* terminated; this argument is normally the value returned by a
712	<	* previous call to {@code arrive} or its variants
713	<	* @return the next arrival phase number, or a negative value
714	<	* if terminated or argument is negative
712	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
713	>	* @return the next arrival phase number, or the argument if it is
714	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
715	>	* if terminated
716		* @throws InterruptedException if thread interrupted while waiting
717		*/
718		public int awaitAdvanceInterruptibly(int phase)
719		throws InterruptedException {
720	+	final Phaser root = this.root;
721	+	long s = (root == this) ? state : reconcileState();
722	+	int p = (int)(s >>> PHASE_SHIFT);
723		if (phase < 0)
724		return phase;
725	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
726	<	if (p != phase)
727	<	return p;
728	<	QNode node = new QNode(this, phase, true, false, 0L);
729	<	p = internalAwaitAdvance(phase, node);
730	<	if (node.wasInterrupted)
731	<	throw new InterruptedException();
609	<	else
610	<	return p;
725	>	if (p == phase) {
726	>	QNode node = new QNode(this, phase, true, false, 0L);
727	>	p = root.internalAwaitAdvance(phase, node);
728	>	if (node.wasInterrupted)
729	>	throw new InterruptedException();
730	>	}
731	>	return p;
732		}
733
734		/**
735	<	* Awaits the phase of the barrier to advance from the given phase
735	>	* Awaits the phase of this phaser to advance from the given phase
736		* value or the given timeout to elapse, throwing {@code
737		* InterruptedException} if interrupted while waiting, or
738	<	* returning immediately if the current phase of the barrier is
739	<	* not equal to the given phase value or this barrier is
619	<	* terminated.
738	>	* returning immediately if the current phase is not equal to the
739	>	* given phase value or this phaser is terminated.
740		*
741		* @param phase an arrival phase number, or negative value if
742		* terminated; this argument is normally the value returned by a
743	<	* previous call to {@code arrive} or its variants
743	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
744		* @param timeout how long to wait before giving up, in units of
745		*        {@code unit}
746		* @param unit a {@code TimeUnit} determining how to interpret the
747		*        {@code timeout} parameter
748	<	* @return the next arrival phase number, or a negative value
749	<	* if terminated or argument is negative
748	>	* @return the next arrival phase number, or the argument if it is
749	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
750	>	* if terminated
751		* @throws InterruptedException if thread interrupted while waiting
752		* @throws TimeoutException if timed out while waiting
753		*/
#	Line 634 | Line 755 | public class Phaser {
755		long timeout, TimeUnit unit)
756		throws InterruptedException, TimeoutException {
757		long nanos = unit.toNanos(timeout);
758	+	final Phaser root = this.root;
759	+	long s = (root == this) ? state : reconcileState();
760	+	int p = (int)(s >>> PHASE_SHIFT);
761		if (phase < 0)
762		return phase;
763	<	int p = (int)((parent==null? state : reconcileState()) >>> PHASE_SHIFT);
764	<	if (p != phase)
765	<	return p;
766	<	QNode node = new QNode(this, phase, true, true, nanos);
767	<	p = internalAwaitAdvance(phase, node);
768	<	if (node.wasInterrupted)
769	<	throw new InterruptedException();
770	<	else if (p == phase)
771	<	throw new TimeoutException();
648	<	else
649	<	return p;
763	>	if (p == phase) {
764	>	QNode node = new QNode(this, phase, true, true, nanos);
765	>	p = root.internalAwaitAdvance(phase, node);
766	>	if (node.wasInterrupted)
767	>	throw new InterruptedException();
768	>	else if (p == phase)
769	>	throw new TimeoutException();
770	>	}
771	>	return p;
772		}
773
774		/**
775	<	* Forces this barrier to enter termination state. Counts of
776	<	* arrived and registered parties are unaffected. If this phaser
777	<	* has a parent, it too is terminated. This method may be useful
778	<	* for coordinating recovery after one or more tasks encounter
775	>	* Forces this phaser to enter termination state.  Counts of
776	>	* registered parties are unaffected.  If this phaser is a member
777	>	* of a tiered set of phasers, then all of the phasers in the set
778	>	* are terminated.  If this phaser is already terminated, this
779	>	* method has no effect.  This method may be useful for
780	>	* coordinating recovery after one or more tasks encounter
781		* unexpected exceptions.
782		*/
783		public void forceTermination() {
784	<	Phaser r = root;    // force at root then reconcile
784	>	// Only need to change root state
785	>	final Phaser root = this.root;
786		long s;
787	<	while ((s = r.state) >= 0)
788	<	UNSAFE.compareAndSwapLong(r, stateOffset, s, s | TERMINATION_PHASE);
789	<	reconcileState();
790	<	releaseWaiters(0); // signal all threads
791	<	releaseWaiters(1);
787	>	while ((s = root.state) >= 0) {
788	>	if (UNSAFE.compareAndSwapLong(root, stateOffset,
789	>	s, s | TERMINATION_BIT)) {
790	>	// signal all threads
791	>	releaseWaiters(0);
792	>	releaseWaiters(1);
793	>	return;
794	>	}
795	>	}
796		}
797
798		/**
799		* Returns the current phase number. The maximum phase number is
800		* {@code Integer.MAX_VALUE}, after which it restarts at
801	<	* zero. Upon termination, the phase number is negative.
801	>	* zero. Upon termination, the phase number is negative,
802	>	* in which case the prevailing phase prior to termination
803	>	* may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
804		*
805		* @return the phase number, or a negative value if terminated
806		*/
807		public final int getPhase() {
808	<	return (int)((parent == null? state : reconcileState()) >>> PHASE_SHIFT);
808	>	return (int)(root.state >>> PHASE_SHIFT);
809		}
810
811		/**
812	<	* Returns the number of parties registered at this barrier.
812	>	* Returns the number of parties registered at this phaser.
813		*
814		* @return the number of parties
815		*/
816		public int getRegisteredParties() {
817	<	return partiesOf(parent == null? state : reconcileState());
817	>	return partiesOf(state);
818		}
819
820		/**
821		* Returns the number of registered parties that have arrived at
822	<	* the current phase of this barrier.
822	>	* the current phase of this phaser. If this phaser has terminated,
823	>	* the returned value is meaningless and arbitrary.
824		*
825		* @return the number of arrived parties
826		*/
827		public int getArrivedParties() {
828	<	return arrivedOf(parent == null? state : reconcileState());
828	>	return arrivedOf(reconcileState());
829		}
830
831		/**
832		* Returns the number of registered parties that have not yet
833	<	* arrived at the current phase of this barrier.
833	>	* arrived at the current phase of this phaser. If this phaser has
834	>	* terminated, the returned value is meaningless and arbitrary.
835		*
836		* @return the number of unarrived parties
837		*/
838		public int getUnarrivedParties() {
839	<	return unarrivedOf(parent == null? state : reconcileState());
839	>	return unarrivedOf(reconcileState());
840		}
841
842		/**
#	Line 726 | Line 859 | public class Phaser {
859		}
860
861		/**
862	<	* Returns {@code true} if this barrier has been terminated.
862	>	* Returns {@code true} if this phaser has been terminated.
863		*
864	<	* @return {@code true} if this barrier has been terminated
864	>	* @return {@code true} if this phaser has been terminated
865		*/
866		public boolean isTerminated() {
867	<	return (parent == null? state : reconcileState()) < 0;
867	>	return root.state < 0L;
868		}
869
870		/**
871		* Overridable method to perform an action upon impending phase
872		* advance, and to control termination. This method is invoked
873	<	* upon arrival of the party tripping the barrier (when all other
873	>	* upon arrival of the party advancing this phaser (when all other
874		* waiting parties are dormant).  If this method returns {@code
875	<	* true}, then, rather than advance the phase number, this barrier
876	<	* will be set to a final termination state, and subsequent calls
877	<	* to {@link #isTerminated} will return true. Any (unchecked)
878	<	* Exception or Error thrown by an invocation of this method is
879	<	* propagated to the party attempting to trip the barrier, in
880	<	* which case no advance occurs.
875	>	* true}, this phaser will be set to a final termination state
876	>	* upon advance, and subsequent calls to {@link #isTerminated}
877	>	* will return true. Any (unchecked) Exception or Error thrown by
878	>	* an invocation of this method is propagated to the party
879	>	* attempting to advance this phaser, in which case no advance
880	>	* occurs.
881		*
882		* <p>The arguments to this method provide the state of the phaser
883		* prevailing for the current transition.  The effects of invoking
884	<	* arrival, registration, and waiting methods on this Phaser from
884	>	* arrival, registration, and waiting methods on this phaser from
885		* within {@code onAdvance} are unspecified and should not be
886		* relied on.
887		*
888	<	* <p>If this Phaser is a member of a tiered set of Phasers, then
889	<	* {@code onAdvance} is invoked only for its root Phaser on each
888	>	* <p>If this phaser is a member of a tiered set of phasers, then
889	>	* {@code onAdvance} is invoked only for its root phaser on each
890		* advance.
891		*
892	<	* <p>The default version returns {@code true} when the number of
893	<	* registered parties is zero. Normally, overrides that arrange
894	<	* termination for other reasons should also preserve this
895	<	* property.
892	>	* <p>To support the most common use cases, the default
893	>	* implementation of this method returns {@code true} when the
894	>	* number of registered parties has become zero as the result of a
895	>	* party invoking {@code arriveAndDeregister}.  You can disable
896	>	* this behavior, thus enabling continuation upon future
897	>	* registrations, by overriding this method to always return
898	>	* {@code false}:
899	>	*
900	>	* <pre> {@code
901	>	* Phaser phaser = new Phaser() {
902	>	*   protected boolean onAdvance(int phase, int parties) { return false; }
903	>	* }}</pre>
904		*
905	<	* @param phase the phase number on entering the barrier
905	>	* @param phase the current phase number on entry to this method,
906	>	* before this phaser is advanced
907		* @param registeredParties the current number of registered parties
908	<	* @return {@code true} if this barrier should terminate
908	>	* @return {@code true} if this phaser should terminate
909		*/
910		protected boolean onAdvance(int phase, int registeredParties) {
911	<	return registeredParties <= 0;
911	>	return registeredParties == 0;
912		}
913
914		/**
#	Line 776 | Line 918 | public class Phaser {
918		* followed by the number of registered parties, and {@code
919		* "arrived = "} followed by the number of arrived parties.
920		*
921	<	* @return a string identifying this barrier, as well as its state
921	>	* @return a string identifying this phaser, as well as its state
922		*/
923		public String toString() {
924	<	long s = reconcileState();
924	>	return stateToString(reconcileState());
925	>	}
926	>
927	>	/**
928	>	* Implementation of toString and string-based error messages
929	>	*/
930	>	private String stateToString(long s) {
931		return super.toString() +
932		"[phase = " + phaseOf(s) +
933		" parties = " + partiesOf(s) +
934		" arrived = " + arrivedOf(s) + "]";
935		}
936
937	+	// Waiting mechanics
938	+
939		/**
940	<	* Removes and signals threads from queue for phase
940	>	* Removes and signals threads from queue for phase.
941		*/
942		private void releaseWaiters(int phase) {
943	<	AtomicReference<QNode> head = queueFor(phase);
944	<	QNode q;
945	<	int p;
943	>	QNode q;   // first element of queue
944	>	Thread t;  // its thread
945	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
946		while ((q = head.get()) != null &&
947	<	((p = q.phase) == phase ||
948	<	(int)(root.state >>> PHASE_SHIFT) != p)) {
949	<	if (head.compareAndSet(q, q.next))
950	<	q.signal();
947	>	q.phase != (int)(root.state >>> PHASE_SHIFT)) {
948	>	if (head.compareAndSet(q, q.next) &&
949	>	(t = q.thread) != null) {
950	>	q.thread = null;
951	>	LockSupport.unpark(t);
952	>	}
953		}
954		}
955
956		/**
957	<	* Tries to enqueue given node in the appropriate wait queue.
958	<	*
959	<	* @return true if successful
960	<	*/
961	<	private boolean tryEnqueue(int phase, QNode node) {
962	<	releaseWaiters(phase-1); // ensure old queue clean
963	<	AtomicReference<QNode> head = queueFor(phase);
964	<	QNode q = head.get();
965	<	return ((q == null || q.phase == phase) &&
966	<	(int)(root.state >>> PHASE_SHIFT) == phase &&
967	<	head.compareAndSet(node.next = q, node));
957	>	* Variant of releaseWaiters that additionally tries to remove any
958	>	* nodes no longer waiting for advance due to timeout or
959	>	* interrupt. Currently, nodes are removed only if they are at
960	>	* head of queue, which suffices to reduce memory footprint in
961	>	* most usages.
962	>	*
963	>	* @return current phase on exit
964	>	*/
965	>	private int abortWait(int phase) {
966	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
967	>	for (;;) {
968	>	Thread t;
969	>	QNode q = head.get();
970	>	int p = (int)(root.state >>> PHASE_SHIFT);
971	>	if (q == null || ((t = q.thread) != null && q.phase == p))
972	>	return p;
973	>	if (head.compareAndSet(q, q.next) && t != null) {
974	>	q.thread = null;
975	>	LockSupport.unpark(t);
976	>	}
977	>	}
978		}
979
980		/** The number of CPUs, for spin control */
#	Line 826 | Line 988 | public class Phaser {
988		* avoid it when threads regularly arrive: When a thread in
989		* internalAwaitAdvance notices another arrival before blocking,
990		* and there appear to be enough CPUs available, it spins
991	<	* SPINS_PER_ARRIVAL more times before continuing to try to
992	<	* block. The value trades off good-citizenship vs big unnecessary
831	<	* slowdowns.
991	>	* SPINS_PER_ARRIVAL more times before blocking. The value trades
992	>	* off good-citizenship vs big unnecessary slowdowns.
993		*/
994	<	static final int SPINS_PER_ARRIVAL = NCPU < 2? 1 : 1 << 8;
994	>	static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
995
996		/**
997		* Possibly blocks and waits for phase to advance unless aborted.
998	+	* Call only from root node.
999		*
1000		* @param phase current phase
1001		* @param node if non-null, the wait node to track interrupt and timeout;
#	Line 841 | Line 1003 | public class Phaser {
1003		* @return current phase
1004		*/
1005		private int internalAwaitAdvance(int phase, QNode node) {
1006	<	Phaser current = this;       // to eventually wait at root if tiered
1007	<	Phaser par = parent;
1008	<	boolean queued = false;
1006	>	releaseWaiters(phase-1);          // ensure old queue clean
1007	>	boolean queued = false;           // true when node is enqueued
1008	>	int lastUnarrived = 0;            // to increase spins upon change
1009		int spins = SPINS_PER_ARRIVAL;
848	–	int lastUnarrived = -1;      // to increase spins upon change
1010		long s;
1011		int p;
1012	<	while ((p = (int)((s = current.state) >>> PHASE_SHIFT)) == phase) {
1013	<	int unarrived = (int)(s & UNARRIVED_MASK);
1014	<	if (unarrived != lastUnarrived) {
1015	<	if ((lastUnarrived = unarrived) < NCPU)
1012	>	while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
1013	>	if (node == null) {           // spinning in noninterruptible mode
1014	>	int unarrived = (int)s & UNARRIVED_MASK;
1015	>	if (unarrived != lastUnarrived &&
1016	>	(lastUnarrived = unarrived) < NCPU)
1017		spins += SPINS_PER_ARRIVAL;
1018	+	boolean interrupted = Thread.interrupted();
1019	+	if (interrupted || --spins < 0) { // need node to record intr
1020	+	node = new QNode(this, phase, false, false, 0L);
1021	+	node.wasInterrupted = interrupted;
1022	+	}
1023		}
1024	<	else if (unarrived == 0 && par != null) {
858	<	current = par;       // if all arrived, use parent
859	<	par = par.parent;
860	<	}
861	<	else if (spins > 0)
862	<	--spins;
863	<	else if (node == null)
864	<	node = new QNode(this, phase, false, false, 0L);
865	<	else if (node.isReleasable())
1024	>	else if (node.isReleasable()) // done or aborted
1025		break;
1026	<	else if (!queued)
1027	<	queued = tryEnqueue(phase, node);
1026	>	else if (!queued) {           // push onto queue
1027	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
1028	>	QNode q = node.next = head.get();
1029	>	if ((q == null || q.phase == phase) &&
1030	>	(int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
1031	>	queued = head.compareAndSet(q, node);
1032	>	}
1033		else {
1034		try {
1035		ForkJoinPool.managedBlock(node);
#	Line 874 | Line 1038 | public class Phaser {
1038		}
1039		}
1040		}
1041	+
1042		if (node != null) {
1043		if (node.thread != null)
1044	<	node.thread = null;
1045	<	if (!node.interruptible && node.wasInterrupted)
1044	>	node.thread = null;       // avoid need for unpark()
1045	>	if (node.wasInterrupted && !node.interruptible)
1046		Thread.currentThread().interrupt();
1047	+	if (p == phase && (p = (int)(state >>> PHASE_SHIFT)) == phase)
1048	+	return abortWait(phase); // possibly clean up on abort
1049		}
1050	<	if (p == phase)
884	<	p = (int)(reconcileState() >>> PHASE_SHIFT);
885	<	if (p != phase)
886	<	releaseWaiters(phase);
1050	>	releaseWaiters(phase);
1051		return p;
1052		}
1053
#	Line 908 | Line 1072 | public class Phaser {
1072		this.interruptible = interruptible;
1073		this.nanos = nanos;
1074		this.timed = timed;
1075	<	this.lastTime = timed? System.nanoTime() : 0L;
1075	>	this.lastTime = timed ? System.nanoTime() : 0L;
1076		thread = Thread.currentThread();
1077		}
1078
1079		public boolean isReleasable() {
1080	<	Thread t = thread;
1081	<	if (t != null) {
1082	<	if (phaser.getPhase() != phase)
919	<	t = null;
920	<	else {
921	<	if (Thread.interrupted())
922	<	wasInterrupted = true;
923	<	if (interruptible && wasInterrupted)
924	<	t = null;
925	<	else if (timed) {
926	<	if (nanos > 0) {
927	<	long now = System.nanoTime();
928	<	nanos -= now - lastTime;
929	<	lastTime = now;
930	<	}
931	<	if (nanos <= 0)
932	<	t = null;
933	<	}
934	<	}
935	<	if (t != null)
936	<	return false;
1080	>	if (thread == null)
1081	>	return true;
1082	>	if (phaser.getPhase() != phase) {
1083		thread = null;
1084	+	return true;
1085		}
1086	<	return true;
1086	>	if (Thread.interrupted())
1087	>	wasInterrupted = true;
1088	>	if (wasInterrupted && interruptible) {
1089	>	thread = null;
1090	>	return true;
1091	>	}
1092	>	if (timed) {
1093	>	if (nanos > 0L) {
1094	>	long now = System.nanoTime();
1095	>	nanos -= now - lastTime;
1096	>	lastTime = now;
1097	>	}
1098	>	if (nanos <= 0L) {
1099	>	thread = null;
1100	>	return true;
1101	>	}
1102	>	}
1103	>	return false;
1104		}
1105
1106		public boolean block() {
#	Line 948 | Line 1112 | public class Phaser {
1112		LockSupport.parkNanos(this, nanos);
1113		return isReleasable();
1114		}
951	–
952	–	void signal() {
953	–	Thread t = thread;
954	–	if (t != null) {
955	–	thread = null;
956	–	LockSupport.unpark(t);
957	–	}
958	–	}
1115		}
1116
1117		// Unsafe mechanics
1118
1119	<	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1120	<	private static final long stateOffset =
1121	<	objectFieldOffset("state", Phaser.class);
966	<
967	<	private static long objectFieldOffset(String field, Class<?> klazz) {
1119	>	private static final sun.misc.Unsafe UNSAFE;
1120	>	private static final long stateOffset;
1121	>	static {
1122		try {
1123	<	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1124	<	} catch (NoSuchFieldException e) {
1125	<	// Convert Exception to corresponding Error
1126	<	NoSuchFieldError error = new NoSuchFieldError(field);
1127	<	error.initCause(e);
1128	<	throw error;
1123	>	UNSAFE = getUnsafe();
1124	>	Class<?> k = Phaser.class;
1125	>	stateOffset = UNSAFE.objectFieldOffset
1126	>	(k.getDeclaredField("state"));
1127	>	} catch (Exception e) {
1128	>	throw new Error(e);
1129		}
1130		}
1131

Diff Legend

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