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 1.7 by jsr166, Mon Jan 5 03:53:26 2009 UTC vs.
Revision 1.58 by dl, Wed Nov 24 15:48:01 2010 UTC

#	Line 5 | Line 5
5		*/
6
7		package jsr166y;
8	<	import java.util.concurrent.*;
9	<	import java.util.concurrent.atomic.*;
8	>
9	>	import java.util.concurrent.TimeUnit;
10	>	import java.util.concurrent.TimeoutException;
11	>	import java.util.concurrent.atomic.AtomicReference;
12		import java.util.concurrent.locks.LockSupport;
11	–	import sun.misc.Unsafe;
12	–	import java.lang.reflect.*;
13
14		/**
15	<	* A reusable synchronization barrier, similar in functionality to a
16	<	* {@link java.util.concurrent.CyclicBarrier} and {@link
17	<	* java.util.concurrent.CountDownLatch} but supporting more flexible
18	<	* usage.
15	>	* A reusable synchronization barrier, similar in functionality to
16	>	* {@link java.util.concurrent.CyclicBarrier CyclicBarrier} and
17	>	* {@link java.util.concurrent.CountDownLatch CountDownLatch}
18	>	* but supporting more flexible usage.
19	>	*
20	>	* <p> <b>Registration.</b> Unlike the case for other barriers, the
21	>	* number of parties <em>registered</em> to synchronize on a phaser
22	>	* may vary over time.  Tasks may be registered at any time (using
23	>	* methods {@link #register}, {@link #bulkRegister}, or forms of
24	>	* constructors establishing initial numbers of parties), and
25	>	* optionally deregistered upon any arrival (using {@link
26	>	* #arriveAndDeregister}).  As is the case with most basic
27	>	* synchronization constructs, registration and deregistration affect
28	>	* only internal counts; they do not establish any further internal
29	>	* bookkeeping, so tasks cannot query whether they are registered.
30	>	* (However, you can introduce such bookkeeping by subclassing this
31	>	* class.)
32	>	*
33	>	* <p> <b>Synchronization.</b> Like a {@code CyclicBarrier}, a {@code
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
40	>	* Integer.MAX_VALUE}. The use of phase numbers enables independent
41	>	* control of actions upon arrival at a barrier 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> The number of parties synchronizing on a phaser may vary over
48	<	* time.  A task may register to be a party at any time, and may
49	<	* deregister upon arriving at the barrier.  As is the case with most
50	<	* basic synchronization constructs, registration and deregistration
51	<	* affect only internal counts; they do not establish any further
52	<	* internal bookkeeping, so tasks cannot query whether they are
53	<	* registered. (However, you can introduce such bookkeeping in by
54	<	* subclassing this class.)
55	<	*
56	<	* <li> Each generation has an associated phase value, starting at
57	<	* zero, and advancing when all parties reach the barrier (wrapping
58	<	* around to zero after reaching {@code Integer.MAX_VALUE}).
59	<	*
60	<	* <li> Like a CyclicBarrier, a Phaser may be repeatedly awaited.
61	<	* Method {@code arriveAndAwaitAdvance} has effect analogous to
62	<	* {@code CyclicBarrier.await}.  However, Phasers separate two
63	<	* aspects of coordination, that may also be invoked independently:
64	<	*
65	<	* <ul>
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}.
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.
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
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},
72	>	*       which will ensure sufficient parallelism to execute tasks
73	>	*       when others are blocked waiting for a phase to advance.
74		*
42	–	*   <li> Arriving at a barrier. Methods {@code arrive} and
43	–	*       {@code arriveAndDeregister} do not block, but return
44	–	*       the phase value current upon entry to the method.
45	–	*
46	–	*   <li> Awaiting others. Method {@code awaitAdvance} requires an
47	–	*       argument indicating the entry phase, and returns when the
48	–	*       barrier advances to a new phase.
75		* </ul>
76		*
77	<	*
78	<	* <li> Barrier actions, performed by the task triggering a phase
79	<	* advance while others may be waiting, are arranged by overriding
80	<	* method {@code onAdvance}, that also controls termination.
81	<	* Overriding this method may be used to similar but more flexible
82	<	* effect as providing a barrier action to a CyclicBarrier.
83	<	*
58	<	* <li> Phasers may enter a <em>termination</em> state in which all
59	<	* await actions immediately return, indicating (via a negative phase
60	<	* value) that execution is complete.  Termination is triggered by
61	<	* executing the overridable {@code onAdvance} method that is invoked
62	<	* each time the barrier is about to be tripped. When a Phaser is
63	<	* controlling an action with a fixed number of iterations, it is
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
86	<	* {@code forceTermination} is also available to abruptly release
87	<	* waiting threads and allow them to terminate.
88	<	*
89	<	* <li> Phasers may be tiered to reduce contention. Phasers with large
90	<	* numbers of parties that would otherwise experience heavy
91	<	* synchronization contention costs may instead be arranged in trees.
92	<	* This will typically greatly increase throughput even though it
93	<	* incurs somewhat greater per-operation overhead.
94	<	*
95	<	* <li> By default, {@code awaitAdvance} continues to wait even if
96	<	* the waiting thread is interrupted. And unlike the case in
97	<	* CyclicBarriers, exceptions encountered while tasks wait
98	<	* interruptibly or with timeout do not change the state of the
99	<	* barrier. If necessary, you can perform any associated recovery
100	<	* within handlers of those exceptions, often after invoking
101	<	* {@code forceTermination}.
102	<	*
103	<	* </ul>
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.,
90	>	* constructed in tree structures) to reduce contention. Phasers with
91	>	* large numbers of parties that would otherwise experience heavy
92	>	* synchronization contention costs may instead be set up so that
93	>	* groups of sub-phasers share a common parent.  This may greatly
94	>	* increase throughput even though it incurs greater per-operation
95	>	* overhead.
96	>	*
97	>	* <p><b>Monitoring.</b> While synchronization methods may be invoked
98	>	* only by registered parties, the current state of a phaser may be
99	>	* monitored by any caller.  At any given moment there are {@link
100	>	* #getRegisteredParties} parties in total, of which {@link
101	>	* #getArrivedParties} have arrived at the current phase ({@link
102	>	* #getPhase}).  When the remaining ({@link #getUnarrivedParties})
103	>	* parties arrive, the phase advances.  The values returned by these
104	>	* methods may reflect transient states and so are not in general
105	>	* useful for synchronization control.  Method {@link #toString}
106	>	* returns snapshots of these state queries in a form convenient for
107	>	* informal monitoring.
108		*
109		* <p><b>Sample usages:</b>
110		*
111	<	* <p>A Phaser may be used instead of a {@code CountdownLatch} to control
112	<	* a one-shot action serving a variable number of parties. The typical
113	<	* idiom is for the method setting this up to first register, then
114	<	* start the actions, then deregister, as in:
115	<	*
116	<	* <pre>
117	<	*  void runTasks(List&lt;Runnable&gt; list) {
118	<	*    final Phaser phaser = new Phaser(1); // "1" to register self
119	<	*    for (Runnable r : list) {
120	<	*      phaser.register();
121	<	*      new Thread() {
122	<	*        public void run() {
123	<	*          phaser.arriveAndAwaitAdvance(); // await all creation
124	<	*          r.run();
125	<	*          phaser.arriveAndDeregister();   // signal completion
126	<	*        }
127	<	*      }.start();
111	>	* <p>A {@code Phaser} may be used instead of a {@code CountDownLatch}
112	>	* to control a one-shot action serving a variable number of parties.
113	>	* The typical idiom is for the method setting this up to first
114	>	* register, then start the actions, then deregister, as in:
115	>	*
116	>	*  <pre> {@code
117	>	* void runTasks(List<Runnable> tasks) {
118	>	*   final Phaser phaser = new Phaser(1); // "1" to register self
119	>	*   // create and start threads
120	>	*   for (Runnable task : tasks) {
121	>	*     phaser.register();
122	>	*     new Thread() {
123	>	*       public void run() {
124	>	*         phaser.arriveAndAwaitAdvance(); // await all creation
125	>	*         task.run();
126	>	*       }
127	>	*     }.start();
128		*   }
129		*
130	<	*   doSomethingOnBehalfOfWorkers();
131	<	*   phaser.arrive(); // allow threads to start
132	<	*   int p = phaser.arriveAndDeregister(); // deregister self  ...
109	<	*   p = phaser.awaitAdvance(p); // ... and await arrival
110	<	*   otherActions(); // do other things while tasks execute
111	<	*   phaser.awaitAdvance(p); // awit final completion
112	<	* }
113	<	* </pre>
130	>	*   // allow threads to start and deregister self
131	>	*   phaser.arriveAndDeregister();
132	>	* }}</pre>
133		*
134		* <p>One way to cause a set of threads to repeatedly perform actions
135		* for a given number of iterations is to override {@code onAdvance}:
136		*
137	<	* <pre>
138	<	*  void startTasks(List&lt;Runnable&gt; list, final int iterations) {
139	<	*    final Phaser phaser = new Phaser() {
140	<	*       public boolean onAdvance(int phase, int registeredParties) {
141	<	*         return phase &gt;= iterations || registeredParties == 0;
137	>	*  <pre> {@code
138	>	* void startTasks(List<Runnable> tasks, final int iterations) {
139	>	*   final Phaser phaser = new Phaser() {
140	>	*     protected boolean onAdvance(int phase, int registeredParties) {
141	>	*       return phase >= iterations || registeredParties == 0;
142	>	*     }
143	>	*   };
144	>	*   phaser.register();
145	>	*   for (final Runnable task : tasks) {
146	>	*     phaser.register();
147	>	*     new Thread() {
148	>	*       public void run() {
149	>	*         do {
150	>	*           task.run();
151	>	*           phaser.arriveAndAwaitAdvance();
152	>	*         } while (!phaser.isTerminated());
153		*       }
154	<	*    };
125	<	*    phaser.register();
126	<	*    for (Runnable r : list) {
127	<	*      phaser.register();
128	<	*      new Thread() {
129	<	*        public void run() {
130	<	*           do {
131	<	*             r.run();
132	<	*             phaser.arriveAndAwaitAdvance();
133	<	*           } while(!phaser.isTerminated();
134	<	*        }
135	<	*      }.start();
154	>	*     }.start();
155		*   }
156		*   phaser.arriveAndDeregister(); // deregister self, don't wait
157	<	* }
158	<	* </pre>
157	>	* }}</pre>
158	>	*
159	>	* If the main task must later await termination, it
160	>	* may re-register and then execute a similar loop:
161	>	*  <pre> {@code
162	>	*   // ...
163	>	*   phaser.register();
164	>	*   while (!phaser.isTerminated())
165	>	*     phaser.arriveAndAwaitAdvance();}</pre>
166	>	*
167	>	* <p>Related constructions may be used to await particular phase numbers
168	>	* in contexts where you are sure that the phase will never wrap around
169	>	* {@code Integer.MAX_VALUE}. For example:
170	>	*
171	>	*  <pre> {@code
172	>	* void awaitPhase(Phaser phaser, int phase) {
173	>	*   int p = phaser.register(); // assumes caller not already registered
174	>	*   while (p < phase) {
175	>	*     if (phaser.isTerminated())
176	>	*       // ... deal with unexpected termination
177	>	*     else
178	>	*       p = phaser.arriveAndAwaitAdvance();
179	>	*   }
180	>	*   phaser.arriveAndDeregister();
181	>	* }}</pre>
182	>	*
183		*
184	<	* <p> To create a set of tasks using a tree of Phasers,
184	>	* <p>To create a set of tasks using a tree of phasers,
185		* you could use code of the following form, assuming a
186	<	* Task class with a constructor accepting a Phaser that
187	<	* it registers for upon construction:
188	<	* <pre>
189	<	*  void build(Task[] actions, int lo, int hi, Phaser b) {
190	<	*    int step = (hi - lo) / TASKS_PER_PHASER;
191	<	*    if (step &gt; 1) {
192	<	*       int i = lo;
193	<	*       while (i &lt; hi) {
194	<	*         int r = Math.min(i + step, hi);
195	<	*         build(actions, i, r, new Phaser(b));
196	<	*         i = r;
197	<	*       }
198	<	*    }
199	<	*    else {
200	<	*      for (int i = lo; i &lt; hi; ++i)
201	<	*        actions[i] = new Task(b);
202	<	*        // assumes new Task(b) performs b.register()
203	<	*    }
161	<	*  }
162	<	*  // .. initially called, for n tasks via
163	<	*  build(new Task[n], 0, n, new Phaser());
164	<	* </pre>
186	>	* Task class with a constructor accepting a phaser that
187	>	* it registers with upon construction:
188	>	*
189	>	*  <pre> {@code
190	>	* void build(Task[] actions, int lo, int hi, Phaser ph) {
191	>	*   if (hi - lo > TASKS_PER_PHASER) {
192	>	*     for (int i = lo; i < hi; i += TASKS_PER_PHASER) {
193	>	*       int j = Math.min(i + TASKS_PER_PHASER, hi);
194	>	*       build(actions, i, j, new Phaser(ph));
195	>	*     }
196	>	*   } else {
197	>	*     for (int i = lo; i < hi; ++i)
198	>	*       actions[i] = new Task(ph);
199	>	*       // assumes new Task(ph) performs ph.register()
200	>	*   }
201	>	* }
202	>	* // .. initially called, for n tasks via
203	>	* build(new Task[n], 0, n, new Phaser());}</pre>
204		*
205		* The best value of {@code TASKS_PER_PHASER} depends mainly on
206		* expected barrier synchronization rates. A value as low as four may
207		* be appropriate for extremely small per-barrier task bodies (thus
208		* high rates), or up to hundreds for extremely large ones.
209		*
171	–	* </pre>
172	–	*
210		* <p><b>Implementation notes</b>: This implementation restricts the
211		* maximum number of parties to 65535. Attempts to register additional
212	<	* parties result in IllegalStateExceptions. However, you can and
212	>	* parties result in {@code IllegalStateException}. However, you can and
213		* should create tiered phasers to accommodate arbitrarily large sets
214		* of participants.
215	+	*
216	+	* @since 1.7
217	+	* @author Doug Lea
218		*/
219		public class Phaser {
220		/*
#	Line 187 | Line 227 | public class Phaser {
227		* Barrier state representation. Conceptually, a barrier contains
228		* four values:
229		*
230	<	* * parties -- the number of parties to wait (16 bits)
231	<	* * unarrived -- the number of parties yet to hit barrier (16 bits)
232	<	* * phase -- the generation of the barrier (31 bits)
233	<	* * terminated -- set if barrier is terminated (1 bit)
230	>	* * unarrived -- the number of parties yet to hit barrier (bits  0-15)
231	>	* * parties -- the number of parties to wait              (bits 16-31)
232	>	* * phase -- the generation of the barrier                (bits 32-62)
233	>	* * terminated -- set if barrier is terminated            (bit  63 / sign)
234		*
235		* However, to efficiently maintain atomicity, these values are
236		* packed into a single (atomic) long. Termination uses the sign
237		* bit of 32 bit representation of phase, so phase is set to -1 on
238	<	* termination. Good performace relies on keeping state decoding
238	>	* termination. Good performance relies on keeping state decoding
239		* and encoding simple, and keeping race windows short.
200	–	*
201	–	* Note: there are some cheats in arrive() that rely on unarrived
202	–	* being lowest 16 bits.
240		*/
241		private volatile long state;
242
243	<	private static final int ushortBits = 16;
244	<	private static final int ushortMask =  (1 << ushortBits) - 1;
245	<	private static final int phaseMask = 0x7fffffff;
243	>	private static final int  MAX_PARTIES    = 0xffff;
244	>	private static final int  MAX_PHASE      = 0x7fffffff;
245	>	private static final int  PARTIES_SHIFT  = 16;
246	>	private static final int  PHASE_SHIFT    = 32;
247	>	private static final int  UNARRIVED_MASK = 0xffff;
248	>	private static final long PARTIES_MASK   = 0xffff0000L; // for masking long
249	>	private static final long ONE_ARRIVAL    = 1L;
250	>	private static final long ONE_PARTY      = 1L << PARTIES_SHIFT;
251	>	private static final long TERMINATION_PHASE  = -1L << PHASE_SHIFT;
252	>
253	>	// The following unpacking methods are usually manually inlined
254
255		private static int unarrivedOf(long s) {
256	<	return (int)(s & ushortMask);
256	>	return (int)s & UNARRIVED_MASK;
257		}
258
259		private static int partiesOf(long s) {
260	<	return (int)(s & (ushortMask << 16)) >>> 16;
260	>	return (int)s >>> PARTIES_SHIFT;
261		}
262
263		private static int phaseOf(long s) {
264	<	return (int)(s >>> 32);
264	>	return (int) (s >>> PHASE_SHIFT);
265		}
266
267		private static int arrivedOf(long s) {
268		return partiesOf(s) - unarrivedOf(s);
269		}
270
226	–	private static long stateFor(int phase, int parties, int unarrived) {
227	–	return (((long)phase) << 32) | ((parties << 16) | unarrived);
228	–	}
229	–
230	–	private static long trippedStateFor(int phase, int parties) {
231	–	return (((long)phase) << 32) | ((parties << 16) | parties);
232	–	}
233	–
234	–	private static IllegalStateException badBounds(int parties, int unarrived) {
235	–	return new IllegalStateException
236	–	("Attempt to set " + unarrived +
237	–	" unarrived of " + parties + " parties");
238	–	}
239	–
271		/**
272		* The parent of this phaser, or null if none
273		*/
274		private final Phaser parent;
275
276		/**
277	<	* The root of Phaser tree. Equals this if not in a tree.  Used to
277	>	* The root of phaser tree. Equals this if not in a tree.  Used to
278		* support faster state push-down.
279		*/
280		private final Phaser root;
281
251	–	// Wait queues
252	–
282		/**
283	<	* Heads of Treiber stacks waiting for nonFJ threads. To eliminate
284	<	* contention while releasing some threads while adding others, we
283	>	* Heads of Treiber stacks for waiting threads. To eliminate
284	>	* contention when releasing some threads while adding others, we
285		* use two of them, alternating across even and odd phases.
286	+	* Subphasers share queues with root to speed up releases.
287		*/
288	<	private final AtomicReference<QNode> evenQ = new AtomicReference<QNode>();
289	<	private final AtomicReference<QNode> oddQ  = new AtomicReference<QNode>();
288	>	private final AtomicReference<QNode> evenQ;
289	>	private final AtomicReference<QNode> oddQ;
290
291		private AtomicReference<QNode> queueFor(int phase) {
292	<	return (phase & 1) == 0? evenQ : oddQ;
292	>	return ((phase & 1) == 0) ? evenQ : oddQ;
293		}
294
295		/**
296	<	* Returns current state, first resolving lagged propagation from
297	<	* root if necessary.
296	>	* Main implementation for methods arrive and arriveAndDeregister.
297	>	* Manually tuned to speed up and minimize race windows for the
298	>	* common case of just decrementing unarrived field.
299	>	*
300	>	* @param adj - adjustment to apply to state -- either
301	>	* ONE_ARRIVAL (for arrive) or
302	>	* ONE_ARRIVAL|ONE_PARTY (for arriveAndDeregister)
303		*/
304	<	private long getReconciledState() {
305	<	return parent == null? state : reconcileState();
304	>	private int doArrive(long adj) {
305	>	for (;;) {
306	>	long s = state;
307	>	int phase = (int)(s >>> PHASE_SHIFT);
308	>	if (phase < 0)
309	>	return phase;
310	>	int unarrived = (int)s & UNARRIVED_MASK;
311	>	if (unarrived == 0)
312	>	checkBadArrive(s);
313	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adj)) {
314	>	if (unarrived == 1) {
315	>	long p = s & PARTIES_MASK; // unshifted parties field
316	>	long lu = p >>> PARTIES_SHIFT;
317	>	int u = (int)lu;
318	>	int nextPhase = (phase + 1) & MAX_PHASE;
319	>	long next = ((long)nextPhase << PHASE_SHIFT) | p | lu;
320	>	final Phaser parent = this.parent;
321	>	if (parent == null) {
322	>	if (onAdvance(phase, u))
323	>	next |= TERMINATION_PHASE; // obliterate phase
324	>	UNSAFE.compareAndSwapLong(this, stateOffset, s, next);
325	>	releaseWaiters(phase);
326	>	}
327	>	else {
328	>	parent.doArrive((u == 0) ?
329	>	ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL);
330	>	if ((int)(parent.state >>> PHASE_SHIFT) != nextPhase ||
331	>	((int)(state >>> PHASE_SHIFT) != nextPhase &&
332	>	!UNSAFE.compareAndSwapLong(this, stateOffset,
333	>	s, next)))
334	>	reconcileState();
335	>	}
336	>	}
337	>	return phase;
338	>	}
339	>	}
340	>	}
341	>
342	>	/**
343	>	* Rechecks state and throws bounds exceptions on arrival -- called
344	>	* only if unarrived is apparently zero.
345	>	*/
346	>	private void checkBadArrive(long s) {
347	>	if (reconcileState() == s)
348	>	throw new IllegalStateException
349	>	("Attempted arrival of unregistered party for " +
350	>	stateToString(s));
351	>	}
352	>
353	>	/**
354	>	* Implementation of register, bulkRegister
355	>	*
356	>	* @param registrations number to add to both parties and
357	>	* unarrived fields. Must be greater than zero.
358	>	*/
359	>	private int doRegister(int registrations) {
360	>	// adjustment to state
361	>	long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
362	>	final Phaser parent = this.parent;
363	>	for (;;) {
364	>	long s = (parent == null) ? state : reconcileState();
365	>	int parties = (int)s >>> PARTIES_SHIFT;
366	>	int phase = (int)(s >>> PHASE_SHIFT);
367	>	if (phase < 0)
368	>	return phase;
369	>	else if (parties != 0 && ((int)s & UNARRIVED_MASK) == 0)
370	>	internalAwaitAdvance(phase, null); // wait for onAdvance
371	>	else if (registrations > MAX_PARTIES - parties)
372	>	throw new IllegalStateException(badRegister(s));
373	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s + adj))
374	>	return phase;
375	>	}
376		}
377
378		/**
379	<	* Recursively resolves state.
379	>	* Returns message string for out of bounds exceptions on registration.
380	>	*/
381	>	private String badRegister(long s) {
382	>	return "Attempt to register more than " +
383	>	MAX_PARTIES + " parties for " + stateToString(s);
384	>	}
385	>
386	>	/**
387	>	* Recursively resolves lagged phase propagation from root if necessary.
388		*/
389		private long reconcileState() {
390	<	Phaser p = parent;
390	>	Phaser par = parent;
391		long s = state;
392	<	if (p != null) {
393	<	while (unarrivedOf(s) == 0 && phaseOf(s) != phaseOf(root.state)) {
394	<	long parentState = p.getReconciledState();
395	<	int parentPhase = phaseOf(parentState);
396	<	int phase = phaseOf(s = state);
397	<	if (phase != parentPhase) {
398	<	long next = trippedStateFor(parentPhase, partiesOf(s));
399	<	if (casState(s, next)) {
400	<	releaseWaiters(phase);
401	<	s = next;
402	<	}
392	>	if (par != null) {
393	>	Phaser rt = root;
394	>	int phase, rPhase;
395	>	while ((phase = (int)(s >>> PHASE_SHIFT)) >= 0 &&
396	>	(rPhase = (int)(rt.state >>> PHASE_SHIFT)) != phase) {
397	>	if ((int)(par.state >>> PHASE_SHIFT) != rPhase)
398	>	par.reconcileState();
399	>	else if (rPhase < 0 || ((int)s & UNARRIVED_MASK) == 0) {
400	>	long u = s & PARTIES_MASK; // reset unarrived to parties
401	>	long next = ((((long) rPhase) << PHASE_SHIFT) | u |
402	>	(u >>> PARTIES_SHIFT));
403	>	if (state == s &&
404	>	UNSAFE.compareAndSwapLong(this, stateOffset,
405	>	s, s = next))
406	>	break;
407		}
408	+	s = state;
409		}
410		}
411		return s;
412		}
413
414		/**
415	<	* Creates a new Phaser without any initially registered parties,
416	<	* initial phase number 0, and no parent.
415	>	* Creates a new phaser without any initially registered parties,
416	>	* initial phase number 0, and no parent. Any thread using this
417	>	* phaser will need to first register for it.
418		*/
419		public Phaser() {
420	<	this(null);
420	>	this(null, 0);
421		}
422
423		/**
424	<	* Creates a new Phaser with the given numbers of registered
424	>	* Creates a new phaser with the given number of registered
425		* unarrived parties, initial phase number 0, and no parent.
426	<	* @param parties the number of parties required to trip barrier.
426	>	*
427	>	* @param parties the number of parties required to trip barrier
428		* @throws IllegalArgumentException if parties less than zero
429	<	* or greater than the maximum number of parties supported.
429	>	* or greater than the maximum number of parties supported
430		*/
431		public Phaser(int parties) {
432		this(null, parties);
433		}
434
435		/**
436	<	* Creates a new Phaser with the given parent, without any
437	<	* initially registered parties. If parent is non-null this phaser
438	<	* is registered with the parent and its initial phase number is
319	<	* the same as that of parent phaser.
320	<	* @param parent the parent phaser.
436	>	* Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
437	>	*
438	>	* @param parent the parent phaser
439		*/
440		public Phaser(Phaser parent) {
441	<	int phase = 0;
324	<	this.parent = parent;
325	<	if (parent != null) {
326	<	this.root = parent.root;
327	<	phase = parent.register();
328	<	}
329	<	else
330	<	this.root = this;
331	<	this.state = trippedStateFor(phase, 0);
441	>	this(parent, 0);
442		}
443
444		/**
445	<	* Creates a new Phaser with the given parent and numbers of
446	<	* registered unarrived parties. If parent is non-null this phaser
447	<	* is registered with the parent and its initial phase number is
448	<	* the same as that of parent phaser.
449	<	* @param parent the parent phaser.
450	<	* @param parties the number of parties required to trip barrier.
445	>	* Creates a new phaser with the given parent and number of
446	>	* registered unarrived parties. If parent is non-null, this
447	>	* phaser is registered with the parent and its initial phase
448	>	* number is the same as that of parent phaser.  If the number of
449	>	* parties is zero, the parent phaser will not proceed until this
450	>	* child phaser registers parties and advances, or this child
451	>	* phaser deregisters with its parent, or the parent is otherwise
452	>	* terminated.  This child Phaser will be deregistered from its
453	>	* parent automatically upon any invocation of the child's {@link
454	>	* #arriveAndDeregister} method that results in the child's number
455	>	* of registered parties becoming zero. (Although rarely
456	>	* appropriate, this child may also explicity deregister from its
457	>	* parent using {@code getParent().arriveAndDeregister()}.)  After
458	>	* deregistration, the child cannot re-register. (Instead, you can
459	>	* create a new child Phaser.)
460	>	*
461	>	* @param parent the parent phaser
462	>	* @param parties the number of parties required to trip barrier
463		* @throws IllegalArgumentException if parties less than zero
464	<	* or greater than the maximum number of parties supported.
464	>	* or greater than the maximum number of parties supported
465		*/
466		public Phaser(Phaser parent, int parties) {
467	<	if (parties < 0 || parties > ushortMask)
467	>	if (parties >>> PARTIES_SHIFT != 0)
468		throw new IllegalArgumentException("Illegal number of parties");
469	<	int phase = 0;
469	>	int phase;
470		this.parent = parent;
471		if (parent != null) {
472	<	this.root = parent.root;
473	<	phase = parent.register();
472	>	Phaser r = parent.root;
473	>	this.root = r;
474	>	this.evenQ = r.evenQ;
475	>	this.oddQ = r.oddQ;
476	>	phase = parent.doRegister(1);
477		}
478	<	else
478	>	else {
479		this.root = this;
480	<	this.state = trippedStateFor(phase, parties);
480	>	this.evenQ = new AtomicReference<QNode>();
481	>	this.oddQ = new AtomicReference<QNode>();
482	>	phase = 0;
483	>	}
484	>	long p = (long)parties;
485	>	this.state = (((long)phase) << PHASE_SHIFT) | p | (p << PARTIES_SHIFT);
486		}
487
488		/**
489		* Adds a new unarrived party to this phaser.
490	<	* @return the current barrier phase number upon registration
490	>	* If an ongoing invocation of {@link #onAdvance} is in progress,
491	>	* this method may wait until its completion before registering.
492	>	*
493	>	* @return the arrival phase number to which this registration applied
494		* @throws IllegalStateException if attempting to register more
495	<	* than the maximum supported number of parties.
495	>	* than the maximum supported number of parties
496		*/
497		public int register() {
498		return doRegister(1);
#	Line 367 | Line 500 | public class Phaser {
500
501		/**
502		* Adds the given number of new unarrived parties to this phaser.
503	<	* @param parties the number of parties required to trip barrier.
504	<	* @return the current barrier phase number upon registration
503	>	* If an ongoing invocation of {@link #onAdvance} is in progress,
504	>	* this method may wait until its completion before registering.
505	>	*
506	>	* @param parties the number of additional parties required to trip barrier
507	>	* @return the arrival phase number to which this registration applied
508		* @throws IllegalStateException if attempting to register more
509	<	* than the maximum supported number of parties.
509	>	* than the maximum supported number of parties
510	>	* @throws IllegalArgumentException if {@code parties < 0}
511		*/
512		public int bulkRegister(int parties) {
513		if (parties < 0)
#	Line 381 | Line 518 | public class Phaser {
518		}
519
520		/**
384	–	* Shared code for register, bulkRegister
385	–	*/
386	–	private int doRegister(int registrations) {
387	–	int phase;
388	–	for (;;) {
389	–	long s = getReconciledState();
390	–	phase = phaseOf(s);
391	–	int unarrived = unarrivedOf(s) + registrations;
392	–	int parties = partiesOf(s) + registrations;
393	–	if (phase < 0)
394	–	break;
395	–	if (parties > ushortMask || unarrived > ushortMask)
396	–	throw badBounds(parties, unarrived);
397	–	if (phase == phaseOf(root.state) &&
398	–	casState(s, stateFor(phase, parties, unarrived)))
399	–	break;
400	–	}
401	–	return phase;
402	–	}
403	–
404	–	/**
521		* Arrives at the barrier, but does not wait for others.  (You can
522	<	* in turn wait for others via {@link #awaitAdvance}).
522	>	* in turn wait for others via {@link #awaitAdvance}).  It is an
523	>	* unenforced usage error for an unregistered party to invoke this
524	>	* method.
525		*
526	<	* @return the barrier phase number upon entry to this method, or a
409	<	* negative value if terminated;
526	>	* @return the arrival phase number, or a negative value if terminated
527		* @throws IllegalStateException if not terminated and the number
528	<	* of unarrived parties would become negative.
528	>	* of unarrived parties would become negative
529		*/
530		public int arrive() {
531	<	int phase;
415	<	for (;;) {
416	<	long s = state;
417	<	phase = phaseOf(s);
418	<	int parties = partiesOf(s);
419	<	int unarrived = unarrivedOf(s) - 1;
420	<	if (unarrived > 0) {        // Not the last arrival
421	<	if (casState(s, s - 1)) // s-1 adds one arrival
422	<	break;
423	<	}
424	<	else if (unarrived == 0) {  // the last arrival
425	<	Phaser par = parent;
426	<	if (par == null) {      // directly trip
427	<	if (casState
428	<	(s,
429	<	trippedStateFor(onAdvance(phase, parties)? -1 :
430	<	((phase + 1) & phaseMask), parties))) {
431	<	releaseWaiters(phase);
432	<	break;
433	<	}
434	<	}
435	<	else {                  // cascade to parent
436	<	if (casState(s, s - 1)) { // zeroes unarrived
437	<	par.arrive();
438	<	reconcileState();
439	<	break;
440	<	}
441	<	}
442	<	}
443	<	else if (phase < 0) // Don't throw exception if terminated
444	<	break;
445	<	else if (phase != phaseOf(root.state)) // or if unreconciled
446	<	reconcileState();
447	<	else
448	<	throw badBounds(parties, unarrived);
449	<	}
450	<	return phase;
531	>	return doArrive(ONE_ARRIVAL);
532		}
533
534		/**
535	<	* Arrives at the barrier, and deregisters from it, without
536	<	* waiting for others. Deregistration reduces number of parties
535	>	* Arrives at the barrier and deregisters from it without waiting
536	>	* for others. Deregistration reduces the number of parties
537		* required to trip the barrier in future phases.  If this phaser
538		* has a parent, and deregistration causes this phaser to have
539	<	* zero parties, this phaser is also deregistered from its parent.
539	>	* zero parties, this phaser also arrives at and is deregistered
540	>	* from its parent.  It is an unenforced usage error for an
541	>	* unregistered party to invoke this method.
542		*
543	<	* @return the current barrier phase number upon entry to
461	<	* this method, or a negative value if terminated;
543	>	* @return the arrival phase number, or a negative value if terminated
544		* @throws IllegalStateException if not terminated and the number
545	<	* of registered or unarrived parties would become negative.
545	>	* of registered or unarrived parties would become negative
546		*/
547		public int arriveAndDeregister() {
548	<	// similar code to arrive, but too different to merge
467	<	Phaser par = parent;
468	<	int phase;
469	<	for (;;) {
470	<	long s = state;
471	<	phase = phaseOf(s);
472	<	int parties = partiesOf(s) - 1;
473	<	int unarrived = unarrivedOf(s) - 1;
474	<	if (parties >= 0) {
475	<	if (unarrived > 0 || (unarrived == 0 && par != null)) {
476	<	if (casState
477	<	(s,
478	<	stateFor(phase, parties, unarrived))) {
479	<	if (unarrived == 0) {
480	<	par.arriveAndDeregister();
481	<	reconcileState();
482	<	}
483	<	break;
484	<	}
485	<	continue;
486	<	}
487	<	if (unarrived == 0) {
488	<	if (casState
489	<	(s,
490	<	trippedStateFor(onAdvance(phase, parties)? -1 :
491	<	((phase + 1) & phaseMask), parties))) {
492	<	releaseWaiters(phase);
493	<	break;
494	<	}
495	<	continue;
496	<	}
497	<	if (phase < 0)
498	<	break;
499	<	if (par != null && phase != phaseOf(root.state)) {
500	<	reconcileState();
501	<	continue;
502	<	}
503	<	}
504	<	throw badBounds(parties, unarrived);
505	<	}
506	<	return phase;
548	>	return doArrive(ONE_ARRIVAL|ONE_PARTY);
549		}
550
551		/**
552		* Arrives at the barrier and awaits others. Equivalent in effect
553	<	* to {@code awaitAdvance(arrive())}.  If you instead need to
554	<	* await with interruption of timeout, and/or deregister upon
555	<	* arrival, you can arrange them using analogous constructions.
556	<	* @return the phase on entry to this method
553	>	* to {@code awaitAdvance(arrive())}.  If you need to await with
554	>	* interruption or timeout, you can arrange this with an analogous
555	>	* construction using one of the other forms of the {@code
556	>	* awaitAdvance} method.  If instead you need to deregister upon
557	>	* arrival, use {@link #arriveAndDeregister}. It is an unenforced
558	>	* usage error for an unregistered party to invoke this method.
559	>	*
560	>	* @return the arrival phase number, or a negative number if terminated
561		* @throws IllegalStateException if not terminated and the number
562	<	* of unarrived parties would become negative.
562	>	* of unarrived parties would become negative
563		*/
564		public int arriveAndAwaitAdvance() {
565		return awaitAdvance(arrive());
566		}
567
568		/**
569	<	* Awaits the phase of the barrier to advance from the given
570	<	* value, or returns immediately if argument is negative or this
571	<	* barrier is terminated.
572	<	* @param phase the phase on entry to this method
573	<	* @return the phase on exit from this method
569	>	* Awaits the phase of the barrier to advance from the given phase
570	>	* value, returning immediately if the current phase of the
571	>	* barrier is not equal to the given phase value or this barrier
572	>	* is terminated.
573	>	*
574	>	* @param phase an arrival phase number, or negative value if
575	>	* terminated; this argument is normally the value returned by a
576	>	* previous call to {@code arrive} or its variants
577	>	* @return the next arrival phase number, or a negative value
578	>	* if terminated or argument is negative
579		*/
580		public int awaitAdvance(int phase) {
581		if (phase < 0)
582		return phase;
583	<	long s = getReconciledState();
584	<	int p = phaseOf(s);
585	<	if (p != phase)
535	<	return p;
536	<	if (unarrivedOf(s) == 0)
537	<	parent.awaitAdvance(phase);
538	<	// Fall here even if parent waited, to reconcile and help release
539	<	return untimedWait(phase);
583	>	long s = (parent == null) ? state : reconcileState();
584	>	int p = (int)(s >>> PHASE_SHIFT);
585	>	return (p != phase) ? p : internalAwaitAdvance(phase, null);
586		}
587
588		/**
589	<	* Awaits the phase of the barrier to advance from the given
590	<	* value, or returns immediately if argumet is negative or this
591	<	* barrier is terminated, or throws InterruptedException if
592	<	* interrupted while waiting.
593	<	* @param phase the phase on entry to this method
594	<	* @return the phase on exit from this method
589	>	* Awaits the phase of the barrier to advance from the given phase
590	>	* value, throwing {@code InterruptedException} if interrupted
591	>	* while waiting, or returning immediately if the current phase of
592	>	* the barrier is not equal to the given phase value or this
593	>	* barrier is terminated.
594	>	*
595	>	* @param phase an arrival phase number, or negative value if
596	>	* terminated; this argument is normally the value returned by a
597	>	* previous call to {@code arrive} or its variants
598	>	* @return the next arrival phase number, or a negative value
599	>	* if terminated or argument is negative
600		* @throws InterruptedException if thread interrupted while waiting
601		*/
602	<	public int awaitAdvanceInterruptibly(int phase) throws InterruptedException {
602	>	public int awaitAdvanceInterruptibly(int phase)
603	>	throws InterruptedException {
604		if (phase < 0)
605		return phase;
606	<	long s = getReconciledState();
607	<	int p = phaseOf(s);
608	<	if (p != phase)
609	<	return p;
610	<	if (unarrivedOf(s) != 0)
611	<	parent.awaitAdvanceInterruptibly(phase);
612	<	return interruptibleWait(phase);
606	>	long s = (parent == null) ? state : reconcileState();
607	>	int p = (int)(s >>> PHASE_SHIFT);
608	>	if (p == phase) {
609	>	QNode node = new QNode(this, phase, true, false, 0L);
610	>	p = internalAwaitAdvance(phase, node);
611	>	if (node.wasInterrupted)
612	>	throw new InterruptedException();
613	>	}
614	>	return p;
615		}
616
617		/**
618	<	* Awaits the phase of the barrier to advance from the given value
619	<	* or the given timeout elapses, or returns immediately if
620	<	* argument is negative or this barrier is terminated.
621	<	* @param phase the phase on entry to this method
622	<	* @return the phase on exit from this method
618	>	* Awaits the phase of the barrier to advance from the given phase
619	>	* value or the given timeout to elapse, throwing {@code
620	>	* InterruptedException} if interrupted while waiting, or
621	>	* returning immediately if the current phase of the barrier is
622	>	* not equal to the given phase value or this barrier is
623	>	* terminated.
624	>	*
625	>	* @param phase an arrival phase number, or negative value if
626	>	* terminated; this argument is normally the value returned by a
627	>	* previous call to {@code arrive} or its variants
628	>	* @param timeout how long to wait before giving up, in units of
629	>	*        {@code unit}
630	>	* @param unit a {@code TimeUnit} determining how to interpret the
631	>	*        {@code timeout} parameter
632	>	* @return the next arrival phase number, or a negative value
633	>	* if terminated or argument is negative
634		* @throws InterruptedException if thread interrupted while waiting
635		* @throws TimeoutException if timed out while waiting
636		*/
637	<	public int awaitAdvanceInterruptibly(int phase, long timeout, TimeUnit unit)
637	>	public int awaitAdvanceInterruptibly(int phase,
638	>	long timeout, TimeUnit unit)
639		throws InterruptedException, TimeoutException {
640		if (phase < 0)
641		return phase;
642	<	long s = getReconciledState();
643	<	int p = phaseOf(s);
644	<	if (p != phase)
645	<	return p;
646	<	if (unarrivedOf(s) == 0)
647	<	parent.awaitAdvanceInterruptibly(phase, timeout, unit);
648	<	return timedWait(phase, unit.toNanos(timeout));
642	>	long s = (parent == null) ? state : reconcileState();
643	>	int p = (int)(s >>> PHASE_SHIFT);
644	>	if (p == phase) {
645	>	long nanos = unit.toNanos(timeout);
646	>	QNode node = new QNode(this, phase, true, true, nanos);
647	>	p = internalAwaitAdvance(phase, node);
648	>	if (node.wasInterrupted)
649	>	throw new InterruptedException();
650	>	else if (p == phase)
651	>	throw new TimeoutException();
652	>	}
653	>	return p;
654		}
655
656		/**
657	<	* Forces this barrier to enter termination state. Counts of
658	<	* arrived and registered parties are unaffected. If this phaser
659	<	* has a parent, it too is terminated. This method may be useful
660	<	* for coordinating recovery after one or more tasks encounter
661	<	* unexpected exceptions.
657	>	* Forces this barrier to enter termination state.  Counts of
658	>	* arrived and registered parties are unaffected.  If this phaser
659	>	* is a member of a tiered set of phasers, then all of the phasers
660	>	* in the set are terminated.  If this phaser is already
661	>	* terminated, this method has no effect.  This method may be
662	>	* useful for coordinating recovery after one or more tasks
663	>	* encounter unexpected exceptions.
664		*/
665		public void forceTermination() {
666	<	for (;;) {
667	<	long s = getReconciledState();
668	<	int phase = phaseOf(s);
669	<	int parties = partiesOf(s);
670	<	int unarrived = unarrivedOf(s);
671	<	if (phase < 0 ||
672	<	casState(s, stateFor(-1, parties, unarrived))) {
600	<	releaseWaiters(0);
666	>	// Only need to change root state
667	>	final Phaser root = this.root;
668	>	long s;
669	>	while ((s = root.state) >= 0) {
670	>	if (UNSAFE.compareAndSwapLong(root, stateOffset,
671	>	s, s | TERMINATION_PHASE)) {
672	>	releaseWaiters(0); // signal all threads
673		releaseWaiters(1);
602	–	if (parent != null)
603	–	parent.forceTermination();
674		return;
675		}
676		}
#	Line 610 | Line 680 | public class Phaser {
680		* Returns the current phase number. The maximum phase number is
681		* {@code Integer.MAX_VALUE}, after which it restarts at
682		* zero. Upon termination, the phase number is negative.
683	+	*
684		* @return the phase number, or a negative value if terminated
685		*/
686		public final int getPhase() {
687	<	return phaseOf(getReconciledState());
617	<	}
618	<
619	<	/**
620	<	* Returns true if the current phase number equals the given phase.
621	<	* @param phase the phase
622	<	* @return true if the current phase number equals the given phase.
623	<	*/
624	<	public final boolean hasPhase(int phase) {
625	<	return phaseOf(getReconciledState()) == phase;
687	>	return (int)(root.state >>> PHASE_SHIFT);
688		}
689
690		/**
691		* Returns the number of parties registered at this barrier.
692	+	*
693		* @return the number of parties
694		*/
695		public int getRegisteredParties() {
#	Line 634 | Line 697 | public class Phaser {
697		}
698
699		/**
700	<	* Returns the number of parties that have arrived at the current
701	<	* phase of this barrier.
700	>	* Returns the number of registered parties that have arrived at
701	>	* the current phase of this barrier.
702	>	*
703		* @return the number of arrived parties
704		*/
705		public int getArrivedParties() {
706	<	return arrivedOf(state);
706	>	return arrivedOf(parent==null? state : reconcileState());
707		}
708
709		/**
710		* Returns the number of registered parties that have not yet
711		* arrived at the current phase of this barrier.
712	+	*
713		* @return the number of unarrived parties
714		*/
715		public int getUnarrivedParties() {
716	<	return unarrivedOf(state);
716	>	return unarrivedOf(parent==null? state : reconcileState());
717		}
718
719		/**
720	<	* Returns the parent of this phaser, or null if none.
721	<	* @return the parent of this phaser, or null if none.
720	>	* Returns the parent of this phaser, or {@code null} if none.
721	>	*
722	>	* @return the parent of this phaser, or {@code null} if none
723		*/
724		public Phaser getParent() {
725		return parent;
#	Line 662 | Line 728 | public class Phaser {
728		/**
729		* Returns the root ancestor of this phaser, which is the same as
730		* this phaser if it has no parent.
731	<	* @return the root ancestor of this phaser.
731	>	*
732	>	* @return the root ancestor of this phaser
733		*/
734		public Phaser getRoot() {
735		return root;
736		}
737
738		/**
739	<	* Returns true if this barrier has been terminated.
740	<	* @return true if this barrier has been terminated
739	>	* Returns {@code true} if this barrier has been terminated.
740	>	*
741	>	* @return {@code true} if this barrier has been terminated
742		*/
743		public boolean isTerminated() {
744	<	return getPhase() < 0;
744	>	return root.state < 0L;
745		}
746
747		/**
748	<	* Overridable method to perform an action upon phase advance, and
749	<	* to control termination. This method is invoked whenever the
750	<	* barrier is tripped (and thus all other waiting parties are
751	<	* dormant). If it returns true, then, rather than advance the
752	<	* phase number, this barrier will be set to a final termination
753	<	* state, and subsequent calls to {@code isTerminated} will
754	<	* return true.
748	>	* Overridable method to perform an action upon impending phase
749	>	* advance, and to control termination. This method is invoked
750	>	* upon arrival of the party tripping the barrier (when all other
751	>	* waiting parties are dormant).  If this method returns {@code
752	>	* true}, then, rather than advance the phase number, this barrier
753	>	* will be set to a final termination state, and subsequent calls
754	>	* to {@link #isTerminated} will return true. Any (unchecked)
755	>	* Exception or Error thrown by an invocation of this method is
756	>	* propagated to the party attempting to trip the barrier, in
757	>	* which case no advance occurs.
758		*
759	<	* <p> The default version returns true when the number of
759	>	* <p>The arguments to this method provide the state of the phaser
760	>	* prevailing for the current transition.  The effects of invoking
761	>	* arrival, registration, and waiting methods on this Phaser from
762	>	* within {@code onAdvance} are unspecified and should not be
763	>	* relied on.
764	>	*
765	>	* <p>If this Phaser is a member of a tiered set of Phasers, then
766	>	* {@code onAdvance} is invoked only for its root Phaser on each
767	>	* advance.
768	>	*
769	>	* <p>The default version returns {@code true} when the number of
770		* registered parties is zero. Normally, overrides that arrange
771		* termination for other reasons should also preserve this
772		* property.
773		*
693	–	* <p> You may override this method to perform an action with side
694	–	* effects visible to participating tasks, but it is in general
695	–	* only sensible to do so in designs where all parties register
696	–	* before any arrive, and all {@code awaitAdvance} at each phase.
697	–	* Otherwise, you cannot ensure lack of interference. In
698	–	* particular, this method may be invoked more than once per
699	–	* transition if other parties successfully register while the
700	–	* invocation of this method is in progress, thus postponing the
701	–	* transition until those parties also arrive, re-triggering this
702	–	* method.
703	–	*
774		* @param phase the phase number on entering the barrier
775	<	* @param registeredParties the current number of registered
776	<	* parties.
707	<	* @return true if this barrier should terminate
775	>	* @param registeredParties the current number of registered parties
776	>	* @return {@code true} if this barrier should terminate
777		*/
778		protected boolean onAdvance(int phase, int registeredParties) {
779		return registeredParties <= 0;
#	Line 713 | Line 782 | public class Phaser {
782		/**
783		* Returns a string identifying this phaser, as well as its
784		* state.  The state, in brackets, includes the String {@code
785	<	* "phase ="} followed by the phase number, {@code "parties ="}
785	>	* "phase = "} followed by the phase number, {@code "parties = "}
786		* followed by the number of registered parties, and {@code
787	<	* "arrived ="} followed by the number of arrived parties
787	>	* "arrived = "} followed by the number of arrived parties.
788		*
789		* @return a string identifying this barrier, as well as its state
790		*/
791		public String toString() {
792	<	long s = getReconciledState();
724	<	return super.toString() + "[phase = " + phaseOf(s) + " parties = " + partiesOf(s) + " arrived = " + arrivedOf(s) + "]";
792	>	return stateToString(reconcileState());
793		}
794
727	–	// methods for waiting
728	–
729	–	/** The number of CPUs, for spin control */
730	–	static final int NCPUS = Runtime.getRuntime().availableProcessors();
731	–
795		/**
796	<	* The number of times to spin before blocking in timed waits.
734	<	* The value is empirically derived.
796	>	* Implementation of toString and string-based error messages
797		*/
798	<	static final int maxTimedSpins = (NCPUS < 2)? 0 : 32;
799	<
800	<	/**
801	<	* The number of times to spin before blocking in untimed waits.
802	<	* This is greater than timed value because untimed waits spin
741	<	* faster since they don't need to check times on each spin.
742	<	*/
743	<	static final int maxUntimedSpins = maxTimedSpins * 32;
744	<
745	<	/**
746	<	* The number of nanoseconds for which it is faster to spin
747	<	* rather than to use timed park. A rough estimate suffices.
748	<	*/
749	<	static final long spinForTimeoutThreshold = 1000L;
750	<
751	<	/**
752	<	* Wait nodes for Treiber stack representing wait queue for non-FJ
753	<	* tasks.
754	<	*/
755	<	static final class QNode {
756	<	QNode next;
757	<	volatile Thread thread; // nulled to cancel wait
758	<	QNode() {
759	<	thread = Thread.currentThread();
760	<	}
761	<	void signal() {
762	<	Thread t = thread;
763	<	if (t != null) {
764	<	thread = null;
765	<	LockSupport.unpark(t);
766	<	}
767	<	}
798	>	private String stateToString(long s) {
799	>	return super.toString() +
800	>	"[phase = " + phaseOf(s) +
801	>	" parties = " + partiesOf(s) +
802	>	" arrived = " + arrivedOf(s) + "]";
803		}
804
805	+	// Waiting mechanics
806	+
807		/**
808	<	* Removes and signals waiting threads from wait queue
808	>	* Removes and signals threads from queue for phase.
809		*/
810		private void releaseWaiters(int phase) {
811		AtomicReference<QNode> head = queueFor(phase);
812		QNode q;
813	<	while ((q = head.get()) != null) {
813	>	int p;
814	>	while ((q = head.get()) != null &&
815	>	((p = q.phase) == phase ||
816	>	(int)(root.state >>> PHASE_SHIFT) != p)) {
817		if (head.compareAndSet(q, q.next))
818		q.signal();
819		}
820		}
821
822	+	/** The number of CPUs, for spin control */
823	+	private static final int NCPU = Runtime.getRuntime().availableProcessors();
824	+
825		/**
826	<	* Enqueues node and waits unless aborted or signalled.
826	>	* The number of times to spin before blocking while waiting for
827	>	* advance, per arrival while waiting. On multiprocessors, fully
828	>	* blocking and waking up a large number of threads all at once is
829	>	* usually a very slow process, so we use rechargeable spins to
830	>	* avoid it when threads regularly arrive: When a thread in
831	>	* internalAwaitAdvance notices another arrival before blocking,
832	>	* and there appear to be enough CPUs available, it spins
833	>	* SPINS_PER_ARRIVAL more times before blocking. Plus, even on
834	>	* uniprocessors, there is at least one intervening Thread.yield
835	>	* before blocking. The value trades off good-citizenship vs big
836	>	* unnecessary slowdowns.
837		*/
838	<	private int untimedWait(int phase) {
786	<	int spins = maxUntimedSpins;
787	<	QNode node = null;
788	<	boolean interrupted = false;
789	<	boolean queued = false;
790	<	int p;
791	<	while ((p = getPhase()) == phase) {
792	<	interrupted = Thread.interrupted();
793	<	if (node != null) {
794	<	if (!queued) {
795	<	AtomicReference<QNode> head = queueFor(phase);
796	<	queued = head.compareAndSet(node.next = head.get(), node);
797	<	}
798	<	else if (node.thread != null)
799	<	LockSupport.park(this);
800	<	}
801	<	else if (spins <= 0)
802	<	node = new QNode();
803	<	else
804	<	--spins;
805	<	}
806	<	if (node != null)
807	<	node.thread = null;
808	<	if (interrupted)
809	<	Thread.currentThread().interrupt();
810	<	releaseWaiters(phase);
811	<	return p;
812	<	}
838	>	static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
839
840		/**
841	<	* Messier interruptible version
842	<	*/
843	<	private int interruptibleWait(int phase) throws InterruptedException {
844	<	int spins = maxUntimedSpins;
845	<	QNode node = null;
846	<	boolean queued = false;
847	<	boolean interrupted = false;
841	>	* Possibly blocks and waits for phase to advance unless aborted.
842	>	*
843	>	* @param phase current phase
844	>	* @param node if non-null, the wait node to track interrupt and timeout;
845	>	* if null, denotes noninterruptible wait
846	>	* @return current phase
847	>	*/
848	>	private int internalAwaitAdvance(int phase, QNode node) {
849	>	Phaser current = this;       // to eventually wait at root if tiered
850	>	boolean queued = false;      // true when node is enqueued
851	>	int lastUnarrived = -1;      // to increase spins upon change
852	>	int spins = SPINS_PER_ARRIVAL;
853	>	long s;
854		int p;
855	<	while ((p = getPhase()) == phase) {
856	<	if (interrupted = Thread.interrupted())
857	<	break;
858	<	if (node != null) {
859	<	if (!queued) {
860	<	AtomicReference<QNode> head = queueFor(phase);
861	<	queued = head.compareAndSet(node.next = head.get(), node);
855	>	while ((p = (int)((s = current.state) >>> PHASE_SHIFT)) == phase) {
856	>	Phaser par;
857	>	int unarrived = (int)s & UNARRIVED_MASK;
858	>	if (unarrived != lastUnarrived) {
859	>	if (lastUnarrived == -1) // ensure old queue clean
860	>	releaseWaiters(phase-1);
861	>	if ((lastUnarrived = unarrived) < NCPU)
862	>	spins += SPINS_PER_ARRIVAL;
863	>	}
864	>	else if (unarrived == 0 && (par = current.parent) != null) {
865	>	current = par;       // if all arrived, use parent
866	>	par = par.parent;
867	>	lastUnarrived = -1;
868	>	}
869	>	else if (spins > 0) {
870	>	if (--spins == (SPINS_PER_ARRIVAL >>> 1))
871	>	Thread.yield();  // yield midway through spin
872	>	}
873	>	else if (node == null)   // must be noninterruptible
874	>	node = new QNode(this, phase, false, false, 0L);
875	>	else if (node.isReleasable()) {
876	>	if ((p = (int)(root.state >>> PHASE_SHIFT)) != phase)
877	>	break;
878	>	else
879	>	return phase;    // aborted
880	>	}
881	>	else if (!queued) {      // push onto queue
882	>	AtomicReference<QNode> head = queueFor(phase);
883	>	QNode q = head.get();
884	>	if (q == null || q.phase == phase) {
885	>	node.next = q;
886	>	if ((p = (int)(root.state >>> PHASE_SHIFT)) != phase)
887	>	break;       // recheck to avoid stale enqueue
888	>	else
889	>	queued = head.compareAndSet(q, node);
890	>	}
891	>	}
892	>	else {
893	>	try {
894	>	ForkJoinPool.managedBlock(node);
895	>	} catch (InterruptedException ie) {
896	>	node.wasInterrupted = true;
897		}
831	–	else if (node.thread != null)
832	–	LockSupport.park(this);
898		}
834	–	else if (spins <= 0)
835	–	node = new QNode();
836	–	else
837	–	--spins;
899		}
839	–	if (node != null)
840	–	node.thread = null;
841	–	if (interrupted)
842	–	throw new InterruptedException();
900		releaseWaiters(phase);
901	+	if (node != null)
902	+	node.onRelease();
903		return p;
904		}
905
906		/**
907	<	* Even messier timeout version.
907	>	* Wait nodes for Treiber stack representing wait queue
908		*/
909	<	private int timedWait(int phase, long nanos)
910	<	throws InterruptedException, TimeoutException {
911	<	int p;
912	<	if ((p = getPhase()) == phase) {
913	<	long lastTime = System.nanoTime();
914	<	int spins = maxTimedSpins;
915	<	QNode node = null;
916	<	boolean queued = false;
917	<	boolean interrupted = false;
918	<	while ((p = getPhase()) == phase) {
919	<	if (interrupted = Thread.interrupted())
920	<	break;
921	<	long now = System.nanoTime();
922	<	if ((nanos -= now - lastTime) <= 0)
923	<	break;
924	<	lastTime = now;
925	<	if (node != null) {
926	<	if (!queued) {
927	<	AtomicReference<QNode> head = queueFor(phase);
928	<	queued = head.compareAndSet(node.next = head.get(), node);
929	<	}
930	<	else if (node.thread != null &&
931	<	nanos > spinForTimeoutThreshold) {
932	<	LockSupport.parkNanos(this, nanos);
909	>	static final class QNode implements ForkJoinPool.ManagedBlocker {
910	>	final Phaser phaser;
911	>	final int phase;
912	>	final boolean interruptible;
913	>	final boolean timed;
914	>	boolean wasInterrupted;
915	>	long nanos;
916	>	long lastTime;
917	>	volatile Thread thread; // nulled to cancel wait
918	>	QNode next;
919	>
920	>	QNode(Phaser phaser, int phase, boolean interruptible,
921	>	boolean timed, long nanos) {
922	>	this.phaser = phaser;
923	>	this.phase = phase;
924	>	this.interruptible = interruptible;
925	>	this.nanos = nanos;
926	>	this.timed = timed;
927	>	this.lastTime = timed? System.nanoTime() : 0L;
928	>	thread = Thread.currentThread();
929	>	}
930	>
931	>	public boolean isReleasable() {
932	>	Thread t = thread;
933	>	if (t != null) {
934	>	if (phaser.getPhase() != phase)
935	>	t = null;
936	>	else {
937	>	if (Thread.interrupted())
938	>	wasInterrupted = true;
939	>	if (interruptible && wasInterrupted)
940	>	t = null;
941	>	else if (timed) {
942	>	if (nanos > 0) {
943	>	long now = System.nanoTime();
944	>	nanos -= now - lastTime;
945	>	lastTime = now;
946	>	}
947	>	if (nanos <= 0)
948	>	t = null;
949		}
950		}
951	<	else if (spins <= 0)
952	<	node = new QNode();
953	<	else
879	<	--spins;
951	>	if (t != null)
952	>	return false;
953	>	thread = null;
954		}
955	<	if (node != null)
882	<	node.thread = null;
883	<	if (interrupted)
884	<	throw new InterruptedException();
885	<	if (p == phase && (p = getPhase()) == phase)
886	<	throw new TimeoutException();
955	>	return true;
956		}
957	<	releaseWaiters(phase);
958	<	return p;
957	>
958	>	public boolean block() {
959	>	if (isReleasable())
960	>	return true;
961	>	else if (!timed)
962	>	LockSupport.park(this);
963	>	else if (nanos > 0)
964	>	LockSupport.parkNanos(this, nanos);
965	>	return isReleasable();
966	>	}
967	>
968	>	void signal() {
969	>	Thread t = thread;
970	>	if (t != null) {
971	>	thread = null;
972	>	LockSupport.unpark(t);
973	>	}
974	>	}
975	>
976	>	void onRelease() { // actions upon return from internalAwaitAdvance
977	>	if (!interruptible && wasInterrupted)
978	>	Thread.currentThread().interrupt();
979	>	if (thread != null)
980	>	thread = null;
981	>	}
982	>
983		}
984
985	<	// Temporary Unsafe mechanics for preliminary release
985	>	// Unsafe mechanics
986
987	<	static final Unsafe _unsafe;
988	<	static final long stateOffset;
987	>	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
988	>	private static final long stateOffset =
989	>	objectFieldOffset("state", Phaser.class);
990
991	<	static {
991	>	private static long objectFieldOffset(String field, Class<?> klazz) {
992		try {
993	<	if (Phaser.class.getClassLoader() != null) {
994	<	Field f = Unsafe.class.getDeclaredField("theUnsafe");
995	<	f.setAccessible(true);
996	<	_unsafe = (Unsafe)f.get(null);
997	<	}
998	<	else
905	<	_unsafe = Unsafe.getUnsafe();
906	<	stateOffset = _unsafe.objectFieldOffset
907	<	(Phaser.class.getDeclaredField("state"));
908	<	} catch (Exception e) {
909	<	throw new RuntimeException("Could not initialize intrinsics", e);
993	>	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
994	>	} catch (NoSuchFieldException e) {
995	>	// Convert Exception to corresponding Error
996	>	NoSuchFieldError error = new NoSuchFieldError(field);
997	>	error.initCause(e);
998	>	throw error;
999		}
1000		}
1001
1002	<	final boolean casState(long cmp, long val) {
1003	<	return _unsafe.compareAndSwapLong(this, stateOffset, cmp, val);
1002	>	/**
1003	>	* Returns a sun.misc.Unsafe.  Suitable for use in a 3rd party package.
1004	>	* Replace with a simple call to Unsafe.getUnsafe when integrating
1005	>	* into a jdk.
1006	>	*
1007	>	* @return a sun.misc.Unsafe
1008	>	*/
1009	>	private static sun.misc.Unsafe getUnsafe() {
1010	>	try {
1011	>	return sun.misc.Unsafe.getUnsafe();
1012	>	} catch (SecurityException se) {
1013	>	try {
1014	>	return java.security.AccessController.doPrivileged
1015	>	(new java.security
1016	>	.PrivilegedExceptionAction<sun.misc.Unsafe>() {
1017	>	public sun.misc.Unsafe run() throws Exception {
1018	>	java.lang.reflect.Field f = sun.misc
1019	>	.Unsafe.class.getDeclaredField("theUnsafe");
1020	>	f.setAccessible(true);
1021	>	return (sun.misc.Unsafe) f.get(null);
1022	>	}});
1023	>	} catch (java.security.PrivilegedActionException e) {
1024	>	throw new RuntimeException("Could not initialize intrinsics",
1025	>	e.getCause());
1026	>	}
1027	>	}
1028		}
1029		}

Diff Legend

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