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 1.11 by jsr166, Thu Mar 19 04:49:44 2009 UTC vs.
Revision 1.69 by jsr166, Sat Dec 4 22:00:05 2010 UTC

#	Line 6 | Line 6
6
7		package jsr166y;
8
9	<	import java.util.concurrent.*;
10	<	import java.util.concurrent.atomic.*;
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;
12	–	import sun.misc.Unsafe;
13	–	import java.lang.reflect.*;
13
14		/**
15	<	* A reusable synchronization barrier, similar in functionality to a
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	<	* <ul>
21	<	*
22	<	* <li> The number of parties synchronizing on a phaser may vary over
23	<	* time.  A task may register to be a party at any time, and may
24	<	* deregister upon arriving at the barrier.  As is the case with most
25	<	* basic synchronization constructs, registration and deregistration
26	<	* affect only internal counts; they do not establish any further
27	<	* internal bookkeeping, so tasks cannot query whether they are
28	<	* registered. (However, you can introduce such bookkeeping by
29	<	* subclassing this class.)
30	<	*
31	<	* <li> Each generation has an associated phase value, starting at
32	<	* zero, and advancing when all parties reach the barrier (wrapping
33	<	* around to zero after reaching {@code Integer.MAX_VALUE}).
34	<	*
35	<	* <li> Like a CyclicBarrier, a Phaser may be repeatedly awaited.
36	<	* Method {@code arriveAndAwaitAdvance} has effect analogous to
37	<	* {@code CyclicBarrier.await}.  However, Phasers separate two
38	<	* aspects of coordination, that may also be invoked independently:
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 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 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> Arriving at a barrier. Methods {@code arrive} and
48	<	*       {@code arriveAndDeregister} do not block, but return
49	<	*       the phase value current upon entry to the method.
50	<	*
51	<	*   <li> Awaiting others. Method {@code awaitAdvance} requires an
52	<	*       argument indicating the entry phase, and returns when the
53	<	*       barrier advances to a new phase.
54	<	* </ul>
55	<	*
56	<	*
57	<	* <li> Barrier actions, performed by the task triggering a phase
58	<	* advance while others may be waiting, are arranged by overriding
59	<	* method {@code onAdvance}, that also controls termination.
60	<	* Overriding this method may be used to similar but more flexible
61	<	* effect as providing a barrier action to a CyclicBarrier.
62	<	*
63	<	* <li> Phasers may enter a <em>termination</em> state in which all
64	<	* actions immediately return without updating phaser state or waiting
65	<	* for advance, and indicating (via a negative phase value) that
66	<	* execution is complete.  Termination is triggered by executing the
67	<	* overridable {@code onAdvance} method that is invoked each time the
68	<	* barrier is about to be tripped. When a Phaser is controlling an
69	<	* action with a fixed number of iterations, it is often convenient to
70	<	* override this method to cause termination when the current phase
71	<	* number reaches a threshold. Method {@code forceTermination} is also
72	<	* available to abruptly release waiting threads and allow them to
73	<	* terminate.
70	<	*
71	<	* <li> Phasers may be tiered to reduce contention. Phasers with large
72	<	* numbers of parties that would otherwise experience heavy
73	<	* synchronization contention costs may instead be arranged in trees.
74	<	* This will typically greatly increase throughput even though it
75	<	* incurs somewhat greater per-operation overhead.
76	<	*
77	<	* <li> By default, {@code awaitAdvance} continues to wait even if
78	<	* the waiting thread is interrupted. And unlike the case in
79	<	* CyclicBarriers, exceptions encountered while tasks wait
80	<	* interruptibly or with timeout do not change the state of the
81	<	* barrier. If necessary, you can perform any associated recovery
82	<	* within handlers of those exceptions, often after invoking
83	<	* {@code forceTermination}.
84	<	*
85	<	* <li>Phasers ensure lack of starvation when used by ForkJoinTasks.
47	>	*   <li> <b>Arrival.</b> Methods {@link #arrive} and
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 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 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},
72	>	*       which will ensure sufficient parallelism to execute tasks
73	>	*       when others are blocked waiting for a phase to advance.
74		*
75		* </ul>
76		*
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
113	+	* #getRegisteredParties} parties in total, of which {@link
114	+	* #getArrivedParties} have arrived at the current phase ({@link
115	+	* #getPhase}).  When the remaining ({@link #getUnarrivedParties})
116	+	* parties arrive, the phase advances.  The values returned by these
117	+	* methods may reflect transient states and so are not in general
118	+	* useful for synchronization control.  Method {@link #toString}
119	+	* returns snapshots of these state queries in a form convenient for
120	+	* informal monitoring.
121	+	*
122		* <p><b>Sample usages:</b>
123		*
124	<	* <p>A Phaser may be used instead of a {@code CountDownLatch} to control
125	<	* a one-shot action serving a variable number of parties. The typical
126	<	* idiom is for the method setting this up to first register, then
127	<	* start the actions, then deregister, as in:
128	<	*
129	<	* <pre>
130	<	*  void runTasks(List&lt;Runnable&gt; list) {
131	<	*    final Phaser phaser = new Phaser(1); // "1" to register self
132	<	*    for (Runnable r : list) {
133	<	*      phaser.register();
134	<	*      new Thread() {
135	<	*        public void run() {
136	<	*          phaser.arriveAndAwaitAdvance(); // await all creation
137	<	*          r.run();
138	<	*          phaser.arriveAndDeregister();   // signal completion
139	<	*        }
140	<	*      }.start();
124	>	* <p>A {@code Phaser} may be used instead of a {@code CountDownLatch}
125	>	* to control a one-shot action serving a variable number of parties.
126	>	* The typical idiom is for the method setting this up to first
127	>	* register, then start the actions, then deregister, as in:
128	>	*
129	>	*  <pre> {@code
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) {
134	>	*     phaser.register();
135	>	*     new Thread() {
136	>	*       public void run() {
137	>	*         phaser.arriveAndAwaitAdvance(); // await all creation
138	>	*         task.run();
139	>	*       }
140	>	*     }.start();
141		*   }
142		*
143	<	*   doSomethingOnBehalfOfWorkers();
144	<	*   phaser.arrive(); // allow threads to start
145	<	*   int p = phaser.arriveAndDeregister(); // deregister self  ...
113	<	*   p = phaser.awaitAdvance(p); // ... and await arrival
114	<	*   otherActions(); // do other things while tasks execute
115	<	*   phaser.awaitAdvance(p); // await final completion
116	<	* }
117	<	* </pre>
143	>	*   // allow threads to start and deregister self
144	>	*   phaser.arriveAndDeregister();
145	>	* }}</pre>
146		*
147		* <p>One way to cause a set of threads to repeatedly perform actions
148		* for a given number of iterations is to override {@code onAdvance}:
149		*
150	<	* <pre>
151	<	*  void startTasks(List&lt;Runnable&gt; list, final int iterations) {
152	<	*    final Phaser phaser = new Phaser() {
153	<	*       public boolean onAdvance(int phase, int registeredParties) {
154	<	*         return phase &gt;= iterations || registeredParties == 0;
150	>	*  <pre> {@code
151	>	* void startTasks(List<Runnable> tasks, final int iterations) {
152	>	*   final Phaser phaser = new Phaser() {
153	>	*     protected boolean onAdvance(int phase, int registeredParties) {
154	>	*       return phase >= iterations || registeredParties == 0;
155	>	*     }
156	>	*   };
157	>	*   phaser.register();
158	>	*   for (final Runnable task : tasks) {
159	>	*     phaser.register();
160	>	*     new Thread() {
161	>	*       public void run() {
162	>	*         do {
163	>	*           task.run();
164	>	*           phaser.arriveAndAwaitAdvance();
165	>	*         } while (!phaser.isTerminated());
166		*       }
167	<	*    };
129	<	*    phaser.register();
130	<	*    for (Runnable r : list) {
131	<	*      phaser.register();
132	<	*      new Thread() {
133	<	*        public void run() {
134	<	*           do {
135	<	*             r.run();
136	<	*             phaser.arriveAndAwaitAdvance();
137	<	*           } while(!phaser.isTerminated();
138	<	*        }
139	<	*      }.start();
167	>	*     }.start();
168		*   }
169		*   phaser.arriveAndDeregister(); // deregister self, don't wait
170	<	* }
171	<	* </pre>
170	>	* }}</pre>
171	>	*
172	>	* If the main task must later await termination, it
173	>	* may re-register and then execute a similar loop:
174	>	*  <pre> {@code
175	>	*   // ...
176	>	*   phaser.register();
177	>	*   while (!phaser.isTerminated())
178	>	*     phaser.arriveAndAwaitAdvance();}</pre>
179	>	*
180	>	* <p>Related constructions may be used to await particular phase numbers
181	>	* in contexts where you are sure that the phase will never wrap around
182	>	* {@code Integer.MAX_VALUE}. For example:
183	>	*
184	>	*  <pre> {@code
185	>	* void awaitPhase(Phaser phaser, int phase) {
186	>	*   int p = phaser.register(); // assumes caller not already registered
187	>	*   while (p < phase) {
188	>	*     if (phaser.isTerminated())
189	>	*       // ... deal with unexpected termination
190	>	*     else
191	>	*       p = phaser.arriveAndAwaitAdvance();
192	>	*   }
193	>	*   phaser.arriveAndDeregister();
194	>	* }}</pre>
195		*
196	<	* <p> To create a set of tasks using a tree of Phasers,
197	<	* you could use code of the following form, assuming a
198	<	* Task class with a constructor accepting a Phaser that
199	<	* it registers for upon construction:
200	<	* <pre>
201	<	*  void build(Task[] actions, int lo, int hi, Phaser b) {
202	<	*    int step = (hi - lo) / TASKS_PER_PHASER;
203	<	*    if (step &gt; 1) {
204	<	*       int i = lo;
205	<	*       while (i &lt; hi) {
206	<	*         int r = Math.min(i + step, hi);
207	<	*         build(actions, i, r, new Phaser(b));
208	<	*         i = r;
209	<	*       }
210	<	*    }
211	<	*    else {
212	<	*      for (int i = lo; i &lt; hi; ++i)
213	<	*        actions[i] = new Task(b);
214	<	*        // assumes new Task(b) performs b.register()
215	<	*    }
216	<	*  }
166	<	*  // .. initially called, for n tasks via
167	<	*  build(new Task[n], 0, n, new Phaser());
168	<	* </pre>
196	>	*
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[] 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(tasks, i, j, new Phaser(ph));
210	>	*     }
211	>	*   } else {
212	>	*     for (int i = lo; i < hi; ++i)
213	>	*       tasks[i] = new Task(ph);
214	>	*       // assumes new Task(ph) performs ph.register()
215	>	*   }
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		*
175	–	* </pre>
176	–	*
223		* <p><b>Implementation notes</b>: This implementation restricts the
224		* maximum number of parties to 65535. Attempts to register additional
225	<	* parties result in IllegalStateExceptions. However, you can and
225	>	* parties result in {@code IllegalStateException}. However, you can and
226		* should create tiered phasers to accommodate arbitrarily large sets
227		* of participants.
228	+	*
229	+	* @since 1.7
230	+	* @author Doug Lea
231		*/
232		public class Phaser {
233		/*
#	Line 188 | Line 237 | public class Phaser {
237		*/
238
239		/**
240	<	* Barrier state representation. Conceptually, a barrier contains
241	<	* four values:
240	>	* Primary state representation, holding four 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	>	* Except that a phaser with no registered parties is
248	>	* distinguished with 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	<	* * parties -- the number of parties to wait (16 bits)
257	<	* * unarrived -- the number of parties yet to hit barrier (16 bits)
258	<	* * phase -- the generation of the barrier (31 bits)
259	<	* * terminated -- set if barrier is terminated (1 bit)
260	<	*
199	<	* However, to efficiently maintain atomicity, these values are
200	<	* packed into a single (atomic) long. Termination uses the sign
201	<	* bit of 32 bit representation of phase, so phase is set to -1 on
202	<	* termination. Good performance relies on keeping state decoding
203	<	* and encoding simple, and keeping race windows short.
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	<	* Note: there are some cheats in arrive() that rely on unarrived
263	<	* count being lowest 16 bits.
262	>	* The phase of a subphaser is allowed to lag that of its
263	>	* ancestors until it is actually accessed.  Method reconcileState
264	>	* is usually attempted only only when the number of unarrived
265	>	* parties appears to be zero, which indicates a potential lag in
266	>	* updating phase after the root advanced.
267		*/
268		private volatile long state;
269
270	<	private static final int ushortBits = 16;
271	<	private static final int ushortMask = 0xffff;
272	<	private static final int phaseMask  = 0x7fffffff;
270	>	private static final int  MAX_PARTIES     = 0xffff;
271	>	private static final int  MAX_PHASE       = 0x7fffffff;
272	>	private static final int  PARTIES_SHIFT   = 16;
273	>	private static final int  PHASE_SHIFT     = 32;
274	>	private static final int  UNARRIVED_MASK  = 0xffff;      // to mask ints
275	>	private static final long PARTIES_MASK    = 0xffff0000L; // to mask longs
276	>	private static final long TERMINATION_BIT = 1L << 63;
277	>
278	>	// some special values
279	>	private static final int  ONE_ARRIVAL     = 1;
280	>	private static final int  ONE_PARTY       = 1 << PARTIES_SHIFT;
281	>	private static final int  EMPTY           = 1;
282	>
283	>	// The following unpacking methods are usually manually inlined
284
285		private static int unarrivedOf(long s) {
286	<	return (int)(s & ushortMask);
286	>	int counts = (int)s;
287	>	return (counts == EMPTY) ? 0 : counts & UNARRIVED_MASK;
288		}
289
290		private static int partiesOf(long s) {
291	<	return ((int)s) >>> 16;
291	>	return (int)s >>> PARTIES_SHIFT;
292		}
293
294		private static int phaseOf(long s) {
295	<	return (int)(s >>> 32);
295	>	return (int) (s >>> PHASE_SHIFT);
296		}
297
298		private static int arrivedOf(long s) {
299	<	return partiesOf(s) - unarrivedOf(s);
300	<	}
301	<
230	<	private static long stateFor(int phase, int parties, int unarrived) {
231	<	return ((((long)phase) << 32) | (((long)parties) << 16) |
232	<	(long)unarrived);
233	<	}
234	<
235	<	private static long trippedStateFor(int phase, int parties) {
236	<	long lp = (long)parties;
237	<	return (((long)phase) << 32) | (lp << 16) | lp;
238	<	}
239	<
240	<	/**
241	<	* Returns message string for bad bounds exceptions
242	<	*/
243	<	private static String badBounds(int parties, int unarrived) {
244	<	return ("Attempt to set " + unarrived +
245	<	" unarrived of " + parties + " parties");
299	>	int counts = (int)s;
300	>	return (counts == EMPTY) ? 0 :
301	>	(counts >>> PARTIES_SHIFT) - (counts & UNARRIVED_MASK);
302		}
303
304		/**
#	Line 251 | Line 307 | public class Phaser {
307		private final Phaser parent;
308
309		/**
310	<	* The root of Phaser tree. Equals this if not in a tree.  Used to
255	<	* support faster state push-down.
310	>	* The root of phaser tree. Equals this if not in a tree.
311		*/
312		private final Phaser root;
313
259	–	// Wait queues
260	–
314		/**
315		* Heads of Treiber stacks for waiting threads. To eliminate
316	<	* contention while releasing some threads while adding others, we
316	>	* contention when releasing some threads while adding others, we
317		* use two of them, alternating across even and odd phases.
318	+	* Subphasers share queues with root to speed up releases.
319		*/
320	<	private final AtomicReference<QNode> evenQ = new AtomicReference<QNode>();
321	<	private final AtomicReference<QNode> oddQ  = new AtomicReference<QNode>();
320	>	private final AtomicReference<QNode> evenQ;
321	>	private final AtomicReference<QNode> oddQ;
322
323		private AtomicReference<QNode> queueFor(int phase) {
324	<	return (phase & 1) == 0? evenQ : oddQ;
324	>	return ((phase & 1) == 0) ? evenQ : oddQ;
325	>	}
326	>
327	>	/**
328	>	* Returns message string for bounds exceptions on arrival.
329	>	*/
330	>	private String badArrive(long s) {
331	>	return "Attempted arrival of unregistered party for " +
332	>	stateToString(s);
333	>	}
334	>
335	>	/**
336	>	* Returns message string for bounds exceptions on registration.
337	>	*/
338	>	private String badRegister(long s) {
339	>	return "Attempt to register more than " +
340	>	MAX_PARTIES + " parties for " + stateToString(s);
341		}
342
343		/**
344	<	* Returns current state, first resolving lagged propagation from
345	<	* root if necessary.
344	>	* Main implementation for methods arrive and arriveAndDeregister.
345	>	* Manually tuned to speed up and minimize race windows for the
346	>	* common case of just decrementing unarrived field.
347	>	*
348	>	* @param deregister false for arrive, true for arriveAndDeregister
349		*/
350	<	private long getReconciledState() {
351	<	return parent == null? state : reconcileState();
350	>	private int doArrive(boolean deregister) {
351	>	int adj = deregister ? ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL;
352	>	final Phaser root = this.root;
353	>	for (;;) {
354	>	long s = (root == this) ? state : reconcileState();
355	>	int phase = (int)(s >>> PHASE_SHIFT);
356	>	int counts = (int)s;
357	>	int unarrived = (counts & UNARRIVED_MASK) - 1;
358	>	if (phase < 0)
359	>	return phase;
360	>	else if (counts == EMPTY || unarrived < 0) {
361	>	if (root == this || reconcileState() == s)
362	>	throw new IllegalStateException(badArrive(s));
363	>	}
364	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adj)) {
365	>	if (unarrived == 0) {
366	>	long n = s & PARTIES_MASK;  // base of next state
367	>	int nextUnarrived = ((int)n) >>> PARTIES_SHIFT;
368	>	if (root != this)
369	>	return parent.doArrive(nextUnarrived == 0);
370	>	if (onAdvance(phase, nextUnarrived))
371	>	n |= TERMINATION_BIT;
372	>	else if (nextUnarrived == 0)
373	>	n |= EMPTY;
374	>	else
375	>	n |= nextUnarrived;
376	>	n |= ((long)((phase + 1) & MAX_PHASE)) << PHASE_SHIFT;
377	>	UNSAFE.compareAndSwapLong(this, stateOffset, s, n);
378	>	releaseWaiters(phase);
379	>	}
380	>	return phase;
381	>	}
382	>	}
383		}
384
385		/**
386	<	* Recursively resolves state.
386	>	* Implementation of register, bulkRegister
387	>	*
388	>	* @param registrations number to add to both parties and
389	>	* unarrived fields. Must be greater than zero.
390	>	*/
391	>	private int doRegister(int registrations) {
392	>	// adjustment to state
393	>	long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
394	>	Phaser par = parent;
395	>	int phase;
396	>	for (;;) {
397	>	long s = state;
398	>	int counts = (int)s;
399	>	int parties = counts >>> PARTIES_SHIFT;
400	>	int unarrived = counts & UNARRIVED_MASK;
401	>	if (registrations > MAX_PARTIES - parties)
402	>	throw new IllegalStateException(badRegister(s));
403	>	else if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
404	>	break;
405	>	else if (counts != EMPTY) {             // not 1st registration
406	>	if (par == null || reconcileState() == s) {
407	>	if (unarrived == 0)             // wait out advance
408	>	root.internalAwaitAdvance(phase, null);
409	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset,
410	>	s, s + adj))
411	>	break;
412	>	}
413	>	}
414	>	else if (par == null) {                 // 1st root registration
415	>	long next = (((long) phase) << PHASE_SHIFT) | adj;
416	>	if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
417	>	break;
418	>	}
419	>	else {
420	>	synchronized (this) {               // 1st sub registration
421	>	if (state == s) {               // recheck under lock
422	>	par.doRegister(1);
423	>	do {                        // force current phase
424	>	phase = (int)(root.state >>> PHASE_SHIFT);
425	>	// assert phase < 0 || (int)state == EMPTY;
426	>	} while (!UNSAFE.compareAndSwapLong
427	>	(this, stateOffset, state,
428	>	(((long) phase) << PHASE_SHIFT) | adj));
429	>	break;
430	>	}
431	>	}
432	>	}
433	>	}
434	>	return phase;
435	>	}
436	>
437	>	/**
438	>	* Resolves lagged phase propagation from root if necessary.
439		*/
440		private long reconcileState() {
441	<	Phaser p = parent;
441	>	Phaser rt = root;
442		long s = state;
443	<	if (p != null) {
444	<	while (unarrivedOf(s) == 0 && phaseOf(s) != phaseOf(root.state)) {
445	<	long parentState = p.getReconciledState();
446	<	int parentPhase = phaseOf(parentState);
447	<	int phase = phaseOf(s = state);
448	<	if (phase != parentPhase) {
449	<	long next = trippedStateFor(parentPhase, partiesOf(s));
450	<	if (casState(s, next)) {
451	<	releaseWaiters(phase);
452	<	s = next;
453	<	}
443	>	if (rt != this) {
444	>	int phase;
445	>	while ((phase = (int)(rt.state >>> PHASE_SHIFT)) !=
446	>	(int)(s >>> PHASE_SHIFT)) {
447	>	// assert phase < 0 || unarrivedOf(s) == 0
448	>	long t;                             // to reread s
449	>	long p = s & PARTIES_MASK;          // unshifted parties field
450	>	long n = (((long) phase) << PHASE_SHIFT) | p;
451	>	if (phase >= 0) {
452	>	if (p == 0L)
453	>	n |= EMPTY;                 // reset to empty
454	>	else
455	>	n |= p >>> PARTIES_SHIFT;   // set unarr to parties
456		}
457	+	if ((t = state) == s &&
458	+	UNSAFE.compareAndSwapLong(this, stateOffset, s, s = n))
459	+	break;
460	+	s = t;
461		}
462		}
463		return s;
464		}
465
466		/**
467	<	* Creates a new Phaser without any initially registered parties,
468	<	* initial phase number 0, and no parent. Any thread using this
469	<	* Phaser will need to first register for it.
467	>	* Creates a new phaser with no initially registered parties, no
468	>	* parent, and initial phase number 0. Any thread using this
469	>	* phaser will need to first register for it.
470		*/
471		public Phaser() {
472	<	this(null);
472	>	this(null, 0);
473		}
474
475		/**
476	<	* Creates a new Phaser with the given numbers of registered
477	<	* unarrived parties, initial phase number 0, and no parent.
478	<	* @param parties the number of parties required to trip barrier.
476	>	* Creates a new phaser with the given number of registered
477	>	* unarrived parties, no parent, and initial phase number 0.
478	>	*
479	>	* @param parties the number of parties required to advance to the
480	>	* next phase
481		* @throws IllegalArgumentException if parties less than zero
482	<	* or greater than the maximum number of parties supported.
482	>	* or greater than the maximum number of parties supported
483		*/
484		public Phaser(int parties) {
485		this(null, parties);
486		}
487
488		/**
489	<	* Creates a new Phaser with the given parent, without any
490	<	* initially registered parties. If parent is non-null this phaser
491	<	* is registered with the parent and its initial phase number is
328	<	* the same as that of parent phaser.
329	<	* @param parent the parent phaser.
489	>	* Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
490	>	*
491	>	* @param parent the parent phaser
492		*/
493		public Phaser(Phaser parent) {
494	<	int phase = 0;
333	<	this.parent = parent;
334	<	if (parent != null) {
335	<	this.root = parent.root;
336	<	phase = parent.register();
337	<	}
338	<	else
339	<	this.root = this;
340	<	this.state = trippedStateFor(phase, 0);
494	>	this(parent, 0);
495		}
496
497		/**
498	<	* Creates a new Phaser with the given parent and numbers of
499	<	* registered unarrived parties. If parent is non-null this phaser
500	<	* is registered with the parent and its initial phase number is
501	<	* the same as that of parent phaser.
502	<	* @param parent the parent phaser.
503	<	* @param parties the number of parties required to trip barrier.
498	>	* Creates a new phaser with the given parent and number of
499	>	* registered unarrived parties.  When the given parent is non-null
500	>	* and the given number of parties is greater than zero, this
501	>	* child phaser is registered with its parent.
502	>	*
503	>	* @param parent the parent phaser
504	>	* @param parties the number of parties required to advance to the
505	>	* next phase
506		* @throws IllegalArgumentException if parties less than zero
507	<	* or greater than the maximum number of parties supported.
507	>	* or greater than the maximum number of parties supported
508		*/
509		public Phaser(Phaser parent, int parties) {
510	<	if (parties < 0 || parties > ushortMask)
510	>	if (parties >>> PARTIES_SHIFT != 0)
511		throw new IllegalArgumentException("Illegal number of parties");
512		int phase = 0;
513		this.parent = parent;
514		if (parent != null) {
515	<	this.root = parent.root;
516	<	phase = parent.register();
515	>	final Phaser root = parent.root;
516	>	this.root = root;
517	>	this.evenQ = root.evenQ;
518	>	this.oddQ = root.oddQ;
519	>	if (parties != 0)
520	>	phase = parent.doRegister(1);
521		}
522	<	else
522	>	else {
523		this.root = this;
524	<	this.state = trippedStateFor(phase, parties);
524	>	this.evenQ = new AtomicReference<QNode>();
525	>	this.oddQ = new AtomicReference<QNode>();
526	>	}
527	>	this.state = (parties == 0) ? (long) EMPTY :
528	>	((((long) phase) << PHASE_SHIFT) |
529	>	(((long) parties) << PARTIES_SHIFT) |
530	>	((long) parties));
531		}
532
533		/**
534	<	* Adds a new unarrived party to this phaser.
535	<	* @return the current barrier phase number upon registration
534	>	* Adds a new unarrived party to this phaser.  If an ongoing
535	>	* invocation of {@link #onAdvance} is in progress, this method
536	>	* may await its completion before returning.  If this phaser has
537	>	* a parent, and this phaser previously had no registered parties,
538	>	* this child phaser is also registered with its parent. If
539	>	* this phaser is terminated, the attempt to register has
540	>	* no effect, and a negative value is returned.
541	>	*
542	>	* @return the arrival phase number to which this registration
543	>	* applied.  If this value is negative, then this phaser has
544	>	* terminated, in which case registration has no effect.
545		* @throws IllegalStateException if attempting to register more
546	<	* than the maximum supported number of parties.
546	>	* than the maximum supported number of parties
547		*/
548		public int register() {
549		return doRegister(1);
#	Line 376 | Line 551 | public class Phaser {
551
552		/**
553		* Adds the given number of new unarrived parties to this phaser.
554	<	* @param parties the number of parties required to trip barrier.
555	<	* @return the current barrier phase number upon registration
554	>	* If an ongoing invocation of {@link #onAdvance} is in progress,
555	>	* this method may await its completion before returning.  If this
556	>	* phaser has a parent, and the given number of parties is greater
557	>	* than zero, and this phaser previously had no registered
558	>	* parties, this child phaser is also registered with its parent.
559	>	* If this phaser is terminated, the attempt to register has no
560	>	* effect, and a negative value is returned.
561	>	*
562	>	* @param parties the number of additional parties required to
563	>	* advance to the next phase
564	>	* @return the arrival phase number to which this registration
565	>	* applied.  If this value is negative, then this phaser has
566	>	* terminated, in which case registration has no effect.
567		* @throws IllegalStateException if attempting to register more
568	<	* than the maximum supported number of parties.
568	>	* than the maximum supported number of parties
569	>	* @throws IllegalArgumentException if {@code parties < 0}
570		*/
571		public int bulkRegister(int parties) {
572		if (parties < 0)
#	Line 390 | Line 577 | public class Phaser {
577		}
578
579		/**
580	<	* Shared code for register, bulkRegister
394	<	*/
395	<	private int doRegister(int registrations) {
396	<	int phase;
397	<	for (;;) {
398	<	long s = getReconciledState();
399	<	phase = phaseOf(s);
400	<	int unarrived = unarrivedOf(s) + registrations;
401	<	int parties = partiesOf(s) + registrations;
402	<	if (phase < 0)
403	<	break;
404	<	if (parties > ushortMask || unarrived > ushortMask)
405	<	throw new IllegalStateException(badBounds(parties, unarrived));
406	<	if (phase == phaseOf(root.state) &&
407	<	casState(s, stateFor(phase, parties, unarrived)))
408	<	break;
409	<	}
410	<	return phase;
411	<	}
412	<
413	<	/**
414	<	* Arrives at the barrier, but does not wait for others.  (You can
415	<	* in turn wait for others via {@link #awaitAdvance}).
580	>	* Arrives at this phaser, without waiting for others to arrive.
581		*
582	<	* @return the barrier phase number upon entry to this method, or a
583	<	* negative value if terminated;
582	>	* <p>It is a usage error for an unregistered party to invoke this
583	>	* method.  However, this error may result in an {@code
584	>	* IllegalStateException} only upon some subsequent operation on
585	>	* this phaser, if ever.
586	>	*
587	>	* @return the arrival phase number, or a negative value if terminated
588		* @throws IllegalStateException if not terminated and the number
589	<	* of unarrived parties would become negative.
589	>	* of unarrived parties would become negative
590		*/
591		public int arrive() {
592	<	int phase;
424	<	for (;;) {
425	<	long s = state;
426	<	phase = phaseOf(s);
427	<	if (phase < 0)
428	<	break;
429	<	int parties = partiesOf(s);
430	<	int unarrived = unarrivedOf(s) - 1;
431	<	if (unarrived > 0) {        // Not the last arrival
432	<	if (casState(s, s - 1)) // s-1 adds one arrival
433	<	break;
434	<	}
435	<	else if (unarrived == 0) {  // the last arrival
436	<	Phaser par = parent;
437	<	if (par == null) {      // directly trip
438	<	if (casState
439	<	(s,
440	<	trippedStateFor(onAdvance(phase, parties)? -1 :
441	<	((phase + 1) & phaseMask), parties))) {
442	<	releaseWaiters(phase);
443	<	break;
444	<	}
445	<	}
446	<	else {                  // cascade to parent
447	<	if (casState(s, s - 1)) { // zeroes unarrived
448	<	par.arrive();
449	<	reconcileState();
450	<	break;
451	<	}
452	<	}
453	<	}
454	<	else if (phase != phaseOf(root.state)) // or if unreconciled
455	<	reconcileState();
456	<	else
457	<	throw new IllegalStateException(badBounds(parties, unarrived));
458	<	}
459	<	return phase;
592	>	return doArrive(false);
593		}
594
595		/**
596	<	* Arrives at the barrier, and deregisters from it, without
597	<	* waiting for others. Deregistration reduces number of parties
598	<	* required to trip the barrier in future phases.  If this phaser
596	>	* Arrives at this phaser and deregisters from it without waiting
597	>	* for others to arrive. Deregistration reduces the number of
598	>	* parties required to advance in future phases.  If this phaser
599		* has a parent, and deregistration causes this phaser to have
600		* zero parties, this phaser is also deregistered from its parent.
601		*
602	<	* @return the current barrier phase number upon entry to
603	<	* this method, or a negative value if terminated;
602	>	* <p>It is a usage error for an unregistered party to invoke this
603	>	* method.  However, this error may result in an {@code
604	>	* IllegalStateException} only upon some subsequent operation on
605	>	* this phaser, if ever.
606	>	*
607	>	* @return the arrival phase number, or a negative value if terminated
608		* @throws IllegalStateException if not terminated and the number
609	<	* of registered or unarrived parties would become negative.
609	>	* of registered or unarrived parties would become negative
610		*/
611		public int arriveAndDeregister() {
612	<	// similar code to arrive, but too different to merge
476	<	Phaser par = parent;
477	<	int phase;
478	<	for (;;) {
479	<	long s = state;
480	<	phase = phaseOf(s);
481	<	if (phase < 0)
482	<	break;
483	<	int parties = partiesOf(s) - 1;
484	<	int unarrived = unarrivedOf(s) - 1;
485	<	if (parties >= 0) {
486	<	if (unarrived > 0 || (unarrived == 0 && par != null)) {
487	<	if (casState
488	<	(s,
489	<	stateFor(phase, parties, unarrived))) {
490	<	if (unarrived == 0) {
491	<	par.arriveAndDeregister();
492	<	reconcileState();
493	<	}
494	<	break;
495	<	}
496	<	continue;
497	<	}
498	<	if (unarrived == 0) {
499	<	if (casState
500	<	(s,
501	<	trippedStateFor(onAdvance(phase, parties)? -1 :
502	<	((phase + 1) & phaseMask), parties))) {
503	<	releaseWaiters(phase);
504	<	break;
505	<	}
506	<	continue;
507	<	}
508	<	if (par != null && phase != phaseOf(root.state)) {
509	<	reconcileState();
510	<	continue;
511	<	}
512	<	}
513	<	throw new IllegalStateException(badBounds(parties, unarrived));
514	<	}
515	<	return phase;
612	>	return doArrive(true);
613		}
614
615		/**
616	<	* Arrives at the barrier and awaits others. Equivalent in effect
617	<	* to {@code awaitAdvance(arrive())}.  If you instead need to
618	<	* await with interruption of timeout, and/or deregister upon
619	<	* arrival, you can arrange them using analogous constructions.
620	<	* @return the phase on entry to this method
616	>	* Arrives at this phaser and awaits others. Equivalent in effect
617	>	* to {@code awaitAdvance(arrive())}.  If you need to await with
618	>	* interruption or timeout, you can arrange this with an analogous
619	>	* construction using one of the other forms of the {@code
620	>	* awaitAdvance} method.  If instead you need to deregister upon
621	>	* arrival, use {@code awaitAdvance(arriveAndDeregister())}.
622	>	*
623	>	* <p>It is a usage error for an unregistered party to invoke this
624	>	* method.  However, this error may result in an {@code
625	>	* IllegalStateException} only upon some subsequent operation on
626	>	* this phaser, if ever.
627	>	*
628	>	* @return the arrival phase number, or the (negative)
629	>	* {@linkplain #getPhase() current phase} if terminated
630		* @throws IllegalStateException if not terminated and the number
631	<	* of unarrived parties would become negative.
631	>	* of unarrived parties would become negative
632		*/
633		public int arriveAndAwaitAdvance() {
634	<	return awaitAdvance(arrive());
634	>	// Specialization of doArrive+awaitAdvance eliminating some reads/paths
635	>	final Phaser root = this.root;
636	>	for (;;) {
637	>	long s = (root == this) ? state : reconcileState();
638	>	int phase = (int)(s >>> PHASE_SHIFT);
639	>	int counts = (int)s;
640	>	int unarrived = (counts & UNARRIVED_MASK) - 1;
641	>	if (phase < 0)
642	>	return phase;
643	>	else if (counts == EMPTY || unarrived < 0) {
644	>	if (reconcileState() == s)
645	>	throw new IllegalStateException(badArrive(s));
646	>	}
647	>	else if (UNSAFE.compareAndSwapLong(this, stateOffset, s,
648	>	s -= ONE_ARRIVAL)) {
649	>	if (unarrived != 0)
650	>	return root.internalAwaitAdvance(phase, null);
651	>	if (root != this)
652	>	return parent.arriveAndAwaitAdvance();
653	>	long n = s & PARTIES_MASK;  // base of next state
654	>	int nextUnarrived = ((int)n) >>> PARTIES_SHIFT;
655	>	if (onAdvance(phase, nextUnarrived))
656	>	n |= TERMINATION_BIT;
657	>	else if (nextUnarrived == 0)
658	>	n |= EMPTY;
659	>	else
660	>	n |= nextUnarrived;
661	>	int nextPhase = (phase + 1) & MAX_PHASE;
662	>	n |= (long)nextPhase << PHASE_SHIFT;
663	>	if (!UNSAFE.compareAndSwapLong(this, stateOffset, s, n))
664	>	return (int)(state >>> PHASE_SHIFT); // terminated
665	>	releaseWaiters(phase);
666	>	return nextPhase;
667	>	}
668	>	}
669		}
670
671		/**
672	<	* Awaits the phase of the barrier to advance from the given
673	<	* value, or returns immediately if argument is negative or this
674	<	* barrier is terminated.
675	<	* @param phase the phase on entry to this method
676	<	* @return the phase on exit from this method
672	>	* Awaits the phase of this phaser to advance from the given phase
673	>	* value, returning immediately if the current phase is not equal
674	>	* to the given phase value or this phaser is terminated.
675	>	*
676	>	* @param phase an arrival phase number, or negative value if
677	>	* terminated; this argument is normally the value returned by a
678	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
679	>	* @return the next arrival phase number, or the argument if it is
680	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
681	>	* if terminated
682		*/
683		public int awaitAdvance(int phase) {
684	+	final Phaser root = this.root;
685	+	long s = (root == this) ? state : reconcileState();
686	+	int p = (int)(s >>> PHASE_SHIFT);
687		if (phase < 0)
688		return phase;
689	<	long s = getReconciledState();
690	<	int p = phaseOf(s);
691	<	if (p != phase)
544	<	return p;
545	<	if (unarrivedOf(s) == 0 && parent != null)
546	<	parent.awaitAdvance(phase);
547	<	// Fall here even if parent waited, to reconcile and help release
548	<	return untimedWait(phase);
689	>	if (p == phase)
690	>	return root.internalAwaitAdvance(phase, null);
691	>	return p;
692		}
693
694		/**
695	<	* Awaits the phase of the barrier to advance from the given
696	<	* value, or returns immediately if argument is negative or this
697	<	* barrier is terminated, or throws InterruptedException if
698	<	* interrupted while waiting.
699	<	* @param phase the phase on entry to this method
700	<	* @return the phase on exit from this method
695	>	* Awaits the phase of this phaser to advance from the given phase
696	>	* value, throwing {@code InterruptedException} if interrupted
697	>	* while waiting, or returning immediately if the current phase is
698	>	* not equal to the given phase value or this phaser is
699	>	* terminated.
700	>	*
701	>	* @param phase an arrival phase number, or negative value if
702	>	* terminated; this argument is normally the value returned by a
703	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
704	>	* @return the next arrival phase number, or the argument if it is
705	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
706	>	* if terminated
707		* @throws InterruptedException if thread interrupted while waiting
708		*/
709	<	public int awaitAdvanceInterruptibly(int phase)
709	>	public int awaitAdvanceInterruptibly(int phase)
710		throws InterruptedException {
711	+	final Phaser root = this.root;
712	+	long s = (root == this) ? state : reconcileState();
713	+	int p = (int)(s >>> PHASE_SHIFT);
714		if (phase < 0)
715		return phase;
716	<	long s = getReconciledState();
717	<	int p = phaseOf(s);
718	<	if (p != phase)
719	<	return p;
720	<	if (unarrivedOf(s) == 0 && parent != null)
721	<	parent.awaitAdvanceInterruptibly(phase);
722	<	return interruptibleWait(phase);
716	>	if (p == phase) {
717	>	QNode node = new QNode(this, phase, true, false, 0L);
718	>	p = root.internalAwaitAdvance(phase, node);
719	>	if (node.wasInterrupted)
720	>	throw new InterruptedException();
721	>	}
722	>	return p;
723		}
724
725		/**
726	<	* Awaits the phase of the barrier to advance from the given value
727	<	* or the given timeout elapses, or returns immediately if
728	<	* argument is negative or this barrier is terminated.
729	<	* @param phase the phase on entry to this method
730	<	* @return the phase on exit from this method
726	>	* Awaits the phase of this phaser to advance from the given phase
727	>	* value or the given timeout to elapse, throwing {@code
728	>	* InterruptedException} if interrupted while waiting, or
729	>	* returning immediately if the current phase is not equal to the
730	>	* given phase value or this phaser is terminated.
731	>	*
732	>	* @param phase an arrival phase number, or negative value if
733	>	* terminated; this argument is normally the value returned by a
734	>	* previous call to {@code arrive} or {@code arriveAndDeregister}.
735	>	* @param timeout how long to wait before giving up, in units of
736	>	*        {@code unit}
737	>	* @param unit a {@code TimeUnit} determining how to interpret the
738	>	*        {@code timeout} parameter
739	>	* @return the next arrival phase number, or the argument if it is
740	>	* negative, or the (negative) {@linkplain #getPhase() current phase}
741	>	* if terminated
742		* @throws InterruptedException if thread interrupted while waiting
743		* @throws TimeoutException if timed out while waiting
744		*/
745	<	public int awaitAdvanceInterruptibly(int phase, long timeout, TimeUnit unit)
745	>	public int awaitAdvanceInterruptibly(int phase,
746	>	long timeout, TimeUnit unit)
747		throws InterruptedException, TimeoutException {
748	+	long nanos = unit.toNanos(timeout);
749	+	final Phaser root = this.root;
750	+	long s = (root == this) ? state : reconcileState();
751	+	int p = (int)(s >>> PHASE_SHIFT);
752		if (phase < 0)
753		return phase;
754	<	long s = getReconciledState();
755	<	int p = phaseOf(s);
756	<	if (p != phase)
757	<	return p;
758	<	if (unarrivedOf(s) == 0 && parent != null)
759	<	parent.awaitAdvanceInterruptibly(phase, timeout, unit);
760	<	return timedWait(phase, unit.toNanos(timeout));
754	>	if (p == phase) {
755	>	QNode node = new QNode(this, phase, true, true, nanos);
756	>	p = root.internalAwaitAdvance(phase, node);
757	>	if (node.wasInterrupted)
758	>	throw new InterruptedException();
759	>	else if (p == phase)
760	>	throw new TimeoutException();
761	>	}
762	>	return p;
763		}
764
765		/**
766	<	* Forces this barrier to enter termination state. Counts of
767	<	* arrived and registered parties are unaffected. If this phaser
768	<	* has a parent, it too is terminated. This method may be useful
769	<	* for coordinating recovery after one or more tasks encounter
766	>	* Forces this phaser to enter termination state.  Counts of
767	>	* registered parties are unaffected.  If this phaser is a member
768	>	* of a tiered set of phasers, then all of the phasers in the set
769	>	* are terminated.  If this phaser is already terminated, this
770	>	* method has no effect.  This method may be useful for
771	>	* coordinating recovery after one or more tasks encounter
772		* unexpected exceptions.
773		*/
774		public void forceTermination() {
775	<	for (;;) {
776	<	long s = getReconciledState();
777	<	int phase = phaseOf(s);
778	<	int parties = partiesOf(s);
779	<	int unarrived = unarrivedOf(s);
780	<	if (phase < 0 ||
781	<	casState(s, stateFor(-1, parties, unarrived))) {
775	>	// Only need to change root state
776	>	final Phaser root = this.root;
777	>	long s;
778	>	while ((s = root.state) >= 0) {
779	>	long next = (s & ~((long)UNARRIVED_MASK)) | TERMINATION_BIT;
780	>	if (UNSAFE.compareAndSwapLong(root, stateOffset, s, next)) {
781	>	// signal all threads
782		releaseWaiters(0);
783		releaseWaiters(1);
612	–	if (parent != null)
613	–	parent.forceTermination();
784		return;
785		}
786		}
#	Line 619 | Line 789 | public class Phaser {
789		/**
790		* Returns the current phase number. The maximum phase number is
791		* {@code Integer.MAX_VALUE}, after which it restarts at
792	<	* zero. Upon termination, the phase number is negative.
792	>	* zero. Upon termination, the phase number is negative,
793	>	* in which case the prevailing phase prior to termination
794	>	* may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
795	>	*
796		* @return the phase number, or a negative value if terminated
797		*/
798		public final int getPhase() {
799	<	return phaseOf(getReconciledState());
799	>	return (int)(root.state >>> PHASE_SHIFT);
800		}
801
802		/**
803	<	* Returns {@code true} if the current phase number equals the given phase.
804	<	* @param phase the phase
632	<	* @return {@code true} if the current phase number equals the given phase
633	<	*/
634	<	public final boolean hasPhase(int phase) {
635	<	return phaseOf(getReconciledState()) == phase;
636	<	}
637	<
638	<	/**
639	<	* Returns the number of parties registered at this barrier.
803	>	* Returns the number of parties registered at this phaser.
804	>	*
805		* @return the number of parties
806		*/
807		public int getRegisteredParties() {
#	Line 644 | Line 809 | public class Phaser {
809		}
810
811		/**
812	<	* Returns the number of parties that have arrived at the current
813	<	* phase of this barrier.
812	>	* Returns the number of registered parties that have arrived at
813	>	* the current phase of this phaser.
814	>	*
815		* @return the number of arrived parties
816		*/
817		public int getArrivedParties() {
818	<	return arrivedOf(state);
818	>	return arrivedOf(reconcileState());
819		}
820
821		/**
822		* Returns the number of registered parties that have not yet
823	<	* arrived at the current phase of this barrier.
823	>	* arrived at the current phase of this phaser.
824	>	*
825		* @return the number of unarrived parties
826		*/
827		public int getUnarrivedParties() {
828	<	return unarrivedOf(state);
828	>	return unarrivedOf(reconcileState());
829		}
830
831		/**
832	<	* Returns the parent of this phaser, or null if none.
833	<	* @return the parent of this phaser, or null if none
832	>	* Returns the parent of this phaser, or {@code null} if none.
833	>	*
834	>	* @return the parent of this phaser, or {@code null} if none
835		*/
836		public Phaser getParent() {
837		return parent;
#	Line 672 | Line 840 | public class Phaser {
840		/**
841		* Returns the root ancestor of this phaser, which is the same as
842		* this phaser if it has no parent.
843	+	*
844		* @return the root ancestor of this phaser
845		*/
846		public Phaser getRoot() {
#	Line 679 | Line 848 | public class Phaser {
848		}
849
850		/**
851	<	* Returns {@code true} if this barrier has been terminated.
852	<	* @return {@code true} if this barrier has been terminated
851	>	* Returns {@code true} if this phaser has been terminated.
852	>	*
853	>	* @return {@code true} if this phaser has been terminated
854		*/
855		public boolean isTerminated() {
856	<	return getPhase() < 0;
856	>	return root.state < 0L;
857		}
858
859		/**
860	<	* Overridable method to perform an action upon phase advance, and
861	<	* to control termination. This method is invoked whenever the
862	<	* barrier is tripped (and thus all other waiting parties are
863	<	* dormant). If it returns true, then, rather than advance the
864	<	* phase number, this barrier will be set to a final termination
865	<	* state, and subsequent calls to {@code isTerminated} will
866	<	* return true.
867	<	*
868	<	* <p> The default version returns true when the number of
869	<	* registered parties is zero. Normally, overrides that arrange
870	<	* termination for other reasons should also preserve this
871	<	* property.
872	<	*
873	<	* <p> You may override this method to perform an action with side
874	<	* effects visible to participating tasks, but it is in general
875	<	* only sensible to do so in designs where all parties register
876	<	* before any arrive, and all {@code awaitAdvance} at each phase.
877	<	* Otherwise, you cannot ensure lack of interference. In
878	<	* particular, this method may be invoked more than once per
879	<	* transition if other parties successfully register while the
880	<	* invocation of this method is in progress, thus postponing the
881	<	* transition until those parties also arrive, re-triggering this
882	<	* method.
860	>	* Overridable method to perform an action upon impending phase
861	>	* advance, and to control termination. This method is invoked
862	>	* upon arrival of the party advancing this phaser (when all other
863	>	* waiting parties are dormant).  If this method returns {@code
864	>	* true}, this phaser will be set to a final termination state
865	>	* upon advance, and subsequent calls to {@link #isTerminated}
866	>	* will return true. Any (unchecked) Exception or Error thrown by
867	>	* an invocation of this method is propagated to the party
868	>	* attempting to advance this phaser, in which case no advance
869	>	* occurs.
870	>	*
871	>	* <p>The arguments to this method provide the state of the phaser
872	>	* prevailing for the current transition.  The effects of invoking
873	>	* arrival, registration, and waiting methods on this phaser from
874	>	* within {@code onAdvance} are unspecified and should not be
875	>	* relied on.
876	>	*
877	>	* <p>If this phaser is a member of a tiered set of phasers, then
878	>	* {@code onAdvance} is invoked only for its root phaser on each
879	>	* advance.
880	>	*
881	>	* <p>To support the most common use cases, the default
882	>	* implementation of this method returns {@code true} when the
883	>	* number of registered parties has become zero as the result of a
884	>	* party invoking {@code arriveAndDeregister}.  You can disable
885	>	* this behavior, thus enabling continuation upon future
886	>	* registrations, by overriding this method to always return
887	>	* {@code false}:
888	>	*
889	>	* <pre> {@code
890	>	* Phaser phaser = new Phaser() {
891	>	*   protected boolean onAdvance(int phase, int parties) { return false; }
892	>	* }}</pre>
893		*
894	<	* @param phase the phase number on entering the barrier
894	>	* @param phase the current phase number on entry to this method,
895	>	* before this phaser is advanced
896		* @param registeredParties the current number of registered parties
897	<	* @return {@code true} if this barrier should terminate
897	>	* @return {@code true} if this phaser should terminate
898		*/
899		protected boolean onAdvance(int phase, int registeredParties) {
900	<	return registeredParties <= 0;
900	>	return registeredParties == 0;
901		}
902
903		/**
#	Line 726 | Line 907 | public class Phaser {
907		* followed by the number of registered parties, and {@code
908		* "arrived = "} followed by the number of arrived parties.
909		*
910	<	* @return a string identifying this barrier, as well as its state
910	>	* @return a string identifying this phaser, as well as its state
911		*/
912		public String toString() {
913	<	long s = getReconciledState();
913	>	return stateToString(reconcileState());
914	>	}
915	>
916	>	/**
917	>	* Implementation of toString and string-based error messages
918	>	*/
919	>	private String stateToString(long s) {
920		return super.toString() +
921		"[phase = " + phaseOf(s) +
922		" parties = " + partiesOf(s) +
923		" arrived = " + arrivedOf(s) + "]";
924		}
925
926	<	// methods for waiting
926	>	// Waiting mechanics
927	>
928	>	/**
929	>	* Removes and signals threads from queue for phase.
930	>	*/
931	>	private void releaseWaiters(int phase) {
932	>	QNode q;   // first element of queue
933	>	Thread t;  // its thread
934	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
935	>	while ((q = head.get()) != null &&
936	>	q.phase != (int)(root.state >>> PHASE_SHIFT)) {
937	>	if (head.compareAndSet(q, q.next) &&
938	>	(t = q.thread) != null) {
939	>	q.thread = null;
940	>	LockSupport.unpark(t);
941	>	}
942	>	}
943	>	}
944	>
945	>	/**
946	>	* Variant of releaseWaiters that additionally tries to remove any
947	>	* nodes no longer waiting for advance due to timeout or
948	>	* interrupt. Currently, nodes are removed only if they are at
949	>	* head of queue, which suffices to reduce memory footprint in
950	>	* most usages.
951	>	*
952	>	* @return current phase on exit
953	>	*/
954	>	private int abortWait(int phase) {
955	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
956	>	for (;;) {
957	>	Thread t;
958	>	QNode q = head.get();
959	>	int p = (int)(root.state >>> PHASE_SHIFT);
960	>	if (q == null || ((t = q.thread) != null && q.phase == p))
961	>	return p;
962	>	if (head.compareAndSet(q, q.next) && t != null) {
963	>	q.thread = null;
964	>	LockSupport.unpark(t);
965	>	}
966	>	}
967	>	}
968	>
969	>	/** The number of CPUs, for spin control */
970	>	private static final int NCPU = Runtime.getRuntime().availableProcessors();
971	>
972	>	/**
973	>	* The number of times to spin before blocking while waiting for
974	>	* advance, per arrival while waiting. On multiprocessors, fully
975	>	* blocking and waking up a large number of threads all at once is
976	>	* usually a very slow process, so we use rechargeable spins to
977	>	* avoid it when threads regularly arrive: When a thread in
978	>	* internalAwaitAdvance notices another arrival before blocking,
979	>	* and there appear to be enough CPUs available, it spins
980	>	* SPINS_PER_ARRIVAL more times before blocking. The value trades
981	>	* off good-citizenship vs big unnecessary slowdowns.
982	>	*/
983	>	static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
984	>
985	>	/**
986	>	* Possibly blocks and waits for phase to advance unless aborted.
987	>	* Call only from root node.
988	>	*
989	>	* @param phase current phase
990	>	* @param node if non-null, the wait node to track interrupt and timeout;
991	>	* if null, denotes noninterruptible wait
992	>	* @return current phase
993	>	*/
994	>	private int internalAwaitAdvance(int phase, QNode node) {
995	>	releaseWaiters(phase-1);          // ensure old queue clean
996	>	boolean queued = false;           // true when node is enqueued
997	>	int lastUnarrived = 0;            // to increase spins upon change
998	>	int spins = SPINS_PER_ARRIVAL;
999	>	long s;
1000	>	int p;
1001	>	while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
1002	>	if (node == null) {           // spinning in noninterruptible mode
1003	>	int unarrived = (int)s & UNARRIVED_MASK;
1004	>	if (unarrived != lastUnarrived &&
1005	>	(lastUnarrived = unarrived) < NCPU)
1006	>	spins += SPINS_PER_ARRIVAL;
1007	>	boolean interrupted = Thread.interrupted();
1008	>	if (interrupted || --spins < 0) { // need node to record intr
1009	>	node = new QNode(this, phase, false, false, 0L);
1010	>	node.wasInterrupted = interrupted;
1011	>	}
1012	>	}
1013	>	else if (node.isReleasable()) // done or aborted
1014	>	break;
1015	>	else if (!queued) {           // push onto queue
1016	>	AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
1017	>	QNode q = node.next = head.get();
1018	>	if ((q == null || q.phase == phase) &&
1019	>	(int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
1020	>	queued = head.compareAndSet(q, node);
1021	>	}
1022	>	else {
1023	>	try {
1024	>	ForkJoinPool.managedBlock(node);
1025	>	} catch (InterruptedException ie) {
1026	>	node.wasInterrupted = true;
1027	>	}
1028	>	}
1029	>	}
1030	>
1031	>	if (node != null) {
1032	>	if (node.thread != null)
1033	>	node.thread = null;       // avoid need for unpark()
1034	>	if (node.wasInterrupted && !node.interruptible)
1035	>	Thread.currentThread().interrupt();
1036	>	if (p == phase && (p = (int)(state >>> PHASE_SHIFT)) == phase)
1037	>	return abortWait(phase); // possibly clean up on abort
1038	>	}
1039	>	releaseWaiters(phase);
1040	>	return p;
1041	>	}
1042
1043		/**
1044		* Wait nodes for Treiber stack representing wait queue
#	Line 744 | Line 1046 | public class Phaser {
1046		static final class QNode implements ForkJoinPool.ManagedBlocker {
1047		final Phaser phaser;
1048		final int phase;
747	–	final long startTime;
748	–	final long nanos;
749	–	final boolean timed;
1049		final boolean interruptible;
1050	<	volatile boolean wasInterrupted = false;
1050	>	final boolean timed;
1051	>	boolean wasInterrupted;
1052	>	long nanos;
1053	>	long lastTime;
1054		volatile Thread thread; // nulled to cancel wait
1055		QNode next;
1056	+
1057		QNode(Phaser phaser, int phase, boolean interruptible,
1058	<	boolean timed, long startTime, long nanos) {
1058	>	boolean timed, long nanos) {
1059		this.phaser = phaser;
1060		this.phase = phase;
758	–	this.timed = timed;
1061		this.interruptible = interruptible;
760	–	this.startTime = startTime;
1062		this.nanos = nanos;
1063	+	this.timed = timed;
1064	+	this.lastTime = timed ? System.nanoTime() : 0L;
1065		thread = Thread.currentThread();
1066		}
1067	+
1068		public boolean isReleasable() {
1069	<	return (thread == null ||
1070	<	phaser.getPhase() != phase ||
1071	<	(interruptible && wasInterrupted) ||
1072	<	(timed && (nanos - (System.nanoTime() - startTime)) <= 0));
1073	<	}
770	<	public boolean block() {
771	<	if (Thread.interrupted()) {
772	<	wasInterrupted = true;
773	<	if (interruptible)
774	<	return true;
775	<	}
776	<	if (!timed)
777	<	LockSupport.park(this);
778	<	else {
779	<	long waitTime = nanos - (System.nanoTime() - startTime);
780	<	if (waitTime <= 0)
781	<	return true;
782	<	LockSupport.parkNanos(this, waitTime);
1069	>	if (thread == null)
1070	>	return true;
1071	>	if (phaser.getPhase() != phase) {
1072	>	thread = null;
1073	>	return true;
1074		}
1075	<	return isReleasable();
1076	<	}
1077	<	void signal() {
787	<	Thread t = thread;
788	<	if (t != null) {
1075	>	if (Thread.interrupted())
1076	>	wasInterrupted = true;
1077	>	if (wasInterrupted && interruptible) {
1078		thread = null;
1079	<	LockSupport.unpark(t);
1079	>	return true;
1080		}
1081	<	}
1082	<	boolean doWait() {
1083	<	if (thread != null) {
1084	<	try {
1085	<	ForkJoinPool.managedBlock(this, false);
1086	<	} catch (InterruptedException ie) {
1087	<	}
1081	>	if (timed) {
1082	>	if (nanos > 0L) {
1083	>	long now = System.nanoTime();
1084	>	nanos -= now - lastTime;
1085	>	lastTime = now;
1086	>	}
1087	>	if (nanos <= 0L) {
1088	>	thread = null;
1089	>	return true;
1090	>	}
1091		}
1092	<	return wasInterrupted;
1092	>	return false;
1093		}
1094
1095	<	}
1096	<
1097	<	/**
1098	<	* Removes and signals waiting threads from wait queue
1099	<	*/
1100	<	private void releaseWaiters(int phase) {
1101	<	AtomicReference<QNode> head = queueFor(phase);
1102	<	QNode q;
811	<	while ((q = head.get()) != null) {
812	<	if (head.compareAndSet(q, q.next))
813	<	q.signal();
1095	>	public boolean block() {
1096	>	if (isReleasable())
1097	>	return true;
1098	>	else if (!timed)
1099	>	LockSupport.park(this);
1100	>	else if (nanos > 0)
1101	>	LockSupport.parkNanos(this, nanos);
1102	>	return isReleasable();
1103		}
1104		}
1105
1106	<	/**
818	<	* Tries to enqueue given node in the appropriate wait queue
819	<	* @return true if successful
820	<	*/
821	<	private boolean tryEnqueue(QNode node) {
822	<	AtomicReference<QNode> head = queueFor(node.phase);
823	<	return head.compareAndSet(node.next = head.get(), node);
824	<	}
1106	>	// Unsafe mechanics
1107
1108	<	/**
1109	<	* Enqueues node and waits unless aborted or signalled.
1110	<	* @return current phase
829	<	*/
830	<	private int untimedWait(int phase) {
831	<	QNode node = null;
832	<	boolean queued = false;
833	<	boolean interrupted = false;
834	<	int p;
835	<	while ((p = getPhase()) == phase) {
836	<	if (Thread.interrupted())
837	<	interrupted = true;
838	<	else if (node == null)
839	<	node = new QNode(this, phase, false, false, 0, 0);
840	<	else if (!queued)
841	<	queued = tryEnqueue(node);
842	<	else
843	<	interrupted = node.doWait();
844	<	}
845	<	if (node != null)
846	<	node.thread = null;
847	<	releaseWaiters(phase);
848	<	if (interrupted)
849	<	Thread.currentThread().interrupt();
850	<	return p;
851	<	}
1108	>	private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1109	>	private static final long stateOffset =
1110	>	objectFieldOffset("state", Phaser.class);
1111
1112	<	/**
1113	<	* Interruptible version
1114	<	* @return current phase
1115	<	*/
1116	<	private int interruptibleWait(int phase) throws InterruptedException {
1117	<	QNode node = null;
1118	<	boolean queued = false;
1119	<	boolean interrupted = false;
1120	<	int p;
862	<	while ((p = getPhase()) == phase && !interrupted) {
863	<	if (Thread.interrupted())
864	<	interrupted = true;
865	<	else if (node == null)
866	<	node = new QNode(this, phase, true, false, 0, 0);
867	<	else if (!queued)
868	<	queued = tryEnqueue(node);
869	<	else
870	<	interrupted = node.doWait();
871	<	}
872	<	if (node != null)
873	<	node.thread = null;
874	<	if (p != phase || (p = getPhase()) != phase)
875	<	releaseWaiters(phase);
876	<	if (interrupted)
877	<	throw new InterruptedException();
878	<	return p;
1112	>	private static long objectFieldOffset(String field, Class<?> klazz) {
1113	>	try {
1114	>	return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1115	>	} catch (NoSuchFieldException e) {
1116	>	// Convert Exception to corresponding Error
1117	>	NoSuchFieldError error = new NoSuchFieldError(field);
1118	>	error.initCause(e);
1119	>	throw error;
1120	>	}
1121		}
1122
1123		/**
1124	<	* Timeout version.
1125	<	* @return current phase
1124	>	* Returns a sun.misc.Unsafe.  Suitable for use in a 3rd party package.
1125	>	* Replace with a simple call to Unsafe.getUnsafe when integrating
1126	>	* into a jdk.
1127	>	*
1128	>	* @return a sun.misc.Unsafe
1129		*/
1130	<	private int timedWait(int phase, long nanos)
886	<	throws InterruptedException, TimeoutException {
887	<	long startTime = System.nanoTime();
888	<	QNode node = null;
889	<	boolean queued = false;
890	<	boolean interrupted = false;
891	<	int p;
892	<	while ((p = getPhase()) == phase && !interrupted) {
893	<	if (Thread.interrupted())
894	<	interrupted = true;
895	<	else if (nanos - (System.nanoTime() - startTime) <= 0)
896	<	break;
897	<	else if (node == null)
898	<	node = new QNode(this, phase, true, true, startTime, nanos);
899	<	else if (!queued)
900	<	queued = tryEnqueue(node);
901	<	else
902	<	interrupted = node.doWait();
903	<	}
904	<	if (node != null)
905	<	node.thread = null;
906	<	if (p != phase || (p = getPhase()) != phase)
907	<	releaseWaiters(phase);
908	<	if (interrupted)
909	<	throw new InterruptedException();
910	<	if (p == phase)
911	<	throw new TimeoutException();
912	<	return p;
913	<	}
914	<
915	<	// Temporary Unsafe mechanics for preliminary release
916	<	private static Unsafe getUnsafe() throws Throwable {
1130	>	private static sun.misc.Unsafe getUnsafe() {
1131		try {
1132	<	return Unsafe.getUnsafe();
1132	>	return sun.misc.Unsafe.getUnsafe();
1133		} catch (SecurityException se) {
1134		try {
1135		return java.security.AccessController.doPrivileged
1136	<	(new java.security.PrivilegedExceptionAction<Unsafe>() {
1137	<	public Unsafe run() throws Exception {
1138	<	return getUnsafePrivileged();
1136	>	(new java.security
1137	>	.PrivilegedExceptionAction<sun.misc.Unsafe>() {
1138	>	public sun.misc.Unsafe run() throws Exception {
1139	>	java.lang.reflect.Field f = sun.misc
1140	>	.Unsafe.class.getDeclaredField("theUnsafe");
1141	>	f.setAccessible(true);
1142	>	return (sun.misc.Unsafe) f.get(null);
1143		}});
1144		} catch (java.security.PrivilegedActionException e) {
1145	<	throw e.getCause();
1145	>	throw new RuntimeException("Could not initialize intrinsics",
1146	>	e.getCause());
1147		}
1148		}
1149		}
931	–
932	–	private static Unsafe getUnsafePrivileged()
933	–	throws NoSuchFieldException, IllegalAccessException {
934	–	Field f = Unsafe.class.getDeclaredField("theUnsafe");
935	–	f.setAccessible(true);
936	–	return (Unsafe)f.get(null);
937	–	}
938	–
939	–	private static long fieldOffset(String fieldName)
940	–	throws NoSuchFieldException {
941	–	return _unsafe.objectFieldOffset
942	–	(Phaser.class.getDeclaredField(fieldName));
943	–	}
944	–
945	–	static final Unsafe _unsafe;
946	–	static final long stateOffset;
947	–
948	–	static {
949	–	try {
950	–	_unsafe = getUnsafe();
951	–	stateOffset = fieldOffset("state");
952	–	} catch (Exception e) {
953	–	throw new RuntimeException("Could not initialize intrinsics", e);
954	–	}
955	–	}
956	–
957	–	final boolean casState(long cmp, long val) {
958	–	return _unsafe.compareAndSwapLong(this, stateOffset, cmp, val);
959	–	}
1150		}

Diff Legend

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