ViewVC Help
View File | Revision Log | Show Annotations | Download File | Root Listing
root/jsr166/jsr166/src/jsr166y/Phaser.java
(Generate patch)

Comparing jsr166/src/jsr166y/Phaser.java (file contents):
Revision 1.3 by jsr166, Fri Jul 25 18:11:53 2008 UTC vs.
Revision 1.67 by jsr166, Fri Dec 3 21:29:34 2010 UTC

# Line 5 | Line 5
5   */
6  
7   package jsr166y;
8 < import jsr166y.forkjoin.*;
9 < import java.util.concurrent.*;
10 < 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;
13  
14   /**
15 < * A reusable synchronization barrier, similar in functionality to a
16 < * {@link java.util.concurrent.CyclicBarrier}, but supporting more
17 < * flexible 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 < * <ul>
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 < * <li> The number of parties synchronizing on the barrier may vary
34 < * over time.  A task may register to be a party in a barrier at any
35 < * time, and may deregister upon arriving at the barrier.  As is the
36 < * case with most basic synchronization constructs, registration
37 < * and deregistration affect only internal counts; they do not
38 < * establish any further internal bookkeeping, so tasks cannot query
39 < * whether they are registered.
40 < *
41 < * <li> Each generation has an associated phase value, starting at
42 < * zero, and advancing when all parties reach the barrier (wrapping
43 < * around to zero after reaching <tt>Integer.MAX_VALUE</tt>).
31 < *
32 < * <li> Like a CyclicBarrier, a Phaser may be repeatedly awaited.
33 < * Method <tt>arriveAndAwaitAdvance</tt> has effect analogous to
34 < * <tt>CyclicBarrier.await</tt>.  However, Phasers separate two
35 < * aspects of coordination, that may be invoked independently:
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 <tt>arrive</tt> and
48 < *       <tt>arriveAndDeregister</tt> do not block, but return
49 < *       the phase value on entry to the method.
50 < *
51 < *   <li> Awaiting others. Method <tt>awaitAdvance</tt> requires an
52 < *       argument indicating the entry phase, and returns when the
53 < *       barrier advances to a new phase.
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 in which all synchronization methods immediately return
79 + * without updating phaser state or waiting for advance, and
80 + * indicating (via a negative phase value) that execution is complete.
81 + * Termination is triggered when an invocation of {@code onAdvance}
82 + * returns {@code true}. The default implementation returns {@code
83 + * true} if a deregistration has caused the number of registered
84 + * parties to become zero.  As illustrated below, when phasers control
85 + * actions with a fixed number of iterations, it is often convenient
86 + * to override this method to cause termination when the current phase
87 + * number reaches a threshold. Method {@link #forceTermination} is
88 + * also available to abruptly release waiting threads and allow them
89 + * to terminate.
90   *
91 < * <li> Barrier actions, performed by the task triggering a phase
92 < * advance while others may be waiting, are arranged by overriding
93 < * method <tt>onAdvance</tt>, that also controls termination.
94 < *
95 < * <li> Phasers may enter a <em>termination</em> state in which all
96 < * await actions immediately return, indicating (via a negative phase
97 < * value) that execution is complete.  Termination is triggered by
56 < * executing the overridable <tt>onAdvance</tt> method that is invoked
57 < * each time the barrier is tripped. When a Phaser is controlling an
58 < * action with a fixed number of iterations, it is often convenient to
59 < * override this method to cause termination when the current phase
60 < * number reaches a threshold.  Method <tt>forceTermination</tt> is
61 < * also available to assist recovery actions upon failure.
62 < *
63 < * <li> Unlike most synchronizers, a Phaser may also be used with
64 < * ForkJoinTasks (as well as plain threads).
65 < *
66 < * <li> By default, <tt>awaitAdvance</tt> continues to wait even if
67 < * the current thread is interrupted. And unlike the case in
68 < * CyclicBarriers, exceptions encountered while tasks wait
69 < * interruptibly or with timeout do not change the state of the
70 < * barrier. If necessary, you can perform any associated recovery
71 < * within handlers of those exceptions.
91 > * <p> <b>Tiering.</b> Phasers may be <em>tiered</em> (i.e.,
92 > * constructed in tree structures) to reduce contention. Phasers with
93 > * large numbers of parties that would otherwise experience heavy
94 > * synchronization contention costs may instead be set up so that
95 > * groups of sub-phasers share a common parent.  This may greatly
96 > * increase throughput even though it incurs greater per-operation
97 > * overhead.
98   *
99 < * </ul>
99 > * <p><b>Monitoring.</b> While synchronization methods may be invoked
100 > * only by registered parties, the current state of a phaser may be
101 > * monitored by any caller.  At any given moment there are {@link
102 > * #getRegisteredParties} parties in total, of which {@link
103 > * #getArrivedParties} have arrived at the current phase ({@link
104 > * #getPhase}).  When the remaining ({@link #getUnarrivedParties})
105 > * parties arrive, the phase advances.  The values returned by these
106 > * methods may reflect transient states and so are not in general
107 > * useful for synchronization control.  Method {@link #toString}
108 > * returns snapshots of these state queries in a form convenient for
109 > * informal monitoring.
110 > *
111 > * <p><b>Sample usages:</b>
112 > *
113 > * <p>A {@code Phaser} may be used instead of a {@code CountDownLatch}
114 > * to control a one-shot action serving a variable number of parties.
115 > * The typical idiom is for the method setting this up to first
116 > * register, then start the actions, then deregister, as in:
117 > *
118 > *  <pre> {@code
119 > * void runTasks(List<Runnable> tasks) {
120 > *   final Phaser phaser = new Phaser(1); // "1" to register self
121 > *   // create and start threads
122 > *   for (Runnable task : tasks) {
123 > *     phaser.register();
124 > *     new Thread() {
125 > *       public void run() {
126 > *         phaser.arriveAndAwaitAdvance(); // await all creation
127 > *         task.run();
128 > *       }
129 > *     }.start();
130 > *   }
131 > *
132 > *   // allow threads to start and deregister self
133 > *   phaser.arriveAndDeregister();
134 > * }}</pre>
135 > *
136 > * <p>One way to cause a set of threads to repeatedly perform actions
137 > * for a given number of iterations is to override {@code onAdvance}:
138 > *
139 > *  <pre> {@code
140 > * void startTasks(List<Runnable> tasks, final int iterations) {
141 > *   final Phaser phaser = new Phaser() {
142 > *     protected boolean onAdvance(int phase, int registeredParties) {
143 > *       return phase >= iterations || registeredParties == 0;
144 > *     }
145 > *   };
146 > *   phaser.register();
147 > *   for (final Runnable task : tasks) {
148 > *     phaser.register();
149 > *     new Thread() {
150 > *       public void run() {
151 > *         do {
152 > *           task.run();
153 > *           phaser.arriveAndAwaitAdvance();
154 > *         } while (!phaser.isTerminated());
155 > *       }
156 > *     }.start();
157 > *   }
158 > *   phaser.arriveAndDeregister(); // deregister self, don't wait
159 > * }}</pre>
160   *
161 < * <p><b>Sample usage:</b>
161 > * If the main task must later await termination, it
162 > * may re-register and then execute a similar loop:
163 > *  <pre> {@code
164 > *   // ...
165 > *   phaser.register();
166 > *   while (!phaser.isTerminated())
167 > *     phaser.arriveAndAwaitAdvance();}</pre>
168   *
169 < * <p>[todo: non-FJ example]
169 > * <p>Related constructions may be used to await particular phase numbers
170 > * in contexts where you are sure that the phase will never wrap around
171 > * {@code Integer.MAX_VALUE}. For example:
172   *
173 < * <p> A Phaser may be used to support a style of programming in
174 < * which a task waits for others to complete, without otherwise
175 < * needing to keep track of which tasks it is waiting for. This is
176 < * similar to the "sync" construct in Cilk and "clocks" in X10.
177 < * Special constructions based on such barriers are available using
178 < * the <tt>LinkedAsyncAction</tt> and <tt>CyclicAction</tt> classes,
179 < * but they can be useful in other contexts as well.  For a simple
180 < * (but not very useful) example, here is a variant of Fibonacci:
87 < *
88 < * <pre>
89 < * class BarrierFibonacci extends RecursiveAction {
90 < *   int argument, result;
91 < *   final Phaser parentBarrier;
92 < *   BarrierFibonacci(int n, Phaser parentBarrier) {
93 < *     this.argument = n;
94 < *     this.parentBarrier = parentBarrier;
95 < *     parentBarrier.register();
173 > *  <pre> {@code
174 > * void awaitPhase(Phaser phaser, int phase) {
175 > *   int p = phaser.register(); // assumes caller not already registered
176 > *   while (p < phase) {
177 > *     if (phaser.isTerminated())
178 > *       // ... deal with unexpected termination
179 > *     else
180 > *       p = phaser.arriveAndAwaitAdvance();
181   *   }
182 < *   protected void compute() {
183 < *     int n = argument;
184 < *     if (n &lt;= 1)
185 < *        result = n;
186 < *     else {
187 < *        Phaser childBarrier = new Phaser(1);
188 < *        BarrierFibonacci f1 = new BarrierFibonacci(n - 1, childBarrier);
189 < *        BarrierFibonacci f2 = new BarrierFibonacci(n - 2, childBarrier);
190 < *        f1.fork();
191 < *        f2.fork();
192 < *        childBarrier.arriveAndAwait();
193 < *        result = f1.result + f2.result;
182 > *   phaser.arriveAndDeregister();
183 > * }}</pre>
184 > *
185 > *
186 > * <p>To create a set of {@code n} tasks using a tree of phasers, you
187 > * could use code of the following form, assuming a Task class with a
188 > * constructor accepting a {@code Phaser} that it registers with upon
189 > * construction. After invocation of {@code build(new Task[n], 0, n,
190 > * new Phaser())}, these tasks could then be started, for example by
191 > * submitting to a pool:
192 > *
193 > *  <pre> {@code
194 > * void build(Task[] tasks, int lo, int hi, Phaser ph) {
195 > *   if (hi - lo > TASKS_PER_PHASER) {
196 > *     for (int i = lo; i < hi; i += TASKS_PER_PHASER) {
197 > *       int j = Math.min(i + TASKS_PER_PHASER, hi);
198 > *       build(tasks, i, j, new Phaser(ph));
199   *     }
200 < *     parentBarrier.arriveAndDeregister();
200 > *   } else {
201 > *     for (int i = lo; i < hi; ++i)
202 > *       tasks[i] = new Task(ph);
203 > *       // assumes new Task(ph) performs ph.register()
204   *   }
205 < * }
206 < * </pre>
205 > * }}</pre>
206 > *
207 > * The best value of {@code TASKS_PER_PHASER} depends mainly on
208 > * expected synchronization rates. A value as low as four may
209 > * be appropriate for extremely small per-phase task bodies (thus
210 > * high rates), or up to hundreds for extremely large ones.
211   *
212   * <p><b>Implementation notes</b>: This implementation restricts the
213 < * maximum number of parties to 65535. Attempts to register
214 < * additional parties result in IllegalStateExceptions.
213 > * maximum number of parties to 65535. Attempts to register additional
214 > * parties result in {@code IllegalStateException}. However, you can and
215 > * should create tiered phasers to accommodate arbitrarily large sets
216 > * of participants.
217 > *
218 > * @since 1.7
219 > * @author Doug Lea
220   */
221   public class Phaser {
222      /*
223       * This class implements an extension of X10 "clocks".  Thanks to
224 <     * Vijay Saraswat for the idea of applying it to ForkJoinTasks,
225 <     * and to Vivek Sarkar for enhancements to extend functionality.
224 >     * Vijay Saraswat for the idea, and to Vivek Sarkar for
225 >     * enhancements to extend functionality.
226       */
227  
228      /**
229 <     * Barrier state representation. Conceptually, a barrier contains
128 <     * four values:
229 >     * Primary state representation, holding four fields:
230       *
231 <     * * parties -- the number of parties to wait (16 bits)
232 <     * * unarrived -- the number of parties yet to hit barrier (16 bits)
233 <     * * phase -- the generation of the barrier (31 bits)
234 <     * * terminated -- set if barrier is terminated (1 bit)
231 >     * * unarrived -- the number of parties yet to hit barrier (bits  0-15)
232 >     * * parties -- the number of parties to wait              (bits 16-31)
233 >     * * phase -- the generation of the barrier                (bits 32-62)
234 >     * * terminated -- set if barrier is terminated            (bit  63 / sign)
235       *
236 <     * However, to efficiently maintain atomicity, these values are
237 <     * packed into a single AtomicLong. Termination uses the sign bit
238 <     * of 32 bit representation of phase, so phase is set to -1 on
239 <     * termination.
240 <     */
241 <    private final AtomicLong state;
242 <
243 <    /**
244 <     * Head of Treiber stack for waiting nonFJ threads.
245 <     */
246 <    private final AtomicReference<QNode> head = new AtomicReference<QNode>();
236 >     * Except that a phaser with no registered parties is
237 >     * distinguished with the otherwise illegal state of having zero
238 >     * parties and one unarrived parties (encoded as EMPTY below).
239 >     *
240 >     * To efficiently maintain atomicity, these values are packed into
241 >     * a single (atomic) long. Good performance relies on keeping
242 >     * state decoding and encoding simple, and keeping race windows
243 >     * short.
244 >     *
245 >     * All state updates are performed via CAS except initial
246 >     * registration of a sub-phaser (i.e., one with a non-null
247 >     * parent).  In this (relatively rare) case, we use built-in
248 >     * synchronization to lock while first registering with its
249 >     * parent.
250 >     *
251 >     * The phase of a subphaser is allowed to lag that of its
252 >     * ancestors until it is actually accessed.  Method reconcileState
253 >     * is usually attempted only only when the number of unarrived
254 >     * parties appears to be zero, which indicates a potential lag in
255 >     * updating phase after the root advanced.
256 >     */
257 >    private volatile long state;
258 >
259 >    private static final int  MAX_PARTIES     = 0xffff;
260 >    private static final int  MAX_PHASE       = 0x7fffffff;
261 >    private static final int  PARTIES_SHIFT   = 16;
262 >    private static final int  PHASE_SHIFT     = 32;
263 >    private static final int  UNARRIVED_MASK  = 0xffff;      // to mask ints
264 >    private static final long PARTIES_MASK    = 0xffff0000L; // to mask longs
265 >    private static final long TERMINATION_BIT = 1L << 63;
266 >
267 >    // some special values
268 >    private static final int  ONE_ARRIVAL     = 1;
269 >    private static final int  ONE_PARTY       = 1 << PARTIES_SHIFT;
270 >    private static final int  EMPTY           = 1;
271  
272 <    private static final int ushortBits = 16;
148 <    private static final int ushortMask =  (1 << ushortBits) - 1;
149 <    private static final int phaseMask = 0x7fffffff;
272 >    // The following unpacking methods are usually manually inlined
273  
274      private static int unarrivedOf(long s) {
275 <        return (int)(s & ushortMask);
275 >        int counts = (int)s;
276 >        return (counts == EMPTY) ? 0 : counts & UNARRIVED_MASK;
277      }
278  
279      private static int partiesOf(long s) {
280 <        return (int)(s & (ushortMask << 16)) >>> 16;
280 >        int counts = (int)s;
281 >        return (counts == EMPTY) ? 0 : counts >>> PARTIES_SHIFT;
282      }
283  
284      private static int phaseOf(long s) {
285 <        return (int)(s >>> 32);
285 >        return (int) (s >>> PHASE_SHIFT);
286      }
287  
288      private static int arrivedOf(long s) {
289 <        return partiesOf(s) - unarrivedOf(s);
289 >        int counts = (int)s;
290 >        return (counts == EMPTY) ? 0 :
291 >            (counts >>> PARTIES_SHIFT) - (counts & UNARRIVED_MASK);
292 >    }
293 >
294 >    /**
295 >     * The parent of this phaser, or null if none
296 >     */
297 >    private final Phaser parent;
298 >
299 >    /**
300 >     * The root of phaser tree. Equals this if not in a tree.
301 >     */
302 >    private final Phaser root;
303 >
304 >    /**
305 >     * Heads of Treiber stacks for waiting threads. To eliminate
306 >     * contention when releasing some threads while adding others, we
307 >     * use two of them, alternating across even and odd phases.
308 >     * Subphasers share queues with root to speed up releases.
309 >     */
310 >    private final AtomicReference<QNode> evenQ;
311 >    private final AtomicReference<QNode> oddQ;
312 >
313 >    private AtomicReference<QNode> queueFor(int phase) {
314 >        return ((phase & 1) == 0) ? evenQ : oddQ;
315      }
316  
317 <    private static long stateFor(int phase, int parties, int unarrived) {
318 <        return (((long)phase) << 32) | ((parties << 16) | unarrived);
317 >    /**
318 >     * Returns message string for bounds exceptions on arrival.
319 >     */
320 >    private String badArrive(long s) {
321 >        return "Attempted arrival of unregistered party for " +
322 >            stateToString(s);
323 >    }
324 >
325 >    /**
326 >     * Returns message string for bounds exceptions on registration.
327 >     */
328 >    private String badRegister(long s) {
329 >        return "Attempt to register more than " +
330 >            MAX_PARTIES + " parties for " + stateToString(s);
331      }
332  
333 <    private static IllegalStateException badBounds(int parties, int unarrived) {
334 <        return new IllegalStateException("Attempt to set " + unarrived +
335 <                                         " unarrived of " + parties + " parties");
333 >    /**
334 >     * Main implementation for methods arrive and arriveAndDeregister.
335 >     * Manually tuned to speed up and minimize race windows for the
336 >     * common case of just decrementing unarrived field.
337 >     *
338 >     * @param deregister false for arrive, true for arriveAndDeregister
339 >     */
340 >    private int doArrive(boolean deregister) {
341 >        int adj = deregister ? ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL;
342 >        long s;
343 >        int phase;
344 >        while ((phase = (int)((s = state) >>> PHASE_SHIFT)) >= 0) {
345 >            int counts = (int)s;
346 >            int unarrived = counts & UNARRIVED_MASK;
347 >            if (counts == EMPTY || unarrived == 0) {
348 >                if (reconcileState() == s)
349 >                    throw new IllegalStateException(badArrive(s));
350 >            }
351 >            else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adj)) {
352 >                if (unarrived == 1) {
353 >                    long n = s & PARTIES_MASK;       // unshifted parties field
354 >                    int u = ((int)n) >>> PARTIES_SHIFT;
355 >                    Phaser par = parent;
356 >                    if (par != null) {
357 >                        par.doArrive(u == 0);
358 >                        reconcileState();
359 >                    }
360 >                    else {
361 >                        n |= (((long)((phase+1) & MAX_PHASE)) << PHASE_SHIFT);
362 >                        if (onAdvance(phase, u))
363 >                            n |= TERMINATION_BIT;
364 >                        else if (u == 0)
365 >                            n |= EMPTY;             // reset to unregistered
366 >                        else
367 >                            n |= (long)u;           // reset unarr to parties
368 >                        // assert state == s || isTerminated();
369 >                        UNSAFE.compareAndSwapLong(this, stateOffset, s, n);
370 >                        releaseWaiters(phase);
371 >                    }
372 >                }
373 >                break;
374 >            }
375 >        }
376 >        return phase;
377      }
378  
379      /**
380 <     * Creates a new Phaser without any initially registered parties,
381 <     * and initial phase number 0.
380 >     * Implementation of register, bulkRegister
381 >     *
382 >     * @param registrations number to add to both parties and
383 >     * unarrived fields. Must be greater than zero.
384 >     */
385 >    private int doRegister(int registrations) {
386 >        // adjustment to state
387 >        long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
388 >        Phaser par = parent;
389 >        int phase;
390 >        for (;;) {
391 >            long s = state;
392 >            int counts = (int)s;
393 >            int parties = counts >>> PARTIES_SHIFT;
394 >            int unarrived = counts & UNARRIVED_MASK;
395 >            if (registrations > MAX_PARTIES - parties)
396 >                throw new IllegalStateException(badRegister(s));
397 >            else if ((phase = (int)(s >>> PHASE_SHIFT)) < 0)
398 >                break;
399 >            else if (counts != EMPTY) {             // not 1st registration
400 >                if (par == null || reconcileState() == s) {
401 >                    if (unarrived == 0)             // wait out advance
402 >                        root.internalAwaitAdvance(phase, null);
403 >                    else if (UNSAFE.compareAndSwapLong(this, stateOffset,
404 >                                                       s, s + adj))
405 >                        break;
406 >                }
407 >            }
408 >            else if (par == null) {                 // 1st root registration
409 >                long next = (((long) phase) << PHASE_SHIFT) | adj;
410 >                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
411 >                    break;
412 >            }
413 >            else {
414 >                synchronized (this) {               // 1st sub registration
415 >                    if (state == s) {               // recheck under lock
416 >                        par.doRegister(1);
417 >                        do {                        // force current phase
418 >                            phase = (int)(root.state >>> PHASE_SHIFT);
419 >                            // assert phase < 0 || (int)state == EMPTY;
420 >                        } while (!UNSAFE.compareAndSwapLong
421 >                                 (this, stateOffset, state,
422 >                                  (((long) phase) << PHASE_SHIFT) | adj));
423 >                        break;
424 >                    }
425 >                }
426 >            }
427 >        }
428 >        return phase;
429 >    }
430 >
431 >    /**
432 >     * Resolves lagged phase propagation from root if necessary.
433 >     */
434 >    private long reconcileState() {
435 >        Phaser rt = root;
436 >        long s = state;
437 >        if (rt != this) {
438 >            int phase;
439 >            while ((phase = (int)(rt.state >>> PHASE_SHIFT)) !=
440 >                   (int)(s >>> PHASE_SHIFT)) {
441 >                // assert phase < 0 || unarrivedOf(s) == 0
442 >                long t;                             // to reread s
443 >                long p = s & PARTIES_MASK;          // unshifted parties field
444 >                long n = (((long) phase) << PHASE_SHIFT) | p;
445 >                if (phase >= 0) {
446 >                    if (p == 0L)
447 >                        n |= EMPTY;                 // reset to empty
448 >                    else
449 >                        n |= p >>> PARTIES_SHIFT;   // set unarr to parties
450 >                }
451 >                if ((t = state) == s &&
452 >                    UNSAFE.compareAndSwapLong(this, stateOffset, s, s = n))
453 >                    break;
454 >                s = t;
455 >            }
456 >        }
457 >        return s;
458 >    }
459 >
460 >    /**
461 >     * Creates a new phaser with no initially registered parties, no
462 >     * parent, and initial phase number 0. Any thread using this
463 >     * phaser will need to first register for it.
464       */
465      public Phaser() {
466 <        state = new AtomicLong(stateFor(0, 0, 0));
466 >        this(null, 0);
467      }
468  
469      /**
470 <     * Creates a new Phaser with the given numbers of registered
471 <     * unarrived parties and initial phase number 0.
472 <     * @param parties the number of parties required to trip barrier.
470 >     * Creates a new phaser with the given number of registered
471 >     * unarrived parties, no parent, and initial phase number 0.
472 >     *
473 >     * @param parties the number of parties required to advance to the
474 >     * next phase
475       * @throws IllegalArgumentException if parties less than zero
476 <     * or greater than the maximum number of parties supported.
476 >     * or greater than the maximum number of parties supported
477       */
478      public Phaser(int parties) {
479 <        if (parties < 0 || parties > ushortMask)
479 >        this(null, parties);
480 >    }
481 >
482 >    /**
483 >     * Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
484 >     *
485 >     * @param parent the parent phaser
486 >     */
487 >    public Phaser(Phaser parent) {
488 >        this(parent, 0);
489 >    }
490 >
491 >    /**
492 >     * Creates a new phaser with the given parent and number of
493 >     * registered unarrived parties. Registration and deregistration
494 >     * of this child phaser with its parent are managed automatically.
495 >     * If the given parent is non-null, whenever this child phaser has
496 >     * any registered parties (as established in this constructor,
497 >     * {@link #register}, or {@link #bulkRegister}), this child phaser
498 >     * is registered with its parent. Whenever the number of
499 >     * registered parties becomes zero as the result of an invocation
500 >     * of {@link #arriveAndDeregister}, this child phaser is
501 >     * deregistered from 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
508 >     */
509 >    public Phaser(Phaser parent, int parties) {
510 >        if (parties >>> PARTIES_SHIFT != 0)
511              throw new IllegalArgumentException("Illegal number of parties");
512 <        state = new AtomicLong(stateFor(0, parties, parties));
512 >        int phase = 0;
513 >        this.parent = parent;
514 >        if (parent != null) {
515 >            Phaser r = parent.root;
516 >            this.root = r;
517 >            this.evenQ = r.evenQ;
518 >            this.oddQ = r.oddQ;
519 >            if (parties != 0)
520 >                phase = parent.doRegister(1);
521 >        }
522 >        else {
523 >            this.root = this;
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 phaser is also registered with its parent.
539 >     *
540 >     * @return the arrival phase number to which this registration applied
541       * @throws IllegalStateException if attempting to register more
542 <     * than the maximum supported number of parties.
542 >     * than the maximum supported number of parties
543       */
544 <    public int register() { // increment both parties and unarrived
545 <        final AtomicLong state = this.state;
205 <        for (;;) {
206 <            long s = state.get();
207 <            int phase = phaseOf(s);
208 <            int parties = partiesOf(s) + 1;
209 <            int unarrived = unarrivedOf(s) + 1;
210 <            if (parties > ushortMask || unarrived > ushortMask)
211 <                throw badBounds(parties, unarrived);
212 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived)))
213 <                return phase;
214 <        }
544 >    public int register() {
545 >        return doRegister(1);
546      }
547  
548      /**
549 <     * Arrives at the barrier, but does not wait for others.  (You can
550 <     * in turn wait for others via {@link #awaitAdvance}).
549 >     * Adds the given number of new unarrived parties to this phaser.
550 >     * If an ongoing invocation of {@link #onAdvance} is in progress,
551 >     * this method may await its completion before returning.  If this
552 >     * phaser has a parent, and the given number of parties is
553 >     * greater than zero, and this phaser previously had no registered
554 >     * parties, this phaser is also registered with its parent.
555       *
556 <     * @return the current barrier phase number upon entry to
557 <     * this method, or a negative value if terminated;
558 <     * @throws IllegalStateException if the number of unarrived
559 <     * parties would become negative.
556 >     * @param parties the number of additional parties required to
557 >     * advance to the next phase
558 >     * @return the arrival phase number to which this registration applied
559 >     * @throws IllegalStateException if attempting to register more
560 >     * than the maximum supported number of parties
561 >     * @throws IllegalArgumentException if {@code parties < 0}
562       */
563 <    public int arrive() { // decrement unarrived. If zero, trip
564 <        final AtomicLong state = this.state;
565 <        for (;;) {
566 <            long s = state.get();
567 <            int phase = phaseOf(s);
568 <            int parties = partiesOf(s);
232 <            int unarrived = unarrivedOf(s) - 1;
233 <            if (unarrived < 0)
234 <                throw badBounds(parties, unarrived);
235 <            if (unarrived == 0 && phase >= 0) {
236 <                trip(phase, parties);
237 <                return phase;
238 <            }
239 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived)))
240 <                return phase;
241 <        }
563 >    public int bulkRegister(int parties) {
564 >        if (parties < 0)
565 >            throw new IllegalArgumentException();
566 >        if (parties == 0)
567 >            return getPhase();
568 >        return doRegister(parties);
569      }
570  
571      /**
572 <     * Arrives at the barrier, and deregisters from it, without
246 <     * waiting for others.
572 >     * Arrives at this phaser, without waiting for others to arrive.
573       *
574 <     * @return the current barrier phase number upon entry to
575 <     * this method, or a negative value if terminated;
576 <     * @throws IllegalStateException if the number of registered or
577 <     * unarrived parties would become negative.
574 >     * <p>It is a usage error for an unregistered party to invoke this
575 >     * method.  However, this error may result in an {@code
576 >     * IllegalStateException} only upon some subsequent operation on
577 >     * this phaser, if ever.
578 >     *
579 >     * @return the arrival phase number, or a negative value if terminated
580 >     * @throws IllegalStateException if not terminated and the number
581 >     * of unarrived parties would become negative
582       */
583 <    public int arriveAndDeregister() { // Same as arrive, plus decrement parties
584 <        final AtomicLong state = this.state;
255 <        for (;;) {
256 <            long s = state.get();
257 <            int phase = phaseOf(s);
258 <            int parties = partiesOf(s) - 1;
259 <            int unarrived = unarrivedOf(s) - 1;
260 <            if (parties < 0 || unarrived < 0)
261 <                throw badBounds(parties, unarrived);
262 <            if (unarrived == 0 && phase >= 0) {
263 <                trip(phase, parties);
264 <                return phase;
265 <            }
266 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived)))
267 <                return phase;
268 <        }
583 >    public int arrive() {
584 >        return doArrive(false);
585      }
586  
587      /**
588 <     * Arrives at the barrier and awaits others. Unlike other arrival
589 <     * methods, this method returns the arrival index of the
590 <     * caller. The caller tripping the barrier returns zero, the
591 <     * previous caller 1, and so on.
592 <     * @return the arrival index
593 <     * @throws IllegalStateException if the number of unarrived
594 <     * parties would become negative.
588 >     * Arrives at this phaser and deregisters from it without waiting
589 >     * for others to arrive. Deregistration reduces the number of
590 >     * parties required to advance in future phases.  If this phaser
591 >     * has a parent, and deregistration causes this phaser to have
592 >     * zero parties, this phaser is also deregistered from its parent.
593 >     *
594 >     * <p>It is a usage error for an unregistered party to invoke this
595 >     * method.  However, this error may result in an {@code
596 >     * IllegalStateException} only upon some subsequent operation on
597 >     * this phaser, if ever.
598 >     *
599 >     * @return the arrival phase number, or a negative value if terminated
600 >     * @throws IllegalStateException if not terminated and the number
601 >     * of registered or unarrived parties would become negative
602 >     */
603 >    public int arriveAndDeregister() {
604 >        return doArrive(true);
605 >    }
606 >
607 >    /**
608 >     * Arrives at this phaser and awaits others. Equivalent in effect
609 >     * to {@code awaitAdvance(arrive())}.  If you need to await with
610 >     * interruption or timeout, you can arrange this with an analogous
611 >     * construction using one of the other forms of the {@code
612 >     * awaitAdvance} method.  If instead you need to deregister upon
613 >     * arrival, use {@code awaitAdvance(arriveAndDeregister())}.
614 >     *
615 >     * <p>It is a usage error for an unregistered party to invoke this
616 >     * method.  However, this error may result in an {@code
617 >     * IllegalStateException} only upon some subsequent operation on
618 >     * this phaser, if ever.
619 >     *
620 >     * @return the arrival phase number, or a negative number if terminated
621 >     * @throws IllegalStateException if not terminated and the number
622 >     * of unarrived parties would become negative
623       */
624      public int arriveAndAwaitAdvance() {
625 <        final AtomicLong state = this.state;
282 <        for (;;) {
283 <            long s = state.get();
284 <            int phase = phaseOf(s);
285 <            int parties = partiesOf(s);
286 <            int unarrived = unarrivedOf(s) - 1;
287 <            if (unarrived < 0)
288 <                throw badBounds(parties, unarrived);
289 <            if (unarrived == 0 && phase >= 0) {
290 <                trip(phase, parties);
291 <                return 0;
292 <            }
293 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived))) {
294 <                awaitAdvance(phase);
295 <                return unarrived;
296 <            }
297 <        }
625 >        return awaitAdvance(doArrive(false));
626      }
627  
628      /**
629 <     * Awaits the phase of the barrier to advance from the given
630 <     * value, or returns immediately if this barrier is terminated.
631 <     * @param phase the phase on entry to this method
632 <     * @return the phase on exit from this method
629 >     * Awaits the phase of this phaser to advance from the given phase
630 >     * value, returning immediately if the current phase is not equal
631 >     * to the given phase value or this phaser is terminated.
632 >     *
633 >     * @param phase an arrival phase number, or negative value if
634 >     * terminated; this argument is normally the value returned by a
635 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
636 >     * @return the next arrival phase number, or a negative value
637 >     * if terminated or argument is negative
638       */
639      public int awaitAdvance(int phase) {
640 +        Phaser rt;
641 +        int p = (int)(state >>> PHASE_SHIFT);
642          if (phase < 0)
643              return phase;
644 <        Thread current = Thread.currentThread();
645 <        if (current instanceof ForkJoinWorkerThread)
646 <            return helpingWait(phase);
647 <        if (untimedWait(current, phase, false))
648 <            current.interrupt();
649 <        return phaseOf(state.get());
644 >        if (p == phase) {
645 >            if ((p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase)
646 >                return rt.internalAwaitAdvance(phase, null);
647 >            reconcileState();
648 >        }
649 >        return p;
650      }
651  
652      /**
653 <     * Awaits the phase of the barrier to advance from the given
654 <     * value, or returns immediately if this barrier is terminated, or
655 <     * throws InterruptedException if interrupted while waiting.
656 <     * @param phase the phase on entry to this method
657 <     * @return the phase on exit from this method
653 >     * Awaits the phase of this phaser to advance from the given phase
654 >     * value, throwing {@code InterruptedException} if interrupted
655 >     * while waiting, or returning immediately if the current phase is
656 >     * not equal to the given phase value or this phaser is
657 >     * terminated.
658 >     *
659 >     * @param phase an arrival phase number, or negative value if
660 >     * terminated; this argument is normally the value returned by a
661 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
662 >     * @return the next arrival phase number, or a negative value
663 >     * if terminated or argument is negative
664       * @throws InterruptedException if thread interrupted while waiting
665       */
666 <    public int awaitAdvanceInterruptibly(int phase) throws InterruptedException {
666 >    public int awaitAdvanceInterruptibly(int phase)
667 >        throws InterruptedException {
668 >        Phaser rt;
669 >        int p = (int)(state >>> PHASE_SHIFT);
670          if (phase < 0)
671              return phase;
672 <        Thread current = Thread.currentThread();
673 <        if (current instanceof ForkJoinWorkerThread)
674 <            return helpingWait(phase);
675 <        else if (Thread.interrupted() || untimedWait(current, phase, true))
676 <            throw new InterruptedException();
677 <        else
678 <            return phaseOf(state.get());
672 >        if (p == phase) {
673 >            if ((p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase) {
674 >                QNode node = new QNode(this, phase, true, false, 0L);
675 >                p = rt.internalAwaitAdvance(phase, node);
676 >                if (node.wasInterrupted)
677 >                    throw new InterruptedException();
678 >            }
679 >            else
680 >                reconcileState();
681 >        }
682 >        return p;
683      }
684  
685      /**
686 <     * Awaits the phase of the barrier to advance from the given value
687 <     * or the given timeout elapses, or returns immediately if this
688 <     * barrier is terminated.
689 <     * @param phase the phase on entry to this method
690 <     * @return the phase on exit from this method
686 >     * Awaits the phase of this phaser to advance from the given phase
687 >     * value or the given timeout to elapse, throwing {@code
688 >     * InterruptedException} if interrupted while waiting, or
689 >     * returning immediately if the current phase is not equal to the
690 >     * given phase value or this phaser is terminated.
691 >     *
692 >     * @param phase an arrival phase number, or negative value if
693 >     * terminated; this argument is normally the value returned by a
694 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
695 >     * @param timeout how long to wait before giving up, in units of
696 >     *        {@code unit}
697 >     * @param unit a {@code TimeUnit} determining how to interpret the
698 >     *        {@code timeout} parameter
699 >     * @return the next arrival phase number, or a negative value
700 >     * if terminated or argument is negative
701       * @throws InterruptedException if thread interrupted while waiting
702       * @throws TimeoutException if timed out while waiting
703       */
704 <    public int awaitAdvanceInterruptibly(int phase, long timeout, TimeUnit unit)
704 >    public int awaitAdvanceInterruptibly(int phase,
705 >                                         long timeout, TimeUnit unit)
706          throws InterruptedException, TimeoutException {
707 +        long nanos = unit.toNanos(timeout);
708 +        Phaser rt;
709 +        int p = (int)(state >>> PHASE_SHIFT);
710          if (phase < 0)
711              return phase;
712 <        long nanos = unit.toNanos(timeout);
713 <        Thread current = Thread.currentThread();
714 <        if (current instanceof ForkJoinWorkerThread)
715 <            return timedHelpingWait(phase, nanos);
716 <        timedWait(current, phase, nanos);
717 <        return phaseOf(state.get());
712 >        if (p == phase) {
713 >            if ((p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase) {
714 >                QNode node = new QNode(this, phase, true, true, nanos);
715 >                p = rt.internalAwaitAdvance(phase, node);
716 >                if (node.wasInterrupted)
717 >                    throw new InterruptedException();
718 >                else if (p == phase)
719 >                    throw new TimeoutException();
720 >            }
721 >            else
722 >                reconcileState();
723 >        }
724 >        return p;
725      }
726  
727      /**
728 <     * Forces this barrier to enter termination state. Counts of
729 <     * arrived and registered parties are unaffected. This method may
730 <     * be useful for coordinating recovery after one or more tasks
731 <     * encounter unexpected exceptions.
728 >     * Forces this phaser to enter termination state.  Counts of
729 >     * registered parties are unaffected.  If this phaser is a member
730 >     * of a tiered set of phasers, then all of the phasers in the set
731 >     * are terminated.  If this phaser is already terminated, this
732 >     * method has no effect.  This method may be useful for
733 >     * coordinating recovery after one or more tasks encounter
734 >     * unexpected exceptions.
735       */
736      public void forceTermination() {
737 <        final AtomicLong state = this.state;
738 <        for (;;) {
739 <            long s = state.get();
740 <            int phase = phaseOf(s);
741 <            int parties = partiesOf(s);
742 <            int unarrived = unarrivedOf(s);
743 <            if (phase < 0 ||
744 <                state.compareAndSet(s, stateFor(-1, parties, unarrived))) {
373 <                if (head.get() != null)
374 <                    releaseWaiters(-1);
737 >        // Only need to change root state
738 >        final Phaser root = this.root;
739 >        long s;
740 >        while ((s = root.state) >= 0) {
741 >            long next = (s & ~(long)(MAX_PARTIES)) | TERMINATION_BIT;
742 >            if (UNSAFE.compareAndSwapLong(root, stateOffset, s, next)) {
743 >                releaseWaiters(0); // signal all threads
744 >                releaseWaiters(1);
745                  return;
746              }
747          }
748      }
749  
750      /**
381     * Resets the barrier with the given numbers of registered unarrived
382     * parties and phase number 0. This method allows repeated reuse
383     * of this barrier, but only if it is somehow known not to be in
384     * use for other purposes.
385     * @param parties the number of parties required to trip barrier.
386     * @throws IllegalArgumentException if parties less than zero
387     * or greater than the maximum number of parties supported.
388     */
389    public void reset(int parties) {
390        if (parties < 0 || parties > ushortMask)
391            throw new IllegalArgumentException("Illegal number of parties");
392        state.set(stateFor(0, parties, parties));
393        if (head.get() != null)
394            releaseWaiters(0);
395    }
396
397    /**
751       * Returns the current phase number. The maximum phase number is
752 <     * <tt>Integer.MAX_VALUE</tt>, after which it restarts at
753 <     * zero. Upon termination, the phase number is negative.
752 >     * {@code Integer.MAX_VALUE}, after which it restarts at
753 >     * zero. Upon termination, the phase number is negative,
754 >     * in which case the prevailing phase prior to termination
755 >     * may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
756 >     *
757       * @return the phase number, or a negative value if terminated
758       */
759 <    public int getPhase() {
760 <        return phaseOf(state.get());
759 >    public final int getPhase() {
760 >        return (int)(root.state >>> PHASE_SHIFT);
761      }
762  
763      /**
764 <     * Returns the number of parties registered at this barrier.
764 >     * Returns the number of parties registered at this phaser.
765 >     *
766       * @return the number of parties
767       */
768      public int getRegisteredParties() {
769 <        return partiesOf(state.get());
769 >        return partiesOf(state);
770      }
771  
772      /**
773 <     * Returns the number of parties that have arrived at the current
774 <     * phase of this barrier.
773 >     * Returns the number of registered parties that have arrived at
774 >     * the current phase of this phaser.
775 >     *
776       * @return the number of arrived parties
777       */
778      public int getArrivedParties() {
779 <        return arrivedOf(state.get());
779 >        return arrivedOf(reconcileState());
780      }
781  
782      /**
783       * Returns the number of registered parties that have not yet
784 <     * arrived at the current phase of this barrier.
784 >     * arrived at the current phase of this phaser.
785 >     *
786       * @return the number of unarrived parties
787       */
788      public int getUnarrivedParties() {
789 <        return unarrivedOf(state.get());
789 >        return unarrivedOf(reconcileState());
790      }
791  
792      /**
793 <     * Returns true if this barrier has been terminated.
794 <     * @return true if this barrier has been terminated
793 >     * Returns the parent of this phaser, or {@code null} if none.
794 >     *
795 >     * @return the parent of this phaser, or {@code null} if none
796 >     */
797 >    public Phaser getParent() {
798 >        return parent;
799 >    }
800 >
801 >    /**
802 >     * Returns the root ancestor of this phaser, which is the same as
803 >     * this phaser if it has no parent.
804 >     *
805 >     * @return the root ancestor of this phaser
806 >     */
807 >    public Phaser getRoot() {
808 >        return root;
809 >    }
810 >
811 >    /**
812 >     * Returns {@code true} if this phaser has been terminated.
813 >     *
814 >     * @return {@code true} if this phaser has been terminated
815       */
816      public boolean isTerminated() {
817 <        return phaseOf(state.get()) < 0;
817 >        return root.state < 0L;
818      }
819  
820      /**
821 <     * Overridable method to perform an action upon phase advance, and
822 <     * to control termination. This method is invoked whenever the
823 <     * barrier is tripped (and thus all other waiting parties are
824 <     * dormant). If it returns true, then, rather than advance the
825 <     * phase number, this barrier will be set to a final termination
826 <     * state, and subsequent calls to <tt>isTerminated</tt> will
827 <     * return true.
828 <     *
829 <     * <p> The default version returns true when the number of
830 <     * registered parties is zero. Normally, overrides that arrange
831 <     * termination for other reasons should also preserve this
832 <     * property.
833 <     *
834 <     * @param phase the phase number on entering the barrier
835 <     * @param registeredParties the current number of registered
836 <     * parties.
837 <     * @return true if this barrier should terminate
821 >     * Overridable method to perform an action upon impending phase
822 >     * advance, and to control termination. This method is invoked
823 >     * upon arrival of the party advancing this phaser (when all other
824 >     * waiting parties are dormant).  If this method returns {@code
825 >     * true}, this phaser will be set to a final termination state
826 >     * upon advance, and subsequent calls to {@link #isTerminated}
827 >     * will return true. Any (unchecked) Exception or Error thrown by
828 >     * an invocation of this method is propagated to the party
829 >     * attempting to advance this phaser, in which case no advance
830 >     * occurs.
831 >     *
832 >     * <p>The arguments to this method provide the state of the phaser
833 >     * prevailing for the current transition.  The effects of invoking
834 >     * arrival, registration, and waiting methods on this phaser from
835 >     * within {@code onAdvance} are unspecified and should not be
836 >     * relied on.
837 >     *
838 >     * <p>If this phaser is a member of a tiered set of phasers, then
839 >     * {@code onAdvance} is invoked only for its root phaser on each
840 >     * advance.
841 >     *
842 >     * <p>To support the most common use cases, the default
843 >     * implementation of this method returns {@code true} when the
844 >     * number of registered parties has become zero as the result of a
845 >     * party invoking {@code arriveAndDeregister}.  You can disable
846 >     * this behavior, thus enabling continuation upon future
847 >     * registrations, by overriding this method to always return
848 >     * {@code false}:
849 >     *
850 >     * <pre> {@code
851 >     * Phaser phaser = new Phaser() {
852 >     *   protected boolean onAdvance(int phase, int parties) { return false; }
853 >     * }}</pre>
854 >     *
855 >     * @param phase the current phase number on entry to this method,
856 >     * before this phaser is advanced
857 >     * @param registeredParties the current number of registered parties
858 >     * @return {@code true} if this phaser should terminate
859       */
860      protected boolean onAdvance(int phase, int registeredParties) {
861 <        return registeredParties <= 0;
861 >        return registeredParties == 0;
862      }
863  
864      /**
865 <     * Returns a string identifying this barrier, as well as its
865 >     * Returns a string identifying this phaser, as well as its
866       * state.  The state, in brackets, includes the String {@code
867 <     * "phase ="} followed by the phase number, {@code "parties ="}
867 >     * "phase = "} followed by the phase number, {@code "parties = "}
868       * followed by the number of registered parties, and {@code
869 <     * "arrived ="} followed by the number of arrived parties
869 >     * "arrived = "} followed by the number of arrived parties.
870       *
871 <     * @return a string identifying this barrier, as well as its state
871 >     * @return a string identifying this phaser, as well as its state
872       */
873      public String toString() {
874 <        long s = state.get();
475 <        return super.toString() + "[phase = " + phaseOf(s) + " parties = " + partiesOf(s) + " arrived = " + arrivedOf(s) + "]";
874 >        return stateToString(reconcileState());
875      }
876  
478    // methods for tripping and waiting
479
877      /**
878 <     * Advance the current phase (or terminate)
878 >     * Implementation of toString and string-based error messages
879       */
880 <    private void trip(int phase, int parties) {
881 <        int next = onAdvance(phase, parties)? -1 : ((phase + 1) & phaseMask);
882 <        state.set(stateFor(next, parties, parties));
883 <        if (head.get() != null)
884 <            releaseWaiters(next);
880 >    private String stateToString(long s) {
881 >        return super.toString() +
882 >            "[phase = " + phaseOf(s) +
883 >            " parties = " + partiesOf(s) +
884 >            " arrived = " + arrivedOf(s) + "]";
885      }
886  
887 <    private int helpingWait(int phase) {
888 <        final AtomicLong state = this.state;
889 <        int p;
890 <        while ((p = phaseOf(state.get())) == phase) {
891 <            ForkJoinTask<?> t = ForkJoinWorkerThread.pollTask();
892 <            if (t != null) {
893 <                if ((p = phaseOf(state.get())) == phase)
894 <                    t.exec();
895 <                else {   // push task and exit if barrier advanced
896 <                    t.fork();
897 <                    break;
898 <                }
887 >    // Waiting mechanics
888 >
889 >    /**
890 >     * Removes and signals threads from queue for phase.
891 >     */
892 >    private void releaseWaiters(int phase) {
893 >        QNode q;   // first element of queue
894 >        int p;     // its phase
895 >        Thread t;  // its thread
896 >        //        assert phase != phaseOf(root.state);
897 >        AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
898 >        while ((q = head.get()) != null &&
899 >               q.phase != (int)(root.state >>> PHASE_SHIFT)) {
900 >            if (head.compareAndSet(q, q.next) &&
901 >                (t = q.thread) != null) {
902 >                q.thread = null;
903 >                LockSupport.unpark(t);
904              }
905          }
504        return p;
906      }
907  
908 <    private int timedHelpingWait(int phase, long nanos) throws TimeoutException {
909 <        final AtomicLong state = this.state;
910 <        long lastTime = System.nanoTime();
908 >    /** The number of CPUs, for spin control */
909 >    private static final int NCPU = Runtime.getRuntime().availableProcessors();
910 >
911 >    /**
912 >     * The number of times to spin before blocking while waiting for
913 >     * advance, per arrival while waiting. On multiprocessors, fully
914 >     * blocking and waking up a large number of threads all at once is
915 >     * usually a very slow process, so we use rechargeable spins to
916 >     * avoid it when threads regularly arrive: When a thread in
917 >     * internalAwaitAdvance notices another arrival before blocking,
918 >     * and there appear to be enough CPUs available, it spins
919 >     * SPINS_PER_ARRIVAL more times before blocking. The value trades
920 >     * off good-citizenship vs big unnecessary slowdowns.
921 >     */
922 >    static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
923 >
924 >    /**
925 >     * Possibly blocks and waits for phase to advance unless aborted.
926 >     * Call only from root node.
927 >     *
928 >     * @param phase current phase
929 >     * @param node if non-null, the wait node to track interrupt and timeout;
930 >     * if null, denotes noninterruptible wait
931 >     * @return current phase
932 >     */
933 >    private int internalAwaitAdvance(int phase, QNode node) {
934 >        releaseWaiters(phase-1);          // ensure old queue clean
935 >        boolean queued = false;           // true when node is enqueued
936 >        int lastUnarrived = 0;            // to increase spins upon change
937 >        int spins = SPINS_PER_ARRIVAL;
938 >        long s;
939          int p;
940 <        while ((p = phaseOf(state.get())) == phase) {
941 <            long now = System.nanoTime();
942 <            nanos -= now - lastTime;
943 <            lastTime = now;
944 <            if (nanos <= 0) {
945 <                if ((p = phaseOf(state.get())) == phase)
946 <                    throw new TimeoutException();
947 <                else
948 <                    break;
940 >        while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
941 >            if (node == null) {           // spinning in noninterruptible mode
942 >                int unarrived = (int)s & UNARRIVED_MASK;
943 >                if (unarrived != lastUnarrived &&
944 >                    (lastUnarrived = unarrived) < NCPU)
945 >                    spins += SPINS_PER_ARRIVAL;
946 >                boolean interrupted = Thread.interrupted();
947 >                if (interrupted || --spins < 0) { // need node to record intr
948 >                    node = new QNode(this, phase, false, false, 0L);
949 >                    node.wasInterrupted = interrupted;
950 >                }
951              }
952 <            ForkJoinTask<?> t = ForkJoinWorkerThread.pollTask();
953 <            if (t != null) {
954 <                if ((p = phaseOf(state.get())) == phase)
955 <                    t.exec();
956 <                else {   // push task and exit if barrier advanced
957 <                    t.fork();
958 <                    break;
952 >            else if (node.isReleasable()) // done or aborted
953 >                break;
954 >            else if (!queued) {           // push onto queue
955 >                AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
956 >                QNode q = node.next = head.get();
957 >                if ((q == null || q.phase == phase) &&
958 >                    (int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
959 >                    queued = head.compareAndSet(q, node);
960 >            }
961 >            else {
962 >                try {
963 >                    ForkJoinPool.managedBlock(node);
964 >                } catch (InterruptedException ie) {
965 >                    node.wasInterrupted = true;
966                  }
967              }
968          }
969 +
970 +        if (node != null) {
971 +            if (node.thread != null)
972 +                node.thread = null;       // avoid need for unpark()
973 +            if (node.wasInterrupted && !node.interruptible)
974 +                Thread.currentThread().interrupt();
975 +            if (p == phase && (p = (int)(state >>> PHASE_SHIFT)) == phase)
976 +                return p;                 // recheck abort
977 +        }
978 +        releaseWaiters(phase);
979          return p;
980      }
981  
982      /**
983 <     * Wait nodes for Treiber stack representing wait queue for non-FJ
536 <     * tasks. The waiting scheme is an adaptation of the one used in
537 <     * forkjoin.PoolBarrier.
983 >     * Wait nodes for Treiber stack representing wait queue
984       */
985 <    static final class QNode {
986 <        QNode next;
541 <        volatile Thread thread; // nulled to cancel wait
985 >    static final class QNode implements ForkJoinPool.ManagedBlocker {
986 >        final Phaser phaser;
987          final int phase;
988 <        QNode(Thread t, int c) {
989 <            thread = t;
990 <            phase = c;
988 >        final boolean interruptible;
989 >        final boolean timed;
990 >        boolean wasInterrupted;
991 >        long nanos;
992 >        long lastTime;
993 >        volatile Thread thread; // nulled to cancel wait
994 >        QNode next;
995 >
996 >        QNode(Phaser phaser, int phase, boolean interruptible,
997 >              boolean timed, long nanos) {
998 >            this.phaser = phaser;
999 >            this.phase = phase;
1000 >            this.interruptible = interruptible;
1001 >            this.nanos = nanos;
1002 >            this.timed = timed;
1003 >            this.lastTime = timed ? System.nanoTime() : 0L;
1004 >            thread = Thread.currentThread();
1005          }
547    }
1006  
1007 <    private void releaseWaiters(int currentPhase) {
1008 <        final AtomicReference<QNode> head = this.head;
1009 <        QNode p;
1010 <        while ((p = head.get()) != null && p.phase != currentPhase) {
1011 <            if (head.compareAndSet(p, null)) {
1012 <                do {
555 <                    Thread t = p.thread;
556 <                    if (t != null) {
557 <                        p.thread = null;
558 <                        LockSupport.unpark(t);
559 <                    }
560 <                } while ((p = p.next) != null);
1007 >        public boolean isReleasable() {
1008 >            if (thread == null)
1009 >                return true;
1010 >            if (phaser.getPhase() != phase) {
1011 >                thread = null;
1012 >                return true;
1013              }
1014 +            if (Thread.interrupted())
1015 +                wasInterrupted = true;
1016 +            if (wasInterrupted && interruptible) {
1017 +                thread = null;
1018 +                return true;
1019 +            }
1020 +            if (timed) {
1021 +                if (nanos > 0L) {
1022 +                    long now = System.nanoTime();
1023 +                    nanos -= now - lastTime;
1024 +                    lastTime = now;
1025 +                }
1026 +                if (nanos <= 0L) {
1027 +                    thread = null;
1028 +                    return true;
1029 +                }
1030 +            }
1031 +            return false;
1032          }
563    }
1033  
1034 <    /** The number of CPUs, for spin control */
1035 <    static final int NCPUS = Runtime.getRuntime().availableProcessors();
1036 <
1037 <    /**
1038 <     * The number of times to spin before blocking in timed waits.
1039 <     * The value is empirically derived.
1040 <     */
1041 <    static final int maxTimedSpins = (NCPUS < 2)? 0 : 32;
1042 <
1043 <    /**
575 <     * The number of times to spin before blocking in untimed waits.
576 <     * This is greater than timed value because untimed waits spin
577 <     * faster since they don't need to check times on each spin.
578 <     */
579 <    static final int maxUntimedSpins = maxTimedSpins * 32;
1034 >        public boolean block() {
1035 >            if (isReleasable())
1036 >                return true;
1037 >            else if (!timed)
1038 >                LockSupport.park(this);
1039 >            else if (nanos > 0)
1040 >                LockSupport.parkNanos(this, nanos);
1041 >            return isReleasable();
1042 >        }
1043 >    }
1044  
1045 <    /**
582 <     * The number of nanoseconds for which it is faster to spin
583 <     * rather than to use timed park. A rough estimate suffices.
584 <     */
585 <    static final long spinForTimeoutThreshold = 1000L;
1045 >    // Unsafe mechanics
1046  
1047 <    /**
1048 <     * Enqueues node and waits unless aborted or signalled.
1049 <     */
1050 <    private boolean untimedWait(Thread thread, int currentPhase,
1051 <                               boolean abortOnInterrupt) {
1052 <        final AtomicReference<QNode> head = this.head;
1053 <        final AtomicLong state = this.state;
1054 <        boolean wasInterrupted = false;
1055 <        QNode node = null;
1056 <        boolean queued = false;
1057 <        int spins = maxUntimedSpins;
1058 <        while (phaseOf(state.get()) == currentPhase) {
599 <            QNode h;
600 <            if (node != null && queued) {
601 <                if (node.thread != null) {
602 <                    LockSupport.park();
603 <                    if (Thread.interrupted()) {
604 <                        wasInterrupted = true;
605 <                        if (abortOnInterrupt)
606 <                            break;
607 <                    }
608 <                }
609 <            }
610 <            else if ((h = head.get()) != null && h.phase != currentPhase) {
611 <                if (phaseOf(state.get()) == currentPhase) { // must recheck
612 <                    if (head.compareAndSet(h, h.next)) {
613 <                        Thread t = h.thread; // help clear out old waiters
614 <                        if (t != null) {
615 <                            h.thread = null;
616 <                            LockSupport.unpark(t);
617 <                        }
618 <                    }
619 <                }
620 <                else
621 <                    break;
622 <            }
623 <            else if (node != null)
624 <                queued = head.compareAndSet(node.next = h, node);
625 <            else if (spins <= 0)
626 <                node = new QNode(thread, currentPhase);
627 <            else
628 <                --spins;
1047 >    private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1048 >    private static final long stateOffset =
1049 >        objectFieldOffset("state", Phaser.class);
1050 >
1051 >    private static long objectFieldOffset(String field, Class<?> klazz) {
1052 >        try {
1053 >            return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1054 >        } catch (NoSuchFieldException e) {
1055 >            // Convert Exception to corresponding Error
1056 >            NoSuchFieldError error = new NoSuchFieldError(field);
1057 >            error.initCause(e);
1058 >            throw error;
1059          }
630        if (node != null)
631            node.thread = null;
632        return wasInterrupted;
1060      }
1061  
1062      /**
1063 <     * Messier timeout version
1063 >     * Returns a sun.misc.Unsafe.  Suitable for use in a 3rd party package.
1064 >     * Replace with a simple call to Unsafe.getUnsafe when integrating
1065 >     * into a jdk.
1066 >     *
1067 >     * @return a sun.misc.Unsafe
1068       */
1069 <    private void timedWait(Thread thread, int currentPhase, long nanos)
1070 <        throws InterruptedException, TimeoutException {
1071 <        final AtomicReference<QNode> head = this.head;
1072 <        final AtomicLong state = this.state;
1073 <        long lastTime = System.nanoTime();
1074 <        QNode node = null;
1075 <        boolean queued = false;
1076 <        int spins = maxTimedSpins;
1077 <        while (phaseOf(state.get()) == currentPhase) {
1078 <            QNode h;
1079 <            long now = System.nanoTime();
1080 <            nanos -= now - lastTime;
1081 <            lastTime = now;
1082 <            if (nanos <= 0) {
1083 <                if (node != null)
1084 <                    node.thread = null;
1085 <                if (phaseOf(state.get()) == currentPhase)
655 <                    throw new TimeoutException();
656 <                else
657 <                    break;
658 <            }
659 <            else if (node != null && queued) {
660 <                if (node.thread != null &&
661 <                    nanos > spinForTimeoutThreshold) {
662 <                    //                LockSupport.parkNanos(this, nanos);
663 <                    LockSupport.parkNanos(nanos);
664 <                    if (Thread.interrupted()) {
665 <                        node.thread = null;
666 <                        throw new InterruptedException();
667 <                    }
668 <                }
1069 >    private static sun.misc.Unsafe getUnsafe() {
1070 >        try {
1071 >            return sun.misc.Unsafe.getUnsafe();
1072 >        } catch (SecurityException se) {
1073 >            try {
1074 >                return java.security.AccessController.doPrivileged
1075 >                    (new java.security
1076 >                     .PrivilegedExceptionAction<sun.misc.Unsafe>() {
1077 >                        public sun.misc.Unsafe run() throws Exception {
1078 >                            java.lang.reflect.Field f = sun.misc
1079 >                                .Unsafe.class.getDeclaredField("theUnsafe");
1080 >                            f.setAccessible(true);
1081 >                            return (sun.misc.Unsafe) f.get(null);
1082 >                        }});
1083 >            } catch (java.security.PrivilegedActionException e) {
1084 >                throw new RuntimeException("Could not initialize intrinsics",
1085 >                                           e.getCause());
1086              }
670            else if ((h = head.get()) != null && h.phase != currentPhase) {
671                if (phaseOf(state.get()) == currentPhase) { // must recheck
672                    if (head.compareAndSet(h, h.next)) {
673                        Thread t = h.thread; // help clear out old waiters
674                        if (t != null) {
675                            h.thread = null;
676                            LockSupport.unpark(t);
677                        }
678                    }
679                }
680                else
681                    break;
682            }
683            else if (node != null)
684                queued = head.compareAndSet(node.next = h, node);
685            else if (spins <= 0)
686                node = new QNode(thread, currentPhase);
687            else
688                --spins;
1087          }
690        if (node != null)
691            node.thread = null;
1088      }
693
1089   }

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines