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 1.64 by jsr166, Mon Nov 29 20:58:06 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 tasks using a tree of phasers,
187 > * you could use code of the following form, assuming a
188 > * Task class with a constructor accepting a {@code Phaser} that
189 > * it registers with upon construction:
190 > *
191 > *  <pre> {@code
192 > * void build(Task[] actions, int lo, int hi, Phaser ph) {
193 > *   if (hi - lo > TASKS_PER_PHASER) {
194 > *     for (int i = lo; i < hi; i += TASKS_PER_PHASER) {
195 > *       int j = Math.min(i + TASKS_PER_PHASER, hi);
196 > *       build(actions, i, j, new Phaser(ph));
197   *     }
198 < *     parentBarrier.arriveAndDeregister();
198 > *   } else {
199 > *     for (int i = lo; i < hi; ++i)
200 > *       actions[i] = new Task(ph);
201 > *       // assumes new Task(ph) performs ph.register()
202   *   }
203   * }
204 < * </pre>
204 > * // .. initially called, for n tasks via
205 > * build(new Task[n], 0, n, new Phaser());}</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;
237 >     * packed into a single (atomic) long. Termination uses the sign
238 >     * bit of 32 bit representation of phase, so phase is set to -1 on
239 >     * termination. Good performance relies on keeping state decoding
240 >     * and encoding simple, and keeping race windows short.
241 >     */
242 >    private volatile long state;
243 >
244 >    private static final int  MAX_PARTIES     = 0xffff;
245 >    private static final int  MAX_PHASE       = 0x7fffffff;
246 >    private static final int  PARTIES_SHIFT   = 16;
247 >    private static final int  PHASE_SHIFT     = 32;
248 >    private static final int  UNARRIVED_MASK  = 0xffff;      // to mask ints
249 >    private static final long PARTIES_MASK    = 0xffff0000L; // to mask longs
250 >    private static final long ONE_ARRIVAL     = 1L;
251 >    private static final long ONE_PARTY       = 1L << PARTIES_SHIFT;
252 >    private static final long TERMINATION_BIT = 1L << 63;
253  
254 <    /**
143 <     * Head of Treiber stack for waiting nonFJ threads.
144 <     */
145 <    private final AtomicReference<QNode> head = new AtomicReference<QNode>();
146 <
147 <    private static final int ushortBits = 16;
148 <    private static final int ushortMask =  (1 << ushortBits) - 1;
149 <    private static final int phaseMask = 0x7fffffff;
254 >    // The following unpacking methods are usually manually inlined
255  
256      private static int unarrivedOf(long s) {
257 <        return (int)(s & ushortMask);
257 >        return (int)s & UNARRIVED_MASK;
258      }
259  
260      private static int partiesOf(long s) {
261 <        return (int)(s & (ushortMask << 16)) >>> 16;
261 >        return (int)s >>> PARTIES_SHIFT;
262      }
263  
264      private static int phaseOf(long s) {
265 <        return (int)(s >>> 32);
265 >        return (int) (s >>> PHASE_SHIFT);
266      }
267  
268      private static int arrivedOf(long s) {
269          return partiesOf(s) - unarrivedOf(s);
270      }
271  
272 <    private static long stateFor(int phase, int parties, int unarrived) {
273 <        return (((long)phase) << 32) | ((parties << 16) | unarrived);
274 <    }
272 >    /**
273 >     * The parent of this phaser, or null if none
274 >     */
275 >    private final Phaser parent;
276  
277 <    private static IllegalStateException badBounds(int parties, int unarrived) {
278 <        return new IllegalStateException("Attempt to set " + unarrived +
279 <                                         " unarrived of " + parties + " parties");
280 <    }
277 >    /**
278 >     * The root of phaser tree. Equals this if not in a tree.  Used to
279 >     * support faster state push-down.
280 >     */
281 >    private final Phaser root;
282  
283      /**
284 <     * Creates a new Phaser without any initially registered parties,
285 <     * and initial phase number 0.
284 >     * Heads of Treiber stacks for waiting threads. To eliminate
285 >     * contention when releasing some threads while adding others, we
286 >     * use two of them, alternating across even and odd phases.
287 >     * Subphasers share queues with root to speed up releases.
288       */
289 <    public Phaser() {
290 <        state = new AtomicLong(stateFor(0, 0, 0));
289 >    private final AtomicReference<QNode> evenQ;
290 >    private final AtomicReference<QNode> oddQ;
291 >
292 >    private AtomicReference<QNode> queueFor(int phase) {
293 >        return ((phase & 1) == 0) ? evenQ : oddQ;
294      }
295  
296      /**
297 <     * Creates a new Phaser with the given numbers of registered
186 <     * unarrived parties and initial phase number 0.
187 <     * @param parties the number of parties required to trip barrier.
188 <     * @throws IllegalArgumentException if parties less than zero
189 <     * or greater than the maximum number of parties supported.
297 >     * Returns message string for bounds exceptions on arrival.
298       */
299 <    public Phaser(int parties) {
300 <        if (parties < 0 || parties > ushortMask)
301 <            throw new IllegalArgumentException("Illegal number of parties");
194 <        state = new AtomicLong(stateFor(0, parties, parties));
299 >    private String badArrive(long s) {
300 >        return "Attempted arrival of unregistered party for " +
301 >            stateToString(s);
302      }
303  
304      /**
305 <     * Adds a new unarrived party to this phaser.
199 <     * @return the current barrier phase number upon registration
200 <     * @throws IllegalStateException if attempting to register more
201 <     * than the maximum supported number of parties.
305 >     * Returns message string for bounds exceptions on registration.
306       */
307 <    public int register() { // increment both parties and unarrived
308 <        final AtomicLong state = this.state;
309 <        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 <        }
307 >    private String badRegister(long s) {
308 >        return "Attempt to register more than " +
309 >            MAX_PARTIES + " parties for " + stateToString(s);
310      }
311  
312      /**
313 <     * Arrives at the barrier, but does not wait for others.  (You can
314 <     * in turn wait for others via {@link #awaitAdvance}).
313 >     * Main implementation for methods arrive and arriveAndDeregister.
314 >     * Manually tuned to speed up and minimize race windows for the
315 >     * common case of just decrementing unarrived field.
316       *
317 <     * @return the current barrier phase number upon entry to
318 <     * this method, or a negative value if terminated;
319 <     * @throws IllegalStateException if the number of unarrived
224 <     * parties would become negative.
317 >     * @param adj - adjustment to apply to state -- either
318 >     * ONE_ARRIVAL (for arrive) or
319 >     * ONE_ARRIVAL|ONE_PARTY (for arriveAndDeregister)
320       */
321 <    public int arrive() { // decrement unarrived. If zero, trip
227 <        final AtomicLong state = this.state;
321 >    private int doArrive(long adj) {
322          for (;;) {
323 <            long s = state.get();
324 <            int phase = phaseOf(s);
325 <            int parties = partiesOf(s);
326 <            int unarrived = unarrivedOf(s) - 1;
233 <            if (unarrived < 0)
234 <                throw badBounds(parties, unarrived);
235 <            if (unarrived == 0 && phase >= 0) {
236 <                trip(phase, parties);
323 >            long s = state;
324 >            int unarrived = (int)s & UNARRIVED_MASK;
325 >            int phase = (int)(s >>> PHASE_SHIFT);
326 >            if (phase < 0)
327                  return phase;
328 +            else if (unarrived == 0) {
329 +                if (reconcileState() == s)     // recheck
330 +                    throw new IllegalStateException(badArrive(s));
331              }
332 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived)))
332 >            else if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s-=adj)) {
333 >                if (unarrived == 1) {
334 >                    long p = s & PARTIES_MASK; // unshifted parties field
335 >                    long lu = p >>> PARTIES_SHIFT;
336 >                    int u = (int)lu;
337 >                    int nextPhase = (phase + 1) & MAX_PHASE;
338 >                    long next = ((long)nextPhase << PHASE_SHIFT) | p | lu;
339 >                    final Phaser parent = this.parent;
340 >                    if (parent == null) {
341 >                        if (onAdvance(phase, u))
342 >                            next |= TERMINATION_BIT;
343 >                        UNSAFE.compareAndSwapLong(this, stateOffset, s, next);
344 >                        releaseWaiters(phase);
345 >                    }
346 >                    else {
347 >                        parent.doArrive((u == 0) ?
348 >                                        ONE_ARRIVAL|ONE_PARTY : ONE_ARRIVAL);
349 >                        if ((int)(parent.state >>> PHASE_SHIFT) != nextPhase)
350 >                            reconcileState();
351 >                        else if (state == s)
352 >                            UNSAFE.compareAndSwapLong(this, stateOffset, s,
353 >                                                      next);
354 >                    }
355 >                }
356                  return phase;
357 +            }
358          }
359      }
360  
361      /**
362 <     * Arrives at the barrier, and deregisters from it, without
246 <     * waiting for others.
362 >     * Implementation of register, bulkRegister
363       *
364 <     * @return the current barrier phase number upon entry to
365 <     * this method, or a negative value if terminated;
250 <     * @throws IllegalStateException if the number of registered or
251 <     * unarrived parties would become negative.
364 >     * @param registrations number to add to both parties and
365 >     * unarrived fields. Must be greater than zero.
366       */
367 <    public int arriveAndDeregister() { // Same as arrive, plus decrement parties
368 <        final AtomicLong state = this.state;
367 >    private int doRegister(int registrations) {
368 >        // adjustment to state
369 >        long adj = ((long)registrations << PARTIES_SHIFT) | registrations;
370 >        final Phaser parent = this.parent;
371          for (;;) {
372 <            long s = state.get();
373 <            int phase = phaseOf(s);
374 <            int parties = partiesOf(s) - 1;
375 <            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);
372 >            long s = (parent == null) ? state : reconcileState();
373 >            int parties = (int)s >>> PARTIES_SHIFT;
374 >            int phase = (int)(s >>> PHASE_SHIFT);
375 >            if (phase < 0)
376                  return phase;
377 +            else if (registrations > MAX_PARTIES - parties)
378 +                throw new IllegalStateException(badRegister(s));
379 +            else if ((parties == 0 && parent == null) || // first reg of root
380 +                     ((int)s & UNARRIVED_MASK) != 0) {   // not advancing
381 +                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s + adj))
382 +                    return phase;
383 +            }
384 +            else if (parties != 0)               // wait for onAdvance
385 +                root.internalAwaitAdvance(phase, null);
386 +            else {                               // 1st registration of child
387 +                synchronized (this) {            // register parent first
388 +                    if (reconcileState() == s) { // recheck under lock
389 +                        parent.doRegister(1);    // OK if throws IllegalState
390 +                        for (;;) {               // simpler form of outer loop
391 +                            s = reconcileState();
392 +                            phase = (int)(s >>> PHASE_SHIFT);
393 +                            if (phase < 0 ||
394 +                                UNSAFE.compareAndSwapLong(this, stateOffset,
395 +                                                          s, s + adj))
396 +                                return phase;
397 +                        }
398 +                    }
399 +                }
400              }
266            if (state.compareAndSet(s, stateFor(phase, parties, unarrived)))
267                return phase;
401          }
402      }
403  
404      /**
405 <     * Arrives at the barrier and awaits others. Unlike other arrival
273 <     * methods, this method returns the arrival index of the
274 <     * caller. The caller tripping the barrier returns zero, the
275 <     * previous caller 1, and so on.
276 <     * @return the arrival index
277 <     * @throws IllegalStateException if the number of unarrived
278 <     * parties would become negative.
405 >     * Recursively resolves lagged phase propagation from root if necessary.
406       */
407 <    public int arriveAndAwaitAdvance() {
408 <        final AtomicLong state = this.state;
409 <        for (;;) {
410 <            long s = state.get();
411 <            int phase = phaseOf(s);
412 <            int parties = partiesOf(s);
413 <            int unarrived = unarrivedOf(s) - 1;
414 <            if (unarrived < 0)
415 <                throw badBounds(parties, unarrived);
416 <            if (unarrived == 0 && phase >= 0) {
417 <                trip(phase, parties);
418 <                return 0;
419 <            }
420 <            if (state.compareAndSet(s, stateFor(phase, parties, unarrived))) {
421 <                awaitAdvance(phase);
422 <                return unarrived;
407 >    private long reconcileState() {
408 >        Phaser par = parent;
409 >        long s = state;
410 >        if (par != null) {
411 >            Phaser rt = root;
412 >            int phase, rPhase;
413 >            while ((phase = (int)(s >>> PHASE_SHIFT)) >= 0 &&
414 >                   (rPhase = (int)(rt.state >>> PHASE_SHIFT)) != phase) {
415 >                if (par != rt && (int)(par.state >>> PHASE_SHIFT) != rPhase)
416 >                    par.reconcileState();
417 >                else if (rPhase < 0 || ((int)s & UNARRIVED_MASK) == 0) {
418 >                    long u = s & PARTIES_MASK; // reset unarrived to parties
419 >                    long next = ((((long) rPhase) << PHASE_SHIFT) | u |
420 >                                 (u >>> PARTIES_SHIFT));
421 >                    UNSAFE.compareAndSwapLong(this, stateOffset, s, next);
422 >                }
423 >                s = state;
424              }
425          }
426 +        return s;
427      }
428  
429      /**
430 <     * Awaits the phase of the barrier to advance from the given
431 <     * value, or returns immediately if this barrier is terminated.
432 <     * @param phase the phase on entry to this method
433 <     * @return the phase on exit from this method
430 >     * Creates a new phaser with no initially registered parties, no
431 >     * parent, and initial phase number 0. Any thread using this
432 >     * phaser will need to first register for it.
433 >     */
434 >    public Phaser() {
435 >        this(null, 0);
436 >    }
437 >
438 >    /**
439 >     * Creates a new phaser with the given number of registered
440 >     * unarrived parties, no parent, and initial phase number 0.
441 >     *
442 >     * @param parties the number of parties required to advance to the
443 >     * next phase
444 >     * @throws IllegalArgumentException if parties less than zero
445 >     * or greater than the maximum number of parties supported
446 >     */
447 >    public Phaser(int parties) {
448 >        this(null, parties);
449 >    }
450 >
451 >    /**
452 >     * Equivalent to {@link #Phaser(Phaser, int) Phaser(parent, 0)}.
453 >     *
454 >     * @param parent the parent phaser
455 >     */
456 >    public Phaser(Phaser parent) {
457 >        this(parent, 0);
458 >    }
459 >
460 >    /**
461 >     * Creates a new phaser with the given parent and number of
462 >     * registered unarrived parties. Registration and deregistration
463 >     * of this child phaser with its parent are managed automatically.
464 >     * If the given parent is non-null, whenever this child phaser has
465 >     * any registered parties (as established in this constructor,
466 >     * {@link #register}, or {@link #bulkRegister}), this child phaser
467 >     * is registered with its parent. Whenever the number of
468 >     * registered parties becomes zero as the result of an invocation
469 >     * of {@link #arriveAndDeregister}, this child phaser is
470 >     * deregistered from its parent.
471 >     *
472 >     * @param parent the parent phaser
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
477 >     */
478 >    public Phaser(Phaser parent, int parties) {
479 >        if (parties >>> PARTIES_SHIFT != 0)
480 >            throw new IllegalArgumentException("Illegal number of parties");
481 >        long s = ((long) parties) | (((long) parties) << PARTIES_SHIFT);
482 >        this.parent = parent;
483 >        if (parent != null) {
484 >            Phaser r = parent.root;
485 >            this.root = r;
486 >            this.evenQ = r.evenQ;
487 >            this.oddQ = r.oddQ;
488 >            if (parties != 0)
489 >                s |= ((long)(parent.doRegister(1))) << PHASE_SHIFT;
490 >        }
491 >        else {
492 >            this.root = this;
493 >            this.evenQ = new AtomicReference<QNode>();
494 >            this.oddQ = new AtomicReference<QNode>();
495 >        }
496 >        this.state = s;
497 >    }
498 >
499 >    /**
500 >     * Adds a new unarrived party to this phaser.  If an ongoing
501 >     * invocation of {@link #onAdvance} is in progress, this method
502 >     * may await its completion before returning.  If this phaser has
503 >     * a parent, and this phaser previously had no registered parties,
504 >     * this phaser is also registered with its parent.
505 >     *
506 >     * @return the arrival phase number to which this registration applied
507 >     * @throws IllegalStateException if attempting to register more
508 >     * than the maximum supported number of parties
509 >     */
510 >    public int register() {
511 >        return doRegister(1);
512 >    }
513 >
514 >    /**
515 >     * Adds the given number of new unarrived parties to this phaser.
516 >     * If an ongoing invocation of {@link #onAdvance} is in progress,
517 >     * this method may await its completion before returning.  If this
518 >     * phaser has a parent, and the given number of parities is
519 >     * greater than zero, and this phaser previously had no registered
520 >     * parties, this phaser is also registered with its parent.
521 >     *
522 >     * @param parties the number of additional parties required to
523 >     * advance to the next phase
524 >     * @return the arrival phase number to which this registration applied
525 >     * @throws IllegalStateException if attempting to register more
526 >     * than the maximum supported number of parties
527 >     * @throws IllegalArgumentException if {@code parties < 0}
528 >     */
529 >    public int bulkRegister(int parties) {
530 >        if (parties < 0)
531 >            throw new IllegalArgumentException();
532 >        if (parties == 0)
533 >            return getPhase();
534 >        return doRegister(parties);
535 >    }
536 >
537 >    /**
538 >     * Arrives at this phaser, without waiting for others to arrive.
539 >     *
540 >     * <p>It is a usage error for an unregistered party to invoke this
541 >     * method.  However, this error may result in an {@code
542 >     * IllegalStateException} only upon some subsequent operation on
543 >     * this phaser, if ever.
544 >     *
545 >     * @return the arrival phase number, or a negative value if terminated
546 >     * @throws IllegalStateException if not terminated and the number
547 >     * of unarrived parties would become negative
548 >     */
549 >    public int arrive() {
550 >        return doArrive(ONE_ARRIVAL);
551 >    }
552 >
553 >    /**
554 >     * Arrives at this phaser and deregisters from it without waiting
555 >     * for others to arrive. Deregistration reduces the number of
556 >     * parties required to advance in future phases.  If this phaser
557 >     * has a parent, and deregistration causes this phaser to have
558 >     * zero parties, this phaser is also deregistered from its parent.
559 >     *
560 >     * <p>It is a usage error for an unregistered party to invoke this
561 >     * method.  However, this error may result in an {@code
562 >     * IllegalStateException} only upon some subsequent operation on
563 >     * this phaser, if ever.
564 >     *
565 >     * @return the arrival phase number, or a negative value if terminated
566 >     * @throws IllegalStateException if not terminated and the number
567 >     * of registered or unarrived parties would become negative
568 >     */
569 >    public int arriveAndDeregister() {
570 >        return doArrive(ONE_ARRIVAL|ONE_PARTY);
571 >    }
572 >
573 >    /**
574 >     * Arrives at this phaser and awaits others. Equivalent in effect
575 >     * to {@code awaitAdvance(arrive())}.  If you need to await with
576 >     * interruption or timeout, you can arrange this with an analogous
577 >     * construction using one of the other forms of the {@code
578 >     * awaitAdvance} method.  If instead you need to deregister upon
579 >     * arrival, use {@code awaitAdvance(arriveAndDeregister())}.
580 >     *
581 >     * <p>It is a usage error for an unregistered party to invoke this
582 >     * method.  However, this error may result in an {@code
583 >     * IllegalStateException} only upon some subsequent operation on
584 >     * this phaser, if ever.
585 >     *
586 >     * @return the arrival phase number, or a negative number if terminated
587 >     * @throws IllegalStateException if not terminated and the number
588 >     * of unarrived parties would become negative
589 >     */
590 >    public int arriveAndAwaitAdvance() {
591 >        return awaitAdvance(doArrive(ONE_ARRIVAL));
592 >    }
593 >
594 >    /**
595 >     * Awaits the phase of this phaser to advance from the given phase
596 >     * value, returning immediately if the current phase is not equal
597 >     * to the given phase value or this phaser is terminated.
598 >     *
599 >     * @param phase an arrival phase number, or negative value if
600 >     * terminated; this argument is normally the value returned by a
601 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
602 >     * @return the next arrival phase number, or a negative value
603 >     * if terminated or argument is negative
604       */
605      public int awaitAdvance(int phase) {
606 +        Phaser rt;
607 +        int p = (int)(state >>> PHASE_SHIFT);
608          if (phase < 0)
609              return phase;
610 <        Thread current = Thread.currentThread();
611 <        if (current instanceof ForkJoinWorkerThread)
612 <            return helpingWait(phase);
613 <        if (untimedWait(current, phase, false))
313 <            current.interrupt();
314 <        return phaseOf(state.get());
610 >        if (p == phase &&
611 >            (p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase)
612 >            return rt.internalAwaitAdvance(phase, null);
613 >        return p;
614      }
615  
616      /**
617 <     * Awaits the phase of the barrier to advance from the given
618 <     * value, or returns immediately if this barrier is terminated, or
619 <     * throws InterruptedException if interrupted while waiting.
620 <     * @param phase the phase on entry to this method
621 <     * @return the phase on exit from this method
617 >     * Awaits the phase of this phaser to advance from the given phase
618 >     * value, throwing {@code InterruptedException} if interrupted
619 >     * while waiting, or returning immediately if the current phase is
620 >     * not equal to the given phase value or this phaser is
621 >     * terminated.
622 >     *
623 >     * @param phase an arrival phase number, or negative value if
624 >     * terminated; this argument is normally the value returned by a
625 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
626 >     * @return the next arrival phase number, or a negative value
627 >     * if terminated or argument is negative
628       * @throws InterruptedException if thread interrupted while waiting
629       */
630 <    public int awaitAdvanceInterruptibly(int phase) throws InterruptedException {
630 >    public int awaitAdvanceInterruptibly(int phase)
631 >        throws InterruptedException {
632 >        Phaser rt;
633 >        int p = (int)(state >>> PHASE_SHIFT);
634          if (phase < 0)
635              return phase;
636 <        Thread current = Thread.currentThread();
637 <        if (current instanceof ForkJoinWorkerThread)
638 <            return helpingWait(phase);
639 <        else if (Thread.interrupted() || untimedWait(current, phase, true))
640 <            throw new InterruptedException();
641 <        else
642 <            return phaseOf(state.get());
636 >        if (p == phase &&
637 >            (p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase) {
638 >            QNode node = new QNode(this, phase, true, false, 0L);
639 >            p = rt.internalAwaitAdvance(phase, node);
640 >            if (node.wasInterrupted)
641 >                throw new InterruptedException();
642 >        }
643 >        return p;
644      }
645  
646      /**
647 <     * Awaits the phase of the barrier to advance from the given value
648 <     * or the given timeout elapses, or returns immediately if this
649 <     * barrier is terminated.
650 <     * @param phase the phase on entry to this method
651 <     * @return the phase on exit from this method
647 >     * Awaits the phase of this phaser to advance from the given phase
648 >     * value or the given timeout to elapse, throwing {@code
649 >     * InterruptedException} if interrupted while waiting, or
650 >     * returning immediately if the current phase is not equal to the
651 >     * given phase value or this phaser is terminated.
652 >     *
653 >     * @param phase an arrival phase number, or negative value if
654 >     * terminated; this argument is normally the value returned by a
655 >     * previous call to {@code arrive} or {@code arriveAndDeregister}.
656 >     * @param timeout how long to wait before giving up, in units of
657 >     *        {@code unit}
658 >     * @param unit a {@code TimeUnit} determining how to interpret the
659 >     *        {@code timeout} parameter
660 >     * @return the next arrival phase number, or a negative value
661 >     * if terminated or argument is negative
662       * @throws InterruptedException if thread interrupted while waiting
663       * @throws TimeoutException if timed out while waiting
664       */
665 <    public int awaitAdvanceInterruptibly(int phase, long timeout, TimeUnit unit)
665 >    public int awaitAdvanceInterruptibly(int phase,
666 >                                         long timeout, TimeUnit unit)
667          throws InterruptedException, TimeoutException {
668 +        long nanos = unit.toNanos(timeout);
669 +        Phaser rt;
670 +        int p = (int)(state >>> PHASE_SHIFT);
671          if (phase < 0)
672              return phase;
673 <        long nanos = unit.toNanos(timeout);
674 <        Thread current = Thread.currentThread();
675 <        if (current instanceof ForkJoinWorkerThread)
676 <            return timedHelpingWait(phase, nanos);
677 <        timedWait(current, phase, nanos);
678 <        return phaseOf(state.get());
673 >        if (p == phase &&
674 >            (p = (int)((rt = root).state >>> PHASE_SHIFT)) == phase) {
675 >            QNode node = new QNode(this, phase, true, true, nanos);
676 >            p = rt.internalAwaitAdvance(phase, node);
677 >            if (node.wasInterrupted)
678 >                throw new InterruptedException();
679 >            else if (p == phase)
680 >                throw new TimeoutException();
681 >        }
682 >        return p;
683      }
684  
685      /**
686 <     * Forces this barrier to enter termination state. Counts of
687 <     * arrived and registered parties are unaffected. This method may
688 <     * be useful for coordinating recovery after one or more tasks
686 >     * Forces this phaser to enter termination state.  Counts of
687 >     * arrived and registered parties are unaffected.  If this phaser
688 >     * is a member of a tiered set of phasers, then all of the phasers
689 >     * in the set are terminated.  If this phaser is already
690 >     * terminated, this method has no effect.  This method may be
691 >     * useful for coordinating recovery after one or more tasks
692       * encounter unexpected exceptions.
693       */
694      public void forceTermination() {
695 <        final AtomicLong state = this.state;
696 <        for (;;) {
697 <            long s = state.get();
698 <            int phase = phaseOf(s);
699 <            int parties = partiesOf(s);
700 <            int unarrived = unarrivedOf(s);
701 <            if (phase < 0 ||
702 <                state.compareAndSet(s, stateFor(-1, parties, unarrived))) {
373 <                if (head.get() != null)
374 <                    releaseWaiters(-1);
695 >        // Only need to change root state
696 >        final Phaser root = this.root;
697 >        long s;
698 >        while ((s = root.state) >= 0) {
699 >            if (UNSAFE.compareAndSwapLong(root, stateOffset,
700 >                                          s, s | TERMINATION_BIT)) {
701 >                releaseWaiters(0); // signal all threads
702 >                releaseWaiters(1);
703                  return;
704              }
705          }
706      }
707  
708      /**
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    /**
709       * Returns the current phase number. The maximum phase number is
710 <     * <tt>Integer.MAX_VALUE</tt>, after which it restarts at
711 <     * zero. Upon termination, the phase number is negative.
710 >     * {@code Integer.MAX_VALUE}, after which it restarts at
711 >     * zero. Upon termination, the phase number is negative,
712 >     * in which case the prevailing phase prior to termination
713 >     * may be obtained via {@code getPhase() + Integer.MIN_VALUE}.
714 >     *
715       * @return the phase number, or a negative value if terminated
716       */
717 <    public int getPhase() {
718 <        return phaseOf(state.get());
717 >    public final int getPhase() {
718 >        return (int)(root.state >>> PHASE_SHIFT);
719      }
720  
721      /**
722 <     * Returns the number of parties registered at this barrier.
722 >     * Returns the number of parties registered at this phaser.
723 >     *
724       * @return the number of parties
725       */
726      public int getRegisteredParties() {
727 <        return partiesOf(state.get());
727 >        return partiesOf(state);
728      }
729  
730      /**
731 <     * Returns the number of parties that have arrived at the current
732 <     * phase of this barrier.
731 >     * Returns the number of registered parties that have arrived at
732 >     * the current phase of this phaser.
733 >     *
734       * @return the number of arrived parties
735       */
736      public int getArrivedParties() {
737 <        return arrivedOf(state.get());
737 >        long s = state;
738 >        int u = unarrivedOf(s); // only reconcile if possibly needed
739 >        return (u != 0 || parent == null) ?
740 >            partiesOf(s) - u :
741 >            arrivedOf(reconcileState());
742      }
743  
744      /**
745       * Returns the number of registered parties that have not yet
746 <     * arrived at the current phase of this barrier.
746 >     * arrived at the current phase of this phaser.
747 >     *
748       * @return the number of unarrived parties
749       */
750      public int getUnarrivedParties() {
751 <        return unarrivedOf(state.get());
751 >        int u = unarrivedOf(state);
752 >        return (u != 0 || parent == null) ? u : unarrivedOf(reconcileState());
753      }
754  
755      /**
756 <     * Returns true if this barrier has been terminated.
757 <     * @return true if this barrier has been terminated
756 >     * Returns the parent of this phaser, or {@code null} if none.
757 >     *
758 >     * @return the parent of this phaser, or {@code null} if none
759 >     */
760 >    public Phaser getParent() {
761 >        return parent;
762 >    }
763 >
764 >    /**
765 >     * Returns the root ancestor of this phaser, which is the same as
766 >     * this phaser if it has no parent.
767 >     *
768 >     * @return the root ancestor of this phaser
769 >     */
770 >    public Phaser getRoot() {
771 >        return root;
772 >    }
773 >
774 >    /**
775 >     * Returns {@code true} if this phaser has been terminated.
776 >     *
777 >     * @return {@code true} if this phaser has been terminated
778       */
779      public boolean isTerminated() {
780 <        return phaseOf(state.get()) < 0;
780 >        return root.state < 0L;
781      }
782  
783      /**
784 <     * Overridable method to perform an action upon phase advance, and
785 <     * to control termination. This method is invoked whenever the
786 <     * barrier is tripped (and thus all other waiting parties are
787 <     * dormant). If it returns true, then, rather than advance the
788 <     * phase number, this barrier will be set to a final termination
789 <     * state, and subsequent calls to <tt>isTerminated</tt> will
790 <     * return true.
791 <     *
792 <     * <p> The default version returns true when the number of
793 <     * registered parties is zero. Normally, overrides that arrange
794 <     * termination for other reasons should also preserve this
795 <     * property.
796 <     *
797 <     * @param phase the phase number on entering the barrier
798 <     * @param registeredParties the current number of registered
799 <     * parties.
800 <     * @return true if this barrier should terminate
784 >     * Overridable method to perform an action upon impending phase
785 >     * advance, and to control termination. This method is invoked
786 >     * upon arrival of the party advancing this phaser (when all other
787 >     * waiting parties are dormant).  If this method returns {@code
788 >     * true}, then, rather than advance the phase number, this phaser
789 >     * will be set to a final termination state, and subsequent calls
790 >     * to {@link #isTerminated} will return true. Any (unchecked)
791 >     * Exception or Error thrown by an invocation of this method is
792 >     * propagated to the party attempting to advance this phaser, in
793 >     * which case no advance occurs.
794 >     *
795 >     * <p>The arguments to this method provide the state of the phaser
796 >     * prevailing for the current transition.  The effects of invoking
797 >     * arrival, registration, and waiting methods on this phaser from
798 >     * within {@code onAdvance} are unspecified and should not be
799 >     * relied on.
800 >     *
801 >     * <p>If this phaser is a member of a tiered set of phasers, then
802 >     * {@code onAdvance} is invoked only for its root phaser on each
803 >     * advance.
804 >     *
805 >     * <p>To support the most common use cases, the default
806 >     * implementation of this method returns {@code true} when the
807 >     * number of registered parties has become zero as the result of a
808 >     * party invoking {@code arriveAndDeregister}.  You can disable
809 >     * this behavior, thus enabling continuation upon future
810 >     * registrations, by overriding this method to always return
811 >     * {@code false}:
812 >     *
813 >     * <pre> {@code
814 >     * Phaser phaser = new Phaser() {
815 >     *   protected boolean onAdvance(int phase, int parties) { return false; }
816 >     * }}</pre>
817 >     *
818 >     * @param phase the current phase number on entry to this method,
819 >     * before this phaser is advanced
820 >     * @param registeredParties the current number of registered parties
821 >     * @return {@code true} if this phaser should terminate
822       */
823      protected boolean onAdvance(int phase, int registeredParties) {
824 <        return registeredParties <= 0;
824 >        return registeredParties == 0;
825      }
826  
827      /**
828 <     * Returns a string identifying this barrier, as well as its
828 >     * Returns a string identifying this phaser, as well as its
829       * state.  The state, in brackets, includes the String {@code
830 <     * "phase ="} followed by the phase number, {@code "parties ="}
830 >     * "phase = "} followed by the phase number, {@code "parties = "}
831       * followed by the number of registered parties, and {@code
832 <     * "arrived ="} followed by the number of arrived parties
832 >     * "arrived = "} followed by the number of arrived parties.
833       *
834 <     * @return a string identifying this barrier, as well as its state
834 >     * @return a string identifying this phaser, as well as its state
835       */
836      public String toString() {
837 <        long s = state.get();
475 <        return super.toString() + "[phase = " + phaseOf(s) + " parties = " + partiesOf(s) + " arrived = " + arrivedOf(s) + "]";
837 >        return stateToString(reconcileState());
838      }
839  
478    // methods for tripping and waiting
479
840      /**
841 <     * Advance the current phase (or terminate)
841 >     * Implementation of toString and string-based error messages
842       */
843 <    private void trip(int phase, int parties) {
844 <        int next = onAdvance(phase, parties)? -1 : ((phase + 1) & phaseMask);
845 <        state.set(stateFor(next, parties, parties));
846 <        if (head.get() != null)
847 <            releaseWaiters(next);
843 >    private String stateToString(long s) {
844 >        return super.toString() +
845 >            "[phase = " + phaseOf(s) +
846 >            " parties = " + partiesOf(s) +
847 >            " arrived = " + arrivedOf(s) + "]";
848      }
849  
850 <    private int helpingWait(int phase) {
851 <        final AtomicLong state = this.state;
852 <        int p;
853 <        while ((p = phaseOf(state.get())) == phase) {
854 <            ForkJoinTask<?> t = ForkJoinWorkerThread.pollTask();
855 <            if (t != null) {
856 <                if ((p = phaseOf(state.get())) == phase)
857 <                    t.exec();
858 <                else {   // push task and exit if barrier advanced
859 <                    t.fork();
860 <                    break;
861 <                }
850 >    // Waiting mechanics
851 >
852 >    /**
853 >     * Removes and signals threads from queue for phase.
854 >     */
855 >    private void releaseWaiters(int phase) {
856 >        QNode q;   // first element of queue
857 >        int p;     // its phase
858 >        Thread t;  // its thread
859 >        AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
860 >        while ((q = head.get()) != null &&
861 >               ((p = q.phase) == phase ||
862 >                (int)(root.state >>> PHASE_SHIFT) != p)) {
863 >            if (head.compareAndSet(q, q.next) &&
864 >                (t = q.thread) != null) {
865 >                q.thread = null;
866 >                LockSupport.unpark(t);
867              }
868          }
504        return p;
869      }
870  
871 <    private int timedHelpingWait(int phase, long nanos) throws TimeoutException {
872 <        final AtomicLong state = this.state;
873 <        long lastTime = System.nanoTime();
871 >    /** The number of CPUs, for spin control */
872 >    private static final int NCPU = Runtime.getRuntime().availableProcessors();
873 >
874 >    /**
875 >     * The number of times to spin before blocking while waiting for
876 >     * advance, per arrival while waiting. On multiprocessors, fully
877 >     * blocking and waking up a large number of threads all at once is
878 >     * usually a very slow process, so we use rechargeable spins to
879 >     * avoid it when threads regularly arrive: When a thread in
880 >     * internalAwaitAdvance notices another arrival before blocking,
881 >     * and there appear to be enough CPUs available, it spins
882 >     * SPINS_PER_ARRIVAL more times before blocking. The value trades
883 >     * off good-citizenship vs big unnecessary slowdowns.
884 >     */
885 >    static final int SPINS_PER_ARRIVAL = (NCPU < 2) ? 1 : 1 << 8;
886 >
887 >    /**
888 >     * Possibly blocks and waits for phase to advance unless aborted.
889 >     * Call only from root node.
890 >     *
891 >     * @param phase current phase
892 >     * @param node if non-null, the wait node to track interrupt and timeout;
893 >     * if null, denotes noninterruptible wait
894 >     * @return current phase
895 >     */
896 >    private int internalAwaitAdvance(int phase, QNode node) {
897 >        releaseWaiters(phase-1);          // ensure old queue clean
898 >        boolean queued = false;           // true when node is enqueued
899 >        int lastUnarrived = 0;            // to increase spins upon change
900 >        int spins = SPINS_PER_ARRIVAL;
901 >        long s;
902          int p;
903 <        while ((p = phaseOf(state.get())) == phase) {
904 <            long now = System.nanoTime();
905 <            nanos -= now - lastTime;
906 <            lastTime = now;
907 <            if (nanos <= 0) {
908 <                if ((p = phaseOf(state.get())) == phase)
909 <                    throw new TimeoutException();
910 <                else
911 <                    break;
912 <            }
521 <            ForkJoinTask<?> t = ForkJoinWorkerThread.pollTask();
522 <            if (t != null) {
523 <                if ((p = phaseOf(state.get())) == phase)
524 <                    t.exec();
525 <                else {   // push task and exit if barrier advanced
526 <                    t.fork();
527 <                    break;
903 >        while ((p = (int)((s = state) >>> PHASE_SHIFT)) == phase) {
904 >            if (node == null) {           // spinning in noninterruptible mode
905 >                int unarrived = (int)s & UNARRIVED_MASK;
906 >                if (unarrived != lastUnarrived &&
907 >                    (lastUnarrived = unarrived) < NCPU)
908 >                    spins += SPINS_PER_ARRIVAL;
909 >                boolean interrupted = Thread.interrupted();
910 >                if (interrupted || --spins < 0) { // need node to record intr
911 >                    node = new QNode(this, phase, false, false, 0L);
912 >                    node.wasInterrupted = interrupted;
913                  }
914              }
915 +            else if (node.isReleasable()) // done or aborted
916 +                break;
917 +            else if (!queued) {           // push onto queue
918 +                AtomicReference<QNode> head = (phase & 1) == 0 ? evenQ : oddQ;
919 +                QNode q = node.next = head.get();
920 +                if ((q == null || q.phase == phase) &&
921 +                    (int)(state >>> PHASE_SHIFT) == phase) // avoid stale enq
922 +                    queued = head.compareAndSet(q, node);
923 +            }
924 +            else {
925 +                try {
926 +                    ForkJoinPool.managedBlock(node);
927 +                } catch (InterruptedException ie) {
928 +                    node.wasInterrupted = true;
929 +                }
930 +            }
931 +        }
932 +
933 +        if (node != null) {
934 +            if (node.thread != null)
935 +                node.thread = null;       // avoid need for unpark()
936 +            if (node.wasInterrupted && !node.interruptible)
937 +                Thread.currentThread().interrupt();
938 +            if ((p = (int)(state >>> PHASE_SHIFT)) == phase)
939 +                return p;                 // recheck abort
940          }
941 +        releaseWaiters(phase);
942          return p;
943      }
944  
945      /**
946 <     * 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.
946 >     * Wait nodes for Treiber stack representing wait queue
947       */
948 <    static final class QNode {
949 <        QNode next;
541 <        volatile Thread thread; // nulled to cancel wait
948 >    static final class QNode implements ForkJoinPool.ManagedBlocker {
949 >        final Phaser phaser;
950          final int phase;
951 <        QNode(Thread t, int c) {
952 <            thread = t;
953 <            phase = c;
951 >        final boolean interruptible;
952 >        final boolean timed;
953 >        boolean wasInterrupted;
954 >        long nanos;
955 >        long lastTime;
956 >        volatile Thread thread; // nulled to cancel wait
957 >        QNode next;
958 >
959 >        QNode(Phaser phaser, int phase, boolean interruptible,
960 >              boolean timed, long nanos) {
961 >            this.phaser = phaser;
962 >            this.phase = phase;
963 >            this.interruptible = interruptible;
964 >            this.nanos = nanos;
965 >            this.timed = timed;
966 >            this.lastTime = timed ? System.nanoTime() : 0L;
967 >            thread = Thread.currentThread();
968          }
547    }
969  
970 <    private void releaseWaiters(int currentPhase) {
971 <        final AtomicReference<QNode> head = this.head;
972 <        QNode p;
973 <        while ((p = head.get()) != null && p.phase != currentPhase) {
974 <            if (head.compareAndSet(p, null)) {
975 <                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);
970 >        public boolean isReleasable() {
971 >            if (thread == null)
972 >                return true;
973 >            if (phaser.getPhase() != phase) {
974 >                thread = null;
975 >                return true;
976              }
977 +            if (Thread.interrupted())
978 +                wasInterrupted = true;
979 +            if (wasInterrupted && interruptible) {
980 +                thread = null;
981 +                return true;
982 +            }
983 +            if (timed) {
984 +                if (nanos > 0L) {
985 +                    long now = System.nanoTime();
986 +                    nanos -= now - lastTime;
987 +                    lastTime = now;
988 +                }
989 +                if (nanos <= 0L) {
990 +                    thread = null;
991 +                    return true;
992 +                }
993 +            }
994 +            return false;
995          }
563    }
564
565    /** The number of CPUs, for spin control */
566    static final int NCPUS = Runtime.getRuntime().availableProcessors();
996  
997 <    /**
998 <     * The number of times to spin before blocking in timed waits.
999 <     * The value is empirically derived.
1000 <     */
1001 <    static final int maxTimedSpins = (NCPUS < 2)? 0 : 32;
1002 <
1003 <    /**
1004 <     * The number of times to spin before blocking in untimed waits.
1005 <     * This is greater than timed value because untimed waits spin
1006 <     * faster since they don't need to check times on each spin.
578 <     */
579 <    static final int maxUntimedSpins = maxTimedSpins * 32;
997 >        public boolean block() {
998 >            if (isReleasable())
999 >                return true;
1000 >            else if (!timed)
1001 >                LockSupport.park(this);
1002 >            else if (nanos > 0)
1003 >                LockSupport.parkNanos(this, nanos);
1004 >            return isReleasable();
1005 >        }
1006 >    }
1007  
1008 <    /**
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;
1008 >    // Unsafe mechanics
1009  
1010 <    /**
1011 <     * Enqueues node and waits unless aborted or signalled.
1012 <     */
1013 <    private boolean untimedWait(Thread thread, int currentPhase,
1014 <                               boolean abortOnInterrupt) {
1015 <        final AtomicReference<QNode> head = this.head;
1016 <        final AtomicLong state = this.state;
1017 <        boolean wasInterrupted = false;
1018 <        QNode node = null;
1019 <        boolean queued = false;
1020 <        int spins = maxUntimedSpins;
1021 <        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;
1010 >    private static final sun.misc.Unsafe UNSAFE = getUnsafe();
1011 >    private static final long stateOffset =
1012 >        objectFieldOffset("state", Phaser.class);
1013 >
1014 >    private static long objectFieldOffset(String field, Class<?> klazz) {
1015 >        try {
1016 >            return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));
1017 >        } catch (NoSuchFieldException e) {
1018 >            // Convert Exception to corresponding Error
1019 >            NoSuchFieldError error = new NoSuchFieldError(field);
1020 >            error.initCause(e);
1021 >            throw error;
1022          }
630        if (node != null)
631            node.thread = null;
632        return wasInterrupted;
1023      }
1024  
1025      /**
1026 <     * Messier timeout version
1026 >     * Returns a sun.misc.Unsafe.  Suitable for use in a 3rd party package.
1027 >     * Replace with a simple call to Unsafe.getUnsafe when integrating
1028 >     * into a jdk.
1029 >     *
1030 >     * @return a sun.misc.Unsafe
1031       */
1032 <    private void timedWait(Thread thread, int currentPhase, long nanos)
1033 <        throws InterruptedException, TimeoutException {
1034 <        final AtomicReference<QNode> head = this.head;
1035 <        final AtomicLong state = this.state;
1036 <        long lastTime = System.nanoTime();
1037 <        QNode node = null;
1038 <        boolean queued = false;
1039 <        int spins = maxTimedSpins;
1040 <        while (phaseOf(state.get()) == currentPhase) {
1041 <            QNode h;
1042 <            long now = System.nanoTime();
1043 <            nanos -= now - lastTime;
1044 <            lastTime = now;
1045 <            if (nanos <= 0) {
1046 <                if (node != null)
1047 <                    node.thread = null;
1048 <                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 <                }
669 <            }
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;
1032 >    private static sun.misc.Unsafe getUnsafe() {
1033 >        try {
1034 >            return sun.misc.Unsafe.getUnsafe();
1035 >        } catch (SecurityException se) {
1036 >            try {
1037 >                return java.security.AccessController.doPrivileged
1038 >                    (new java.security
1039 >                     .PrivilegedExceptionAction<sun.misc.Unsafe>() {
1040 >                        public sun.misc.Unsafe run() throws Exception {
1041 >                            java.lang.reflect.Field f = sun.misc
1042 >                                .Unsafe.class.getDeclaredField("theUnsafe");
1043 >                            f.setAccessible(true);
1044 >                            return (sun.misc.Unsafe) f.get(null);
1045 >                        }});
1046 >            } catch (java.security.PrivilegedActionException e) {
1047 >                throw new RuntimeException("Could not initialize intrinsics",
1048 >                                           e.getCause());
1049              }
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;
1050          }
690        if (node != null)
691            node.thread = null;
1051      }
693
1052   }

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines