--- jsr166/src/jsr166y/Phaser.java	2009/08/02 17:02:06	1.25
+++ jsr166/src/jsr166y/Phaser.java	2010/11/06 16:12:10	1.50
@@ -7,157 +7,195 @@
 package jsr166y;
 
 import java.util.concurrent.*;
-
 import java.util.concurrent.atomic.AtomicReference;
 import java.util.concurrent.locks.LockSupport;
 
 /**
- * A reusable synchronization barrier, similar in functionality to a
+ * A reusable synchronization barrier, similar in functionality to
  * {@link java.util.concurrent.CyclicBarrier CyclicBarrier} and
  * {@link java.util.concurrent.CountDownLatch CountDownLatch}
  * but supporting more flexible usage.
  *
- * <ul>
- *
- * <li> The number of parties synchronizing on a phaser may vary over
- * time.  A task may register to be a party at any time, and may
- * deregister upon arriving at the barrier.  As is the case with most
- * basic synchronization constructs, registration and deregistration
- * affect only internal counts; they do not establish any further
- * internal bookkeeping, so tasks cannot query whether they are
- * registered. (However, you can introduce such bookkeeping by
- * subclassing this class.)
- *
- * <li> Each generation has an associated phase value, starting at
- * zero, and advancing when all parties reach the barrier (wrapping
- * around to zero after reaching {@code Integer.MAX_VALUE}).
- *
- * <li> Like a {@code CyclicBarrier}, a Phaser may be repeatedly
- * awaited.  Method {@link #arriveAndAwaitAdvance} has effect
- * analogous to {@link java.util.concurrent.CyclicBarrier#await
- * CyclicBarrier.await}.  However, phasers separate two aspects of
- * coordination, that may also be invoked independently:
+ * <p> <b>Registration.</b> Unlike the case for other barriers, the
+ * number of parties <em>registered</em> to synchronize on a phaser
+ * may vary over time.  Tasks may be registered at any time (using
+ * methods {@link #register}, {@link #bulkRegister}, or forms of
+ * constructors establishing initial numbers of parties), and
+ * optionally deregistered upon any arrival (using {@link
+ * #arriveAndDeregister}).  As is the case with most basic
+ * synchronization constructs, registration and deregistration affect
+ * only internal counts; they do not establish any further internal
+ * bookkeeping, so tasks cannot query whether they are registered.
+ * (However, you can introduce such bookkeeping by subclassing this
+ * class.)
+ *
+ * <p> <b>Synchronization.</b> Like a {@code CyclicBarrier}, a {@code
+ * Phaser} may be repeatedly awaited.  Method {@link
+ * #arriveAndAwaitAdvance} has effect analogous to {@link
+ * java.util.concurrent.CyclicBarrier#await CyclicBarrier.await}. Each
+ * generation of a {@code Phaser} has an associated phase number. The
+ * phase number starts at zero, and advances when all parties arrive
+ * at the barrier, wrapping around to zero after reaching {@code
+ * Integer.MAX_VALUE}. The use of phase numbers enables independent
+ * control of actions upon arrival at a barrier and upon awaiting
+ * others, via two kinds of methods that may be invoked by any
+ * registered party:
  *
  * <ul>
  *
- *   <li> Arriving at a barrier. Methods {@link #arrive} and
- *       {@link #arriveAndDeregister} do not block, but return
- *       the phase value current upon entry to the method.
- *
- *   <li> Awaiting others. Method {@link #awaitAdvance} requires an
- *       argument indicating the entry phase, and returns when the
- *       barrier advances to a new phase.
- * </ul>
+ *   <li> <b>Arrival.</b> Methods {@link #arrive} and
+ *       {@link #arriveAndDeregister} record arrival at a
+ *       barrier. These methods do not block, but return an associated
+ *       <em>arrival phase number</em>; that is, the phase number of
+ *       the barrier to which the arrival applied. When the final
+ *       party for a given phase arrives, an optional barrier action
+ *       is performed and the phase advances.  Barrier actions,
+ *       performed by the party triggering a phase advance, are
+ *       arranged by overriding method {@link #onAdvance(int, int)},
+ *       which also controls termination. Overriding this method is
+ *       similar to, but more flexible than, providing a barrier
+ *       action to a {@code CyclicBarrier}.
+ *
+ *   <li> <b>Waiting.</b> Method {@link #awaitAdvance} requires an
+ *       argument indicating an arrival phase number, and returns when
+ *       the barrier advances to (or is already at) a different phase.
+ *       Unlike similar constructions using {@code CyclicBarrier},
+ *       method {@code awaitAdvance} continues to wait even if the
+ *       waiting thread is interrupted. Interruptible and timeout
+ *       versions are also available, but exceptions encountered while
+ *       tasks wait interruptibly or with timeout do not change the
+ *       state of the barrier. If necessary, you can perform any
+ *       associated recovery within handlers of those exceptions,
+ *       often after invoking {@code forceTermination}.  Phasers may
+ *       also be used by tasks executing in a {@link ForkJoinPool},
+ *       which will ensure sufficient parallelism to execute tasks
+ *       when others are blocked waiting for a phase to advance.
  *
+ * </ul>
  *
- * <li> Barrier actions, performed by the task triggering a phase
- * advance while others may be waiting, are arranged by overriding
- * method {@link #onAdvance}, that also controls termination.
- * Overriding this method may be used to similar but more flexible
- * effect as providing a barrier action to a {@code CyclicBarrier}.
- *
- * <li> Phasers may enter a <em>termination</em> state in which all
- * actions immediately return without updating phaser state or waiting
- * for advance, and indicating (via a negative phase value) that
- * execution is complete.  Termination is triggered by executing the
- * overridable {@code onAdvance} method that is invoked each time the
- * barrier is about to be tripped. When a phaser is controlling an
- * action with a fixed number of iterations, it is often convenient to
- * override this method to cause termination when the current phase
- * number reaches a threshold. Method {@link #forceTermination} is also
- * available to abruptly release waiting threads and allow them to
- * terminate.
+ * <p> <b>Termination.</b> A {@code Phaser} may enter a
+ * <em>termination</em> state in which all synchronization methods
+ * immediately return without updating phaser state or waiting for
+ * advance, and indicating (via a negative phase value) that execution
+ * is complete.  Termination is triggered when an invocation of {@code
+ * onAdvance} returns {@code true}.  As illustrated below, when
+ * phasers control actions with a fixed number of iterations, it is
+ * often convenient to override this method to cause termination when
+ * the current phase number reaches a threshold. Method {@link
+ * #forceTermination} is also available to abruptly release waiting
+ * threads and allow them to terminate.
  *
- * <li> Phasers may be tiered to reduce contention. Phasers with large
+ * <p> <b>Tiering.</b> Phasers may be <em>tiered</em> (i.e., arranged
+ * in tree structures) to reduce contention. Phasers with large
  * numbers of parties that would otherwise experience heavy
- * synchronization contention costs may instead be arranged in trees.
- * This will typically greatly increase throughput even though it
- * incurs somewhat greater per-operation overhead.
- *
- * <li> By default, {@code awaitAdvance} continues to wait even if
- * the waiting thread is interrupted. And unlike the case in
- * {@code CyclicBarrier}, exceptions encountered while tasks wait
- * interruptibly or with timeout do not change the state of the
- * barrier. If necessary, you can perform any associated recovery
- * within handlers of those exceptions, often after invoking
- * {@code forceTermination}.
- *
- * <li>Phasers ensure lack of starvation when used by ForkJoinTasks.
- *
- * </ul>
+ * synchronization contention costs may instead be set up so that
+ * groups of sub-phasers share a common parent.  This may greatly
+ * increase throughput even though it incurs greater per-operation
+ * overhead.
+ *
+ * <p><b>Monitoring.</b> While synchronization methods may be invoked
+ * only by registered parties, the current state of a phaser may be
+ * monitored by any caller.  At any given moment there are {@link
+ * #getRegisteredParties} parties in total, of which {@link
+ * #getArrivedParties} have arrived at the current phase ({@link
+ * #getPhase}).  When the remaining ({@link #getUnarrivedParties})
+ * parties arrive, the phase advances.  The values returned by these
+ * methods may reflect transient states and so are not in general
+ * useful for synchronization control.  Method {@link #toString}
+ * returns snapshots of these state queries in a form convenient for
+ * informal monitoring.
  *
  * <p><b>Sample usages:</b>
  *
  * <p>A {@code Phaser} may be used instead of a {@code CountDownLatch}
- * to control a one-shot action serving a variable number of
- * parties. The typical idiom is for the method setting this up to
- * first register, then start the actions, then deregister, as in:
+ * to control a one-shot action serving a variable number of parties.
+ * The typical idiom is for the method setting this up to first
+ * register, then start the actions, then deregister, as in:
  *
  *  <pre> {@code
- * void runTasks(List<Runnable> list) {
+ * void runTasks(List<Runnable> tasks) {
  *   final Phaser phaser = new Phaser(1); // "1" to register self
- *   for (Runnable r : list) {
+ *   // create and start threads
+ *   for (Runnable task : tasks) {
  *     phaser.register();
  *     new Thread() {
  *       public void run() {
  *         phaser.arriveAndAwaitAdvance(); // await all creation
- *         r.run();
- *         phaser.arriveAndDeregister();   // signal completion
+ *         task.run();
  *       }
  *     }.start();
  *   }
  *
- *   doSomethingOnBehalfOfWorkers();
- *   phaser.arrive(); // allow threads to start
- *   int p = phaser.arriveAndDeregister(); // deregister self  ...
- *   p = phaser.awaitAdvance(p); // ... and await arrival
- *   otherActions(); // do other things while tasks execute
- *   phaser.awaitAdvance(p); // await final completion
+ *   // allow threads to start and deregister self
+ *   phaser.arriveAndDeregister();
  * }}</pre>
  *
  * <p>One way to cause a set of threads to repeatedly perform actions
  * for a given number of iterations is to override {@code onAdvance}:
  *
  *  <pre> {@code
- * void startTasks(List<Runnable> list, final int iterations) {
+ * void startTasks(List<Runnable> tasks, final int iterations) {
  *   final Phaser phaser = new Phaser() {
- *     public boolean onAdvance(int phase, int registeredParties) {
+ *     protected boolean onAdvance(int phase, int registeredParties) {
  *       return phase >= iterations || registeredParties == 0;
  *     }
  *   };
  *   phaser.register();
- *   for (Runnable r : list) {
+ *   for (final Runnable task : tasks) {
  *     phaser.register();
  *     new Thread() {
  *       public void run() {
  *         do {
- *           r.run();
+ *           task.run();
  *           phaser.arriveAndAwaitAdvance();
- *         } while(!phaser.isTerminated();
+ *         } while (!phaser.isTerminated());
  *       }
  *     }.start();
  *   }
  *   phaser.arriveAndDeregister(); // deregister self, don't wait
  * }}</pre>
  *
+ * If the main task must later await termination, it
+ * may re-register and then execute a similar loop:
+ *  <pre> {@code
+ *   // ...
+ *   phaser.register();
+ *   while (!phaser.isTerminated())
+ *     phaser.arriveAndAwaitAdvance();}</pre>
+ *
+ * <p>Related constructions may be used to await particular phase numbers
+ * in contexts where you are sure that the phase will never wrap around
+ * {@code Integer.MAX_VALUE}. For example:
+ *
+ *  <pre> {@code
+ * void awaitPhase(Phaser phaser, int phase) {
+ *   int p = phaser.register(); // assumes caller not already registered
+ *   while (p < phase) {
+ *     if (phaser.isTerminated())
+ *       // ... deal with unexpected termination
+ *     else
+ *       p = phaser.arriveAndAwaitAdvance();
+ *   }
+ *   phaser.arriveAndDeregister();
+ * }}</pre>
+ *
+ *
  * <p>To create a set of tasks using a tree of phasers,
  * you could use code of the following form, assuming a
  * Task class with a constructor accepting a phaser that
- * it registers for upon construction:
+ * it registers with upon construction:
+ *
  *  <pre> {@code
- * void build(Task[] actions, int lo, int hi, Phaser b) {
- *   int step = (hi - lo) / TASKS_PER_PHASER;
- *   if (step > 1) {
- *     int i = lo;
- *     while (i < hi) {
- *       int r = Math.min(i + step, hi);
- *       build(actions, i, r, new Phaser(b));
- *       i = r;
+ * void build(Task[] actions, int lo, int hi, Phaser ph) {
+ *   if (hi - lo > TASKS_PER_PHASER) {
+ *     for (int i = lo; i < hi; i += TASKS_PER_PHASER) {
+ *       int j = Math.min(i + TASKS_PER_PHASER, hi);
+ *       build(actions, i, j, new Phaser(ph));
  *     }
  *   } else {
  *     for (int i = lo; i < hi; ++i)
- *       actions[i] = new Task(b);
- *       // assumes new Task(b) performs b.register()
+ *       actions[i] = new Task(ph);
+ *       // assumes new Task(ph) performs ph.register()
  *   }
  * }
  * // .. initially called, for n tasks via
@@ -168,11 +206,9 @@ import java.util.concurrent.locks.LockSu
  * be appropriate for extremely small per-barrier task bodies (thus
  * high rates), or up to hundreds for extremely large ones.
  *
- * </pre>
- *
  * <p><b>Implementation notes</b>: This implementation restricts the
  * maximum number of parties to 65535. Attempts to register additional
- * parties result in IllegalStateExceptions. However, you can and
+ * parties result in {@code IllegalStateException}. However, you can and
  * should create tiered phasers to accommodate arbitrarily large sets
  * of participants.
  *
@@ -206,7 +242,6 @@ public class Phaser {
      */
     private volatile long state;
 
-    private static final int ushortBits = 16;
     private static final int ushortMask = 0xffff;
     private static final int phaseMask  = 0x7fffffff;
 
@@ -259,11 +294,12 @@ public class Phaser {
 
     /**
      * Heads of Treiber stacks for waiting threads. To eliminate
-     * contention while releasing some threads while adding others, we
+     * contention when releasing some threads while adding others, we
      * use two of them, alternating across even and odd phases.
+     * Subphasers share queues with root to speed up releases.
      */
-    private final AtomicReference<QNode> evenQ = new AtomicReference<QNode>();
-    private final AtomicReference<QNode> oddQ  = new AtomicReference<QNode>();
+    private final AtomicReference<QNode> evenQ;
+    private final AtomicReference<QNode> oddQ;
 
     private AtomicReference<QNode> queueFor(int phase) {
         return ((phase & 1) == 0) ? evenQ : oddQ;
@@ -281,20 +317,20 @@ public class Phaser {
      * Recursively resolves state.
      */
     private long reconcileState() {
-        Phaser p = parent;
+        Phaser par = parent;
         long s = state;
-        if (p != null) {
-            while (unarrivedOf(s) == 0 && phaseOf(s) != phaseOf(root.state)) {
-                long parentState = p.getReconciledState();
-                int parentPhase = phaseOf(parentState);
-                int phase = phaseOf(s = state);
-                if (phase != parentPhase) {
+        if (par != null) {
+            int phase, rootPhase;
+            while ((phase = phaseOf(s)) >= 0 &&
+                   (rootPhase = phaseOf(root.state)) != phase &&
+                   (rootPhase < 0 || unarrivedOf(s) == 0)) {
+                int parentPhase = phaseOf(par.getReconciledState());
+                if (parentPhase != phase) {
                     long next = trippedStateFor(parentPhase, partiesOf(s));
-                    if (casState(s, next)) {
-                        releaseWaiters(phase);
-                        s = next;
-                    }
+                    if (state == s)
+                        UNSAFE.compareAndSwapLong(this, stateOffset, s, next);
                 }
+                s = state;
             }
         }
         return s;
@@ -306,11 +342,11 @@ public class Phaser {
      * phaser will need to first register for it.
      */
     public Phaser() {
-        this(null);
+        this(null, 0);
     }
 
     /**
-     * Creates a new phaser with the given numbers of registered
+     * Creates a new phaser with the given number of registered
      * unarrived parties, initial phase number 0, and no parent.
      *
      * @param parties the number of parties required to trip barrier
@@ -330,19 +366,11 @@ public class Phaser {
      * @param parent the parent phaser
      */
     public Phaser(Phaser parent) {
-        int phase = 0;
-        this.parent = parent;
-        if (parent != null) {
-            this.root = parent.root;
-            phase = parent.register();
-        }
-        else
-            this.root = this;
-        this.state = trippedStateFor(phase, 0);
+        this(parent, 0);
     }
 
     /**
-     * Creates a new phaser with the given parent and numbers of
+     * Creates a new phaser with the given parent and number of
      * registered unarrived parties. If parent is non-null, this phaser
      * is registered with the parent and its initial phase number is
      * the same as that of parent phaser.
@@ -355,21 +383,30 @@ public class Phaser {
     public Phaser(Phaser parent, int parties) {
         if (parties < 0 || parties > ushortMask)
             throw new IllegalArgumentException("Illegal number of parties");
-        int phase = 0;
+        int phase;
         this.parent = parent;
         if (parent != null) {
-            this.root = parent.root;
+            Phaser r = parent.root;
+            this.root = r;
+            this.evenQ = r.evenQ;
+            this.oddQ = r.oddQ;
             phase = parent.register();
         }
-        else
+        else {
             this.root = this;
+            this.evenQ = new AtomicReference<QNode>();
+            this.oddQ = new AtomicReference<QNode>();
+            phase = 0;
+        }
         this.state = trippedStateFor(phase, parties);
     }
 
     /**
      * Adds a new unarrived party to this phaser.
+     * If an ongoing invocation of {@link #onAdvance} is in progress,
+     * this method may wait until its completion before registering.
      *
-     * @return the current barrier phase number upon registration
+     * @return the arrival phase number to which this registration applied
      * @throws IllegalStateException if attempting to register more
      * than the maximum supported number of parties
      */
@@ -379,11 +416,14 @@ public class Phaser {
 
     /**
      * Adds the given number of new unarrived parties to this phaser.
+     * If an ongoing invocation of {@link #onAdvance} is in progress,
+     * this method may wait until its completion before registering.
      *
-     * @param parties the number of parties required to trip barrier
-     * @return the current barrier phase number upon registration
+     * @param parties the number of additional parties required to trip barrier
+     * @return the arrival phase number to which this registration applied
      * @throws IllegalStateException if attempting to register more
      * than the maximum supported number of parties
+     * @throws IllegalArgumentException if {@code parties < 0}
      */
     public int bulkRegister(int parties) {
         if (parties < 0)
@@ -397,135 +437,131 @@ public class Phaser {
      * Shared code for register, bulkRegister
      */
     private int doRegister(int registrations) {
+        Phaser par = parent;
+        long s;
         int phase;
-        for (;;) {
-            long s = getReconciledState();
-            phase = phaseOf(s);
-            int unarrived = unarrivedOf(s) + registrations;
-            int parties = partiesOf(s) + registrations;
-            if (phase < 0)
-                break;
-            if (parties > ushortMask || unarrived > ushortMask)
+        while ((phase = phaseOf(s = par==null? state:reconcileState())) >= 0) {
+            int p = partiesOf(s);
+            int u = unarrivedOf(s);
+            int unarrived = u + registrations;
+            int parties = p + registrations;
+            if (u == 0 && p != 0)  // if tripped, wait for advance
+                untimedWait(phase);
+            else if (parties > ushortMask)
                 throw new IllegalStateException(badBounds(parties, unarrived));
-            if (phase == phaseOf(root.state) &&
-                casState(s, stateFor(phase, parties, unarrived)))
-                break;
+            else if (par == null || phaseOf(root.state) == phase) {
+                long next = stateFor(phase, parties, unarrived);
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
+                    break;
+            }
         }
         return phase;
     }
 
     /**
      * Arrives at the barrier, but does not wait for others.  (You can
-     * in turn wait for others via {@link #awaitAdvance}).
+     * in turn wait for others via {@link #awaitAdvance}).  It is an
+     * unenforced usage error for an unregistered party to invoke this
+     * method.
      *
-     * @return the barrier phase number upon entry to this method, or a
-     * negative value if terminated
+     * @return the arrival phase number, or a negative value if terminated
      * @throws IllegalStateException if not terminated and the number
      * of unarrived parties would become negative
      */
     public int arrive() {
+        Phaser par = parent;
+        long s;
         int phase;
-        for (;;) {
-            long s = state;
-            phase = phaseOf(s);
-            if (phase < 0)
-                break;
+        while ((phase = phaseOf(s = par==null? state:reconcileState())) >= 0) {
             int parties = partiesOf(s);
             int unarrived = unarrivedOf(s) - 1;
-            if (unarrived > 0) {        // Not the last arrival
-                if (casState(s, s - 1)) // s-1 adds one arrival
-                    break;
+            if (unarrived > 0) {                // Not the last arrival
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, s - 1))
+                    break;                      // s-1 adds one arrival
             }
-            else if (unarrived == 0) {  // the last arrival
-                Phaser par = parent;
-                if (par == null) {      // directly trip
-                    if (casState
-                        (s,
-                         trippedStateFor(onAdvance(phase, parties) ? -1 :
-                                         ((phase + 1) & phaseMask), parties))) {
-                        releaseWaiters(phase);
-                        break;
-                    }
-                }
-                else {                  // cascade to parent
-                    if (casState(s, s - 1)) { // zeroes unarrived
-                        par.arrive();
-                        reconcileState();
-                        break;
-                    }
+            else if (unarrived < 0)
+                throw new IllegalStateException(badBounds(parties, unarrived));
+            else if (par == null) {             // directly trip
+                long next = trippedStateFor(onAdvance(phase, parties) ? -1 :
+                                            ((phase + 1) & phaseMask),
+                                            parties);
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next)) {
+                    releaseWaiters(phase);
+                    break;
                 }
             }
-            else if (phase != phaseOf(root.state)) // or if unreconciled
+            else if (phaseOf(root.state) == phase &&
+                     UNSAFE.compareAndSwapLong(this, stateOffset, s, s - 1)) {
+                par.arrive();                   // cascade to parent
                 reconcileState();
-            else
-                throw new IllegalStateException(badBounds(parties, unarrived));
+                break;
+            }
         }
         return phase;
     }
 
     /**
-     * Arrives at the barrier, and deregisters from it, without
-     * waiting for others. Deregistration reduces number of parties
+     * Arrives at the barrier and deregisters from it without waiting
+     * for others. Deregistration reduces the number of parties
      * required to trip the barrier in future phases.  If this phaser
      * has a parent, and deregistration causes this phaser to have
-     * zero parties, this phaser is also deregistered from its parent.
+     * zero parties, this phaser also arrives at and is deregistered
+     * from its parent.  It is an unenforced usage error for an
+     * unregistered party to invoke this method.
      *
-     * @return the current barrier phase number upon entry to
-     * this method, or a negative value if terminated
+     * @return the arrival phase number, or a negative value if terminated
      * @throws IllegalStateException if not terminated and the number
      * of registered or unarrived parties would become negative
      */
     public int arriveAndDeregister() {
-        // similar code to arrive, but too different to merge
+        // similar to arrive, but too different to merge
         Phaser par = parent;
+        long s;
         int phase;
-        for (;;) {
-            long s = state;
-            phase = phaseOf(s);
-            if (phase < 0)
-                break;
+        while ((phase = phaseOf(s = par==null? state:reconcileState())) >= 0) {
             int parties = partiesOf(s) - 1;
             int unarrived = unarrivedOf(s) - 1;
-            if (parties >= 0) {
-                if (unarrived > 0 || (unarrived == 0 && par != null)) {
-                    if (casState
-                        (s,
-                         stateFor(phase, parties, unarrived))) {
-                        if (unarrived == 0) {
-                            par.arriveAndDeregister();
-                            reconcileState();
-                        }
-                        break;
-                    }
-                    continue;
-                }
-                if (unarrived == 0) {
-                    if (casState
-                        (s,
-                         trippedStateFor(onAdvance(phase, parties) ? -1 :
-                                         ((phase + 1) & phaseMask), parties))) {
-                        releaseWaiters(phase);
-                        break;
-                    }
-                    continue;
+            if (unarrived > 0) {
+                long next = stateFor(phase, parties, unarrived);
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next))
+                    break;
+            }
+            else if (unarrived < 0)
+                throw new IllegalStateException(badBounds(parties, unarrived));
+            else if (par == null) {
+                long next = trippedStateFor(onAdvance(phase, parties)? -1:
+                                            (phase + 1) & phaseMask,
+                                            parties);
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next)) {
+                    releaseWaiters(phase);
+                    break;
                 }
-                if (par != null && phase != phaseOf(root.state)) {
+            }
+            else if (phaseOf(root.state) == phase) {
+                long next = stateFor(phase, parties, 0);
+                if (UNSAFE.compareAndSwapLong(this, stateOffset, s, next)) {
+                    if (parties == 0)
+                        par.arriveAndDeregister();
+                    else
+                        par.arrive();
                     reconcileState();
-                    continue;
+                    break;
                 }
             }
-            throw new IllegalStateException(badBounds(parties, unarrived));
         }
         return phase;
     }
 
     /**
      * Arrives at the barrier and awaits others. Equivalent in effect
-     * to {@code awaitAdvance(arrive())}.  If you instead need to
-     * await with interruption of timeout, and/or deregister upon
-     * arrival, you can arrange them using analogous constructions.
+     * to {@code awaitAdvance(arrive())}.  If you need to await with
+     * interruption or timeout, you can arrange this with an analogous
+     * construction using one of the other forms of the {@code
+     * awaitAdvance} method.  If instead you need to deregister upon
+     * arrival, use {@link #arriveAndDeregister}. It is an unenforced
+     * usage error for an unregistered party to invoke this method.
      *
-     * @return the phase on entry to this method
+     * @return the arrival phase number, or a negative number if terminated
      * @throws IllegalStateException if not terminated and the number
      * of unarrived parties would become negative
      */
@@ -534,71 +570,80 @@ public class Phaser {
     }
 
     /**
-     * Awaits the phase of the barrier to advance from the given
-     * value, or returns immediately if argument is negative or this
-     * barrier is terminated.
-     *
-     * @param phase the phase on entry to this method
-     * @return the phase on exit from this method
+     * Awaits the phase of the barrier to advance from the given phase
+     * value, returning immediately if the current phase of the
+     * barrier is not equal to the given phase value or this barrier
+     * is terminated.
+     *
+     * @param phase an arrival phase number, or negative value if
+     * terminated; this argument is normally the value returned by a
+     * previous call to {@code arrive} or its variants
+     * @return the next arrival phase number, or a negative value
+     * if terminated or argument is negative
      */
     public int awaitAdvance(int phase) {
         if (phase < 0)
             return phase;
-        long s = getReconciledState();
-        int p = phaseOf(s);
+        int p = getPhase();
         if (p != phase)
             return p;
-        if (unarrivedOf(s) == 0 && parent != null)
-            parent.awaitAdvance(phase);
-        // Fall here even if parent waited, to reconcile and help release
         return untimedWait(phase);
     }
 
     /**
-     * Awaits the phase of the barrier to advance from the given
-     * value, or returns immediately if argument is negative or this
-     * barrier is terminated, or throws InterruptedException if
-     * interrupted while waiting.
+     * Awaits the phase of the barrier to advance from the given phase
+     * value, throwing {@code InterruptedException} if interrupted
+     * while waiting, or returning immediately if the current phase of
+     * the barrier is not equal to the given phase value or this
+     * barrier is terminated.
      *
-     * @param phase the phase on entry to this method
-     * @return the phase on exit from this method
+     * @param phase an arrival phase number, or negative value if
+     * terminated; this argument is normally the value returned by a
+     * previous call to {@code arrive} or its variants
+     * @return the next arrival phase number, or a negative value
+     * if terminated or argument is negative
      * @throws InterruptedException if thread interrupted while waiting
      */
     public int awaitAdvanceInterruptibly(int phase)
         throws InterruptedException {
         if (phase < 0)
             return phase;
-        long s = getReconciledState();
-        int p = phaseOf(s);
+        int p = getPhase();
         if (p != phase)
             return p;
-        if (unarrivedOf(s) == 0 && parent != null)
-            parent.awaitAdvanceInterruptibly(phase);
         return interruptibleWait(phase);
     }
 
     /**
-     * Awaits the phase of the barrier to advance from the given value
-     * or the given timeout elapses, or returns immediately if
-     * argument is negative or this barrier is terminated.
-     *
-     * @param phase the phase on entry to this method
-     * @return the phase on exit from this method
+     * Awaits the phase of the barrier to advance from the given phase
+     * value or the given timeout to elapse, throwing {@code
+     * InterruptedException} if interrupted while waiting, or
+     * returning immediately if the current phase of the barrier is
+     * not equal to the given phase value or this barrier is
+     * terminated.
+     *
+     * @param phase an arrival phase number, or negative value if
+     * terminated; this argument is normally the value returned by a
+     * previous call to {@code arrive} or its variants
+     * @param timeout how long to wait before giving up, in units of
+     *        {@code unit}
+     * @param unit a {@code TimeUnit} determining how to interpret the
+     *        {@code timeout} parameter
+     * @return the next arrival phase number, or a negative value
+     * if terminated or argument is negative
      * @throws InterruptedException if thread interrupted while waiting
      * @throws TimeoutException if timed out while waiting
      */
     public int awaitAdvanceInterruptibly(int phase,
                                          long timeout, TimeUnit unit)
         throws InterruptedException, TimeoutException {
+        long nanos = unit.toNanos(timeout);
         if (phase < 0)
             return phase;
-        long s = getReconciledState();
-        int p = phaseOf(s);
+        int p = getPhase();
         if (p != phase)
             return p;
-        if (unarrivedOf(s) == 0 && parent != null)
-            parent.awaitAdvanceInterruptibly(phase, timeout, unit);
-        return timedWait(phase, unit.toNanos(timeout));
+        return timedWait(phase, nanos);
     }
 
     /**
@@ -609,20 +654,15 @@ public class Phaser {
      * unexpected exceptions.
      */
     public void forceTermination() {
-        for (;;) {
-            long s = getReconciledState();
-            int phase = phaseOf(s);
-            int parties = partiesOf(s);
-            int unarrived = unarrivedOf(s);
-            if (phase < 0 ||
-                casState(s, stateFor(-1, parties, unarrived))) {
-                releaseWaiters(0);
-                releaseWaiters(1);
-                if (parent != null)
-                    parent.forceTermination();
-                return;
-            }
-        }
+        Phaser r = root;    // force at root then reconcile
+        long s;
+        while (phaseOf(s = r.state) >= 0)
+            UNSAFE.compareAndSwapLong(r, stateOffset, s,
+                                      stateFor(-1, partiesOf(s),
+                                               unarrivedOf(s)));
+        reconcileState();
+        releaseWaiters(0);  // ensure wakeups on both queues
+        releaseWaiters(1);
     }
 
     /**
@@ -637,32 +677,22 @@ public class Phaser {
     }
 
     /**
-     * Returns {@code true} if the current phase number equals the given phase.
-     *
-     * @param phase the phase
-     * @return {@code true} if the current phase number equals the given phase
-     */
-    public final boolean hasPhase(int phase) {
-        return phaseOf(getReconciledState()) == phase;
-    }
-
-    /**
      * Returns the number of parties registered at this barrier.
      *
      * @return the number of parties
      */
     public int getRegisteredParties() {
-        return partiesOf(state);
+        return partiesOf(getReconciledState());
     }
 
     /**
-     * Returns the number of parties that have arrived at the current
-     * phase of this barrier.
+     * Returns the number of registered parties that have arrived at
+     * the current phase of this barrier.
      *
      * @return the number of arrived parties
      */
     public int getArrivedParties() {
-        return arrivedOf(state);
+        return arrivedOf(getReconciledState());
     }
 
     /**
@@ -672,7 +702,7 @@ public class Phaser {
      * @return the number of unarrived parties
      */
     public int getUnarrivedParties() {
-        return unarrivedOf(state);
+        return unarrivedOf(getReconciledState());
     }
 
     /**
@@ -704,30 +734,33 @@ public class Phaser {
     }
 
     /**
-     * Overridable method to perform an action upon phase advance, and
-     * to control termination. This method is invoked whenever the
-     * barrier is tripped (and thus all other waiting parties are
-     * dormant). If it returns {@code true}, then, rather than advance
-     * the phase number, this barrier will be set to a final
-     * termination state, and subsequent calls to {@link #isTerminated}
-     * will return true.
+     * Overridable method to perform an action upon impending phase
+     * advance, and to control termination. This method is invoked
+     * upon arrival of the party tripping the barrier (when all other
+     * waiting parties are dormant).  If this method returns {@code
+     * true}, then, rather than advance the phase number, this barrier
+     * will be set to a final termination state, and subsequent calls
+     * to {@link #isTerminated} will return true. Any (unchecked)
+     * Exception or Error thrown by an invocation of this method is
+     * propagated to the party attempting to trip the barrier, in
+     * which case no advance occurs.
+     *
+     * <p>The arguments to this method provide the state of the phaser
+     * prevailing for the current transition.  The results and effects
+     * of invoking phase-related methods (including {@code getPhase}
+     * as well as arrival, registration, and waiting methods) from
+     * within {@code onAdvance} are unspecified and should not be
+     * relied on. Similarly, while it is possible to override this
+     * method to produce side-effects visible to participating tasks,
+     * it is in general safe to do so only in designs in which all
+     * parties register before any arrive, and all {@link
+     * #awaitAdvance} at each phase.
      *
      * <p>The default version returns {@code true} when the number of
      * registered parties is zero. Normally, overrides that arrange
      * termination for other reasons should also preserve this
      * property.
      *
-     * <p>You may override this method to perform an action with side
-     * effects visible to participating tasks, but it is in general
-     * only sensible to do so in designs where all parties register
-     * before any arrive, and all {@link #awaitAdvance} at each phase.
-     * Otherwise, you cannot ensure lack of interference. In
-     * particular, this method may be invoked more than once per
-     * transition if other parties successfully register while the
-     * invocation of this method is in progress, thus postponing the
-     * transition until those parties also arrive, re-triggering this
-     * method.
-     *
      * @param phase the phase number on entering the barrier
      * @param registeredParties the current number of registered parties
      * @return {@code true} if this barrier should terminate
@@ -768,6 +801,7 @@ public class Phaser {
         volatile boolean wasInterrupted = false;
         volatile Thread thread; // nulled to cancel wait
         QNode next;
+
         QNode(Phaser phaser, int phase, boolean interruptible,
               boolean timed, long startTime, long nanos) {
             this.phaser = phaser;
@@ -778,12 +812,14 @@ public class Phaser {
             this.nanos = nanos;
             thread = Thread.currentThread();
         }
+
         public boolean isReleasable() {
             return (thread == null ||
                     phaser.getPhase() != phase ||
                     (interruptible && wasInterrupted) ||
                     (timed && (nanos - (System.nanoTime() - startTime)) <= 0));
         }
+
         public boolean block() {
             if (Thread.interrupted()) {
                 wasInterrupted = true;
@@ -800,6 +836,7 @@ public class Phaser {
             }
             return isReleasable();
         }
+
         void signal() {
             Thread t = thread;
             if (t != null) {
@@ -807,16 +844,17 @@ public class Phaser {
                 LockSupport.unpark(t);
             }
         }
+
         boolean doWait() {
             if (thread != null) {
                 try {
-                    ForkJoinPool.managedBlock(this, false);
+                    ForkJoinPool.managedBlock(this);
                 } catch (InterruptedException ie) {
+                    wasInterrupted = true; // can't currently happen
                 }
             }
             return wasInterrupted;
         }
-
     }
 
     /**
@@ -842,6 +880,12 @@ public class Phaser {
     }
 
     /**
+     * The number of times to spin before blocking waiting for advance.
+     */
+    static final int MAX_SPINS =
+        Runtime.getRuntime().availableProcessors() == 1 ? 0 : 1 << 8;
+
+    /**
      * Enqueues node and waits unless aborted or signalled.
      *
      * @return current phase
@@ -850,16 +894,21 @@ public class Phaser {
         QNode node = null;
         boolean queued = false;
         boolean interrupted = false;
+        int spins = MAX_SPINS;
         int p;
         while ((p = getPhase()) == phase) {
             if (Thread.interrupted())
                 interrupted = true;
+            else if (spins > 0) {
+                if (--spins == 0)
+                    Thread.yield();
+            }
             else if (node == null)
                 node = new QNode(this, phase, false, false, 0, 0);
             else if (!queued)
                 queued = tryEnqueue(node);
-            else
-                interrupted = node.doWait();
+            else if (node.doWait())
+                interrupted = true;
         }
         if (node != null)
             node.thread = null;
@@ -877,16 +926,21 @@ public class Phaser {
         QNode node = null;
         boolean queued = false;
         boolean interrupted = false;
+        int spins = MAX_SPINS;
         int p;
         while ((p = getPhase()) == phase && !interrupted) {
             if (Thread.interrupted())
                 interrupted = true;
+            else if (spins > 0) {
+                if (--spins == 0)
+                    Thread.yield();
+            }
             else if (node == null)
                 node = new QNode(this, phase, true, false, 0, 0);
             else if (!queued)
                 queued = tryEnqueue(node);
-            else
-                interrupted = node.doWait();
+            else if (node.doWait())
+                interrupted = true;
         }
         if (node != null)
             node.thread = null;
@@ -907,18 +961,23 @@ public class Phaser {
         QNode node = null;
         boolean queued = false;
         boolean interrupted = false;
+        int spins = MAX_SPINS;
         int p;
         while ((p = getPhase()) == phase && !interrupted) {
             if (Thread.interrupted())
                 interrupted = true;
             else if (nanos - (System.nanoTime() - startTime) <= 0)
                 break;
+            else if (spins > 0) {
+                if (--spins == 0)
+                    Thread.yield();
+            }
             else if (node == null)
                 node = new QNode(this, phase, true, true, startTime, nanos);
             else if (!queued)
                 queued = tryEnqueue(node);
-            else
-                interrupted = node.doWait();
+            else if (node.doWait())
+                interrupted = true;
         }
         if (node != null)
             node.thread = null;
@@ -937,10 +996,6 @@ public class Phaser {
     private static final long stateOffset =
         objectFieldOffset("state", Phaser.class);
 
-    private final boolean casState(long cmp, long val) {
-        return UNSAFE.compareAndSwapLong(this, stateOffset, cmp, val);
-    }
-
     private static long objectFieldOffset(String field, Class<?> klazz) {
         try {
             return UNSAFE.objectFieldOffset(klazz.getDeclaredField(field));