ViewVC Help
View File | Revision Log | Show Annotations | Download File | Root Listing
root/jsr166/jsr166/src/jsr166e/ForkJoinPool.java
(Generate patch)

Comparing jsr166/src/jsr166e/ForkJoinPool.java (file contents):
Revision 1.57 by jsr166, Sat Feb 16 20:50:29 2013 UTC vs.
Revision 1.58 by dl, Wed Jun 19 14:55:40 2013 UTC

# Line 6 | Line 6
6  
7   package jsr166e;
8  
9 + import java.lang.Thread.UncaughtExceptionHandler;
10   import java.util.ArrayList;
11   import java.util.Arrays;
12   import java.util.Collection;
# Line 17 | Line 18 | import java.util.concurrent.ExecutorServ
18   import java.util.concurrent.Future;
19   import java.util.concurrent.RejectedExecutionException;
20   import java.util.concurrent.RunnableFuture;
21 + import java.util.concurrent.ThreadLocalRandom;
22   import java.util.concurrent.TimeUnit;
23  
24   /**
# Line 49 | Line 51 | import java.util.concurrent.TimeUnit;
51   * level; by default, equal to the number of available processors. The
52   * pool attempts to maintain enough active (or available) threads by
53   * dynamically adding, suspending, or resuming internal worker
54 < * threads, even if some tasks are stalled waiting to join
55 < * others. However, no such adjustments are guaranteed in the face of
56 < * blocked I/O or other unmanaged synchronization. The nested {@link
54 > * threads, even if some tasks are stalled waiting to join others.
55 > * However, no such adjustments are guaranteed in the face of blocked
56 > * I/O or other unmanaged synchronization. The nested {@link
57   * ManagedBlocker} interface enables extension of the kinds of
58   * synchronization accommodated.
59   *
# Line 75 | Line 77 | import java.util.concurrent.TimeUnit;
77   * there is little difference among choice of methods.
78   *
79   * <table BORDER CELLPADDING=3 CELLSPACING=1>
80 + * <caption>Summary of task execution methods</caption>
81   *  <tr>
82   *    <td></td>
83   *    <td ALIGN=CENTER> <b>Call from non-fork/join clients</b></td>
84   *    <td ALIGN=CENTER> <b>Call from within fork/join computations</b></td>
85   *  </tr>
86   *  <tr>
87 < *    <td> <b>Arrange async execution</td>
87 > *    <td> <b>Arrange async execution</b></td>
88   *    <td> {@link #execute(ForkJoinTask)}</td>
89   *    <td> {@link ForkJoinTask#fork}</td>
90   *  </tr>
91   *  <tr>
92 < *    <td> <b>Await and obtain result</td>
92 > *    <td> <b>Await and obtain result</b></td>
93   *    <td> {@link #invoke(ForkJoinTask)}</td>
94   *    <td> {@link ForkJoinTask#invoke}</td>
95   *  </tr>
96   *  <tr>
97 < *    <td> <b>Arrange exec and obtain Future</td>
97 > *    <td> <b>Arrange exec and obtain Future</b></td>
98   *    <td> {@link #submit(ForkJoinTask)}</td>
99   *    <td> {@link ForkJoinTask#fork} (ForkJoinTasks <em>are</em> Futures)</td>
100   *  </tr>
101   * </table>
102   *
103   * <p>The common pool is by default constructed with default
104 < * parameters, but these may be controlled by setting three {@link
105 < * System#getProperty system properties} with prefix {@code
106 < * java.util.concurrent.ForkJoinPool.common}: {@code parallelism} --
107 < * an integer greater than zero, {@code threadFactory} -- the class
108 < * name of a {@link ForkJoinWorkerThreadFactory}, and {@code
109 < * exceptionHandler} -- the class name of a {@link
110 < * java.lang.Thread.UncaughtExceptionHandler
111 < * Thread.UncaughtExceptionHandler}. Upon any error in establishing
112 < * these settings, default parameters are used.
104 > * parameters, but these may be controlled by setting three
105 > * {@linkplain System#getProperty system properties}:
106 > * <ul>
107 > * <li>{@code java.util.concurrent.ForkJoinPool.common.parallelism}
108 > * - the parallelism level, a non-negative integer
109 > * <li>{@code java.util.concurrent.ForkJoinPool.common.threadFactory}
110 > * - the class name of a {@link ForkJoinWorkerThreadFactory}
111 > * <li>{@code java.util.concurrent.ForkJoinPool.common.exceptionHandler}
112 > * - the class name of a {@link UncaughtExceptionHandler}
113 > * </ul>
114 > * The system class loader is used to load these classes.
115 > * Upon any error in establishing these settings, default parameters
116 > * are used. It is possible to disable or limit the use of threads in
117 > * the common pool by setting the parallelism property to zero, and/or
118 > * using a factory that may return {@code null}.
119   *
120   * <p><b>Implementation notes</b>: This implementation restricts the
121   * maximum number of running threads to 32767. Attempts to create
# Line 152 | Line 161 | public class ForkJoinPool extends Abstra
161       * (http://research.sun.com/scalable/pubs/index.html) and
162       * "Idempotent work stealing" by Michael, Saraswat, and Vechev,
163       * PPoPP 2009 (http://portal.acm.org/citation.cfm?id=1504186).
164 <     * The main differences ultimately stem from GC requirements that
165 <     * we null out taken slots as soon as we can, to maintain as small
166 <     * a footprint as possible even in programs generating huge
167 <     * numbers of tasks. To accomplish this, we shift the CAS
168 <     * arbitrating pop vs poll (steal) from being on the indices
169 <     * ("base" and "top") to the slots themselves.  So, both a
170 <     * successful pop and poll mainly entail a CAS of a slot from
171 <     * non-null to null.  Because we rely on CASes of references, we
172 <     * do not need tag bits on base or top.  They are simple ints as
173 <     * used in any circular array-based queue (see for example
174 <     * ArrayDeque).  Updates to the indices must still be ordered in a
175 <     * way that guarantees that top == base means the queue is empty,
176 <     * but otherwise may err on the side of possibly making the queue
177 <     * appear nonempty when a push, pop, or poll have not fully
178 <     * committed. Note that this means that the poll operation,
179 <     * considered individually, is not wait-free. One thief cannot
180 <     * successfully continue until another in-progress one (or, if
181 <     * previously empty, a push) completes.  However, in the
182 <     * aggregate, we ensure at least probabilistic non-blockingness.
183 <     * If an attempted steal fails, a thief always chooses a different
184 <     * random victim target to try next. So, in order for one thief to
185 <     * progress, it suffices for any in-progress poll or new push on
186 <     * any empty queue to complete. (This is why we normally use
187 <     * method pollAt and its variants that try once at the apparent
188 <     * base index, else consider alternative actions, rather than
189 <     * method poll.)
164 >     * See also "Correct and Efficient Work-Stealing for Weak Memory
165 >     * Models" by Le, Pop, Cohen, and Nardelli, PPoPP 2013
166 >     * (http://www.di.ens.fr/~zappa/readings/ppopp13.pdf) for an
167 >     * analysis of memory ordering (atomic, volatile etc) issues.  The
168 >     * main differences ultimately stem from GC requirements that we
169 >     * null out taken slots as soon as we can, to maintain as small a
170 >     * footprint as possible even in programs generating huge numbers
171 >     * of tasks. To accomplish this, we shift the CAS arbitrating pop
172 >     * vs poll (steal) from being on the indices ("base" and "top") to
173 >     * the slots themselves.  So, both a successful pop and poll
174 >     * mainly entail a CAS of a slot from non-null to null.  Because
175 >     * we rely on CASes of references, we do not need tag bits on base
176 >     * or top.  They are simple ints as used in any circular
177 >     * array-based queue (see for example ArrayDeque).  Updates to the
178 >     * indices must still be ordered in a way that guarantees that top
179 >     * == base means the queue is empty, but otherwise may err on the
180 >     * side of possibly making the queue appear nonempty when a push,
181 >     * pop, or poll have not fully committed. Note that this means
182 >     * that the poll operation, considered individually, is not
183 >     * wait-free. One thief cannot successfully continue until another
184 >     * in-progress one (or, if previously empty, a push) completes.
185 >     * However, in the aggregate, we ensure at least probabilistic
186 >     * non-blockingness.  If an attempted steal fails, a thief always
187 >     * chooses a different random victim target to try next. So, in
188 >     * order for one thief to progress, it suffices for any
189 >     * in-progress poll or new push on any empty queue to
190 >     * complete. (This is why we normally use method pollAt and its
191 >     * variants that try once at the apparent base index, else
192 >     * consider alternative actions, rather than method poll.)
193       *
194       * This approach also enables support of a user mode in which local
195       * task processing is in FIFO, not LIFO order, simply by using
# Line 196 | Line 208 | public class ForkJoinPool extends Abstra
208       * for work-stealing (this would contaminate lifo/fifo
209       * processing). Instead, we randomly associate submission queues
210       * with submitting threads, using a form of hashing.  The
211 <     * ThreadLocal Submitter class contains a value initially used as
212 <     * a hash code for choosing existing queues, but may be randomly
213 <     * repositioned upon contention with other submitters.  In
214 <     * essence, submitters act like workers except that they are
215 <     * restricted to executing local tasks that they submitted (or in
216 <     * the case of CountedCompleters, others with the same root task).
217 <     * However, because most shared/external queue operations are more
218 <     * expensive than internal, and because, at steady state, external
219 <     * submitters will compete for CPU with workers, ForkJoinTask.join
220 <     * and related methods disable them from repeatedly helping to
221 <     * process tasks if all workers are active.  Insertion of tasks in
222 <     * shared mode requires a lock (mainly to protect in the case of
211 >     * Submitter probe value serves as a hash code for
212 >     * choosing existing queues, and may be randomly repositioned upon
213 >     * contention with other submitters.  In essence, submitters act
214 >     * like workers except that they are restricted to executing local
215 >     * tasks that they submitted (or in the case of CountedCompleters,
216 >     * others with the same root task).  However, because most
217 >     * shared/external queue operations are more expensive than
218 >     * internal, and because, at steady state, external submitters
219 >     * will compete for CPU with workers, ForkJoinTask.join and
220 >     * related methods disable them from repeatedly helping to process
221 >     * tasks if all workers are active.  Insertion of tasks in shared
222 >     * mode requires a lock (mainly to protect in the case of
223       * resizing) but we use only a simple spinlock (using bits in
224       * field qlock), because submitters encountering a busy queue move
225       * on to try or create other queues -- they block only when
# Line 297 | Line 309 | public class ForkJoinPool extends Abstra
309       * has not yet entered the wait queue. We solve this by requiring
310       * a full sweep of all workers (via repeated calls to method
311       * scan()) both before and after a newly waiting worker is added
312 <     * to the wait queue. During a rescan, the worker might release
313 <     * some other queued worker rather than itself, which has the same
314 <     * net effect. Because enqueued workers may actually be rescanning
315 <     * rather than waiting, we set and clear the "parker" field of
316 <     * WorkQueues to reduce unnecessary calls to unpark.  (This
317 <     * requires a secondary recheck to avoid missed signals.)  Note
318 <     * the unusual conventions about Thread.interrupts surrounding
319 <     * parking and other blocking: Because interrupts are used solely
320 <     * to alert threads to check termination, which is checked anyway
321 <     * upon blocking, we clear status (using Thread.interrupted)
322 <     * before any call to park, so that park does not immediately
311 <     * return due to status being set via some other unrelated call to
312 <     * interrupt in user code.
312 >     * to the wait queue.  Because enqueued workers may actually be
313 >     * rescanning rather than waiting, we set and clear the "parker"
314 >     * field of WorkQueues to reduce unnecessary calls to unpark.
315 >     * (This requires a secondary recheck to avoid missed signals.)
316 >     * Note the unusual conventions about Thread.interrupts
317 >     * surrounding parking and other blocking: Because interrupts are
318 >     * used solely to alert threads to check termination, which is
319 >     * checked anyway upon blocking, we clear status (using
320 >     * Thread.interrupted) before any call to park, so that park does
321 >     * not immediately return due to status being set via some other
322 >     * unrelated call to interrupt in user code.
323       *
324       * Signalling.  We create or wake up workers only when there
325       * appears to be at least one task they might be able to find and
326 <     * execute. However, many other threads may notice the same task
327 <     * and each signal to wake up a thread that might take it. So in
328 <     * general, pools will be over-signalled.  When a submission is
329 <     * added or another worker adds a task to a queue that has fewer
330 <     * than two tasks, they signal waiting workers (or trigger
331 <     * creation of new ones if fewer than the given parallelism level
332 <     * -- signalWork), and may leave a hint to the unparked worker to
333 <     * help signal others upon wakeup).  These primary signals are
324 <     * buttressed by others (see method helpSignal) whenever other
325 <     * threads scan for work or do not have a task to process.  On
326 <     * most platforms, signalling (unpark) overhead time is noticeably
326 >     * execute.  When a submission is added or another worker adds a
327 >     * task to a queue that has fewer than two tasks, they signal
328 >     * waiting workers (or trigger creation of new ones if fewer than
329 >     * the given parallelism level -- signalWork).  These primary
330 >     * signals are buttressed by others whenever other threads remove
331 >     * a task from a queue and notice that there are other tasks there
332 >     * as well.  So in general, pools will be over-signalled. On most
333 >     * platforms, signalling (unpark) overhead time is noticeably
334       * long, and the time between signalling a thread and it actually
335       * making progress can be very noticeably long, so it is worth
336       * offloading these delays from critical paths as much as
337 <     * possible.
337 >     * possible. Additionally, workers spin-down gradually, by staying
338 >     * alive so long as they see the ctl state changing.  Similar
339 >     * stability-sensing techniques are also used before blocking in
340 >     * awaitJoin and helpComplete.
341       *
342       * Trimming workers. To release resources after periods of lack of
343       * use, a worker starting to wait when the pool is quiescent will
# Line 440 | Line 450 | public class ForkJoinPool extends Abstra
450       * Common Pool
451       * ===========
452       *
453 <     * The static common Pool always exists after static
453 >     * The static common pool always exists after static
454       * initialization.  Since it (or any other created pool) need
455       * never be used, we minimize initial construction overhead and
456       * footprint to the setup of about a dozen fields, with no nested
# Line 448 | Line 458 | public class ForkJoinPool extends Abstra
458       * fullExternalPush during the first submission to the pool.
459       *
460       * When external threads submit to the common pool, they can
461 <     * perform some subtask processing (see externalHelpJoin and
462 <     * related methods).  We do not need to record whether these
461 >     * perform subtask processing (see externalHelpJoin and related
462 >     * methods).  This caller-helps policy makes it sensible to set
463 >     * common pool parallelism level to one (or more) less than the
464 >     * total number of available cores, or even zero for pure
465 >     * caller-runs.  We do not need to record whether external
466       * submissions are to the common pool -- if not, externalHelpJoin
467       * returns quickly (at the most helping to signal some common pool
468       * workers). These submitters would otherwise be blocked waiting
# Line 519 | Line 532 | public class ForkJoinPool extends Abstra
532           *
533           * @param pool the pool this thread works in
534           * @throws NullPointerException if the pool is null
535 +         * @return the new worker thread
536           */
537          public ForkJoinWorkerThread newThread(ForkJoinPool pool);
538      }
# Line 535 | Line 549 | public class ForkJoinPool extends Abstra
549      }
550  
551      /**
538     * Per-thread records for threads that submit to pools. Currently
539     * holds only pseudo-random seed / index that is used to choose
540     * submission queues in method externalPush. In the future, this may
541     * also incorporate a means to implement different task rejection
542     * and resubmission policies.
543     *
544     * Seeds for submitters and workers/workQueues work in basically
545     * the same way but are initialized and updated using slightly
546     * different mechanics. Both are initialized using the same
547     * approach as in class ThreadLocal, where successive values are
548     * unlikely to collide with previous values. Seeds are then
549     * randomly modified upon collisions using xorshifts, which
550     * requires a non-zero seed.
551     */
552    static final class Submitter {
553        int seed;
554        Submitter(int s) { seed = s; }
555    }
556
557    /**
552       * Class for artificial tasks that are used to replace the target
553       * of local joins if they are removed from an interior queue slot
554       * in WorkQueue.tryRemoveAndExec. We don't need the proxy to
# Line 613 | Line 607 | public class ForkJoinPool extends Abstra
607       * do not want multiple WorkQueue instances or multiple queue
608       * arrays sharing cache lines. (It would be best for queue objects
609       * and their arrays to share, but there is nothing available to
610 <     * help arrange that).  Unfortunately, because they are recorded
611 <     * in a common array, WorkQueue instances are often moved to be
618 <     * adjacent by garbage collectors. To reduce impact, we use field
619 <     * padding that works OK on common platforms; this effectively
620 <     * trades off slightly slower average field access for the sake of
621 <     * avoiding really bad worst-case access. (Until better JVM
622 <     * support is in place, this padding is dependent on transient
623 <     * properties of JVM field layout rules.) We also take care in
624 <     * allocating, sizing and resizing the array. Non-shared queue
625 <     * arrays are initialized by workers before use. Others are
626 <     * allocated on first use.
610 >     * help arrange that). The @Contended annotation alerts JVMs to
611 >     * try to keep instances apart.
612       */
613      static final class WorkQueue {
614          /**
# Line 649 | Line 634 | public class ForkJoinPool extends Abstra
634          // Heuristic padding to ameliorate unfortunate memory placements
635          volatile long pad00, pad01, pad02, pad03, pad04, pad05, pad06;
636  
652        int seed;                  // for random scanning; initialize nonzero
637          volatile int eventCount;   // encoded inactivation count; < 0 if inactive
638          int nextWait;              // encoded record of next event waiter
655        int hint;                  // steal or signal hint (index)
656        int poolIndex;             // index of this queue in pool (or 0)
657        final int mode;            // 0: lifo, > 0: fifo, < 0: shared
639          int nsteals;               // number of steals
640 +        int hint;                  // steal index hint
641 +        short poolIndex;           // index of this queue in pool
642 +        final short mode;          // 0: lifo, > 0: fifo, < 0: shared
643          volatile int qlock;        // 1: locked, -1: terminate; else 0
644          volatile int base;         // index of next slot for poll
645          int top;                   // index of next slot for push
# Line 673 | Line 657 | public class ForkJoinPool extends Abstra
657                    int seed) {
658              this.pool = pool;
659              this.owner = owner;
660 <            this.mode = mode;
661 <            this.seed = seed;
660 >            this.mode = (short)mode;
661 >            this.hint = seed; // store initial seed for runWorker
662              // Place indices in the center of array (that is not yet allocated)
663              base = top = INITIAL_QUEUE_CAPACITY >>> 1;
664          }
# Line 687 | Line 671 | public class ForkJoinPool extends Abstra
671              return (n >= 0) ? 0 : -n; // ignore transient negative
672          }
673  
674 <       /**
674 >        /**
675           * Provides a more accurate estimate of whether this queue has
676           * any tasks than does queueSize, by checking whether a
677           * near-empty queue has at least one unclaimed task.
# Line 712 | Line 696 | public class ForkJoinPool extends Abstra
696           */
697          final void push(ForkJoinTask<?> task) {
698              ForkJoinTask<?>[] a; ForkJoinPool p;
699 <            int s = top, m, n;
699 >            int s = top, n;
700              if ((a = array) != null) {    // ignore if queue removed
701 <                int j = (((m = a.length - 1) & s) << ASHIFT) + ABASE;
702 <                U.putOrderedObject(a, j, task);
703 <                if ((n = (top = s + 1) - base) <= 2) {
704 <                    if ((p = pool) != null)
721 <                        p.signalWork(this);
722 <                }
701 >                int m = a.length - 1;
702 >                U.putOrderedObject(a, ((m & s) << ASHIFT) + ABASE, task);
703 >                if ((n = (top = s + 1) - base) <= 2)
704 >                    (p = pool).signalWork(p.workQueues, this);
705                  else if (n >= m)
706                      growArray();
707              }
708          }
709  
710 <       /**
710 >        /**
711           * Initializes or doubles the capacity of array. Call either
712           * by owner or with lock held -- it is OK for base, but not
713           * top, to move while resizings are in progress.
# Line 783 | Line 765 | public class ForkJoinPool extends Abstra
765              if ((a = array) != null) {
766                  int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
767                  if ((t = (ForkJoinTask<?>)U.getObjectVolatile(a, j)) != null &&
768 <                    base == b &&
769 <                    U.compareAndSwapObject(a, j, t, null)) {
788 <                    base = b + 1;
768 >                    base == b && U.compareAndSwapObject(a, j, t, null)) {
769 >                    U.putOrderedInt(this, QBASE, b + 1);
770                      return t;
771                  }
772              }
# Line 801 | Line 782 | public class ForkJoinPool extends Abstra
782                  int j = (((a.length - 1) & b) << ASHIFT) + ABASE;
783                  t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
784                  if (t != null) {
785 <                    if (base == b &&
786 <                        U.compareAndSwapObject(a, j, t, null)) {
806 <                        base = b + 1;
785 >                    if (U.compareAndSwapObject(a, j, t, null)) {
786 >                        U.putOrderedInt(this, QBASE, b + 1);
787                          return t;
788                      }
789                  }
# Line 860 | Line 840 | public class ForkJoinPool extends Abstra
840                  ForkJoinTask.cancelIgnoringExceptions(t);
841          }
842  
863        /**
864         * Computes next value for random probes.  Scans don't require
865         * a very high quality generator, but also not a crummy one.
866         * Marsaglia xor-shift is cheap and works well enough.  Note:
867         * This is manually inlined in its usages in ForkJoinPool to
868         * avoid writes inside busy scan loops.
869         */
870        final int nextSeed() {
871            int r = seed;
872            r ^= r << 13;
873            r ^= r >>> 17;
874            return seed = r ^= r << 5;
875        }
876
843          // Specialized execution methods
844  
845          /**
880         * Pops and runs tasks until empty.
881         */
882        private void popAndExecAll() {
883            // A bit faster than repeated pop calls
884            ForkJoinTask<?>[] a; int m, s; long j; ForkJoinTask<?> t;
885            while ((a = array) != null && (m = a.length - 1) >= 0 &&
886                   (s = top - 1) - base >= 0 &&
887                   (t = ((ForkJoinTask<?>)
888                         U.getObject(a, j = ((m & s) << ASHIFT) + ABASE)))
889                   != null) {
890                if (U.compareAndSwapObject(a, j, t, null)) {
891                    top = s;
892                    t.doExec();
893                }
894            }
895        }
896
897        /**
846           * Polls and runs tasks until empty.
847           */
848 <        private void pollAndExecAll() {
848 >        final void pollAndExecAll() {
849              for (ForkJoinTask<?> t; (t = poll()) != null;)
850                  t.doExec();
851          }
852  
853          /**
854 +         * Executes a top-level task and any local tasks remaining
855 +         * after execution.
856 +         */
857 +        final void runTask(ForkJoinTask<?> task) {
858 +            if ((currentSteal = task) != null) {
859 +                task.doExec();
860 +                ForkJoinTask<?>[] a = array;
861 +                int md = mode;
862 +                ++nsteals;
863 +                currentSteal = null;
864 +                if (md != 0)
865 +                    pollAndExecAll();
866 +                else if (a != null) {
867 +                    int s, m = a.length - 1;
868 +                    while ((s = top - 1) - base >= 0) {
869 +                        long i = ((m & s) << ASHIFT) + ABASE;
870 +                        ForkJoinTask<?> t = (ForkJoinTask<?>)U.getObject(a, i);
871 +                        if (t == null)
872 +                            break;
873 +                        if (U.compareAndSwapObject(a, i, t, null)) {
874 +                            top = s;
875 +                            t.doExec();
876 +                        }
877 +                    }
878 +                }
879 +            }
880 +        }
881 +      
882 +        /**
883           * If present, removes from queue and executes the given task,
884           * or any other cancelled task. Returns (true) on any CAS
885           * or consistency check failure so caller can retry.
# Line 910 | Line 887 | public class ForkJoinPool extends Abstra
887           * @return false if no progress can be made, else true
888           */
889          final boolean tryRemoveAndExec(ForkJoinTask<?> task) {
890 <            boolean stat = true, removed = false, empty = true;
890 >            boolean stat;
891              ForkJoinTask<?>[] a; int m, s, b, n;
892 <            if ((a = array) != null && (m = a.length - 1) >= 0 &&
892 >            if (task != null && (a = array) != null && (m = a.length - 1) >= 0 &&
893                  (n = (s = top) - (b = base)) > 0) {
894 +                boolean removed = false, empty = true;
895 +                stat = true;
896                  for (ForkJoinTask<?> t;;) {           // traverse from s to b
897 <                    int j = ((--s & m) << ASHIFT) + ABASE;
898 <                    t = (ForkJoinTask<?>)U.getObjectVolatile(a, j);
897 >                    long j = ((--s & m) << ASHIFT) + ABASE;
898 >                    t = (ForkJoinTask<?>)U.getObject(a, j);
899                      if (t == null)                    // inconsistent length
900                          break;
901                      else if (t == task) {
# Line 944 | Line 923 | public class ForkJoinPool extends Abstra
923                          break;
924                      }
925                  }
926 +                if (removed)
927 +                    task.doExec();
928              }
929 <            if (removed)
930 <                task.doExec();
929 >            else
930 >                stat = false;
931              return stat;
932          }
933  
934          /**
935 <         * Polls for and executes the given task or any other task in
936 <         * its CountedCompleter computation.
935 >         * Tries to poll for and execute the given task or any other
936 >         * task in its CountedCompleter computation.
937           */
938 <        final boolean pollAndExecCC(ForkJoinTask<?> root) {
939 <            ForkJoinTask<?>[] a; int b; Object o;
940 <            outer: while ((b = base) - top < 0 && (a = array) != null) {
938 >        final boolean pollAndExecCC(CountedCompleter<?> root) {
939 >            ForkJoinTask<?>[] a; int b; Object o; CountedCompleter<?> t, r;
940 >            if ((b = base) - top < 0 && (a = array) != null) {
941                  long j = (((a.length - 1) & b) << ASHIFT) + ABASE;
942 <                if ((o = U.getObject(a, j)) == null ||
943 <                    !(o instanceof CountedCompleter))
944 <                    break;
945 <                for (CountedCompleter<?> t = (CountedCompleter<?>)o, r = t;;) {
946 <                    if (r == root) {
947 <                        if (base == b &&
948 <                            U.compareAndSwapObject(a, j, t, null)) {
949 <                            base = b + 1;
950 <                            t.doExec();
942 >                if ((o = U.getObjectVolatile(a, j)) == null)
943 >                    return true; // retry
944 >                if (o instanceof CountedCompleter) {
945 >                    for (t = (CountedCompleter<?>)o, r = t;;) {
946 >                        if (r == root) {
947 >                            if (base == b &&
948 >                                U.compareAndSwapObject(a, j, t, null)) {
949 >                                U.putOrderedInt(this, QBASE, b + 1);
950 >                                t.doExec();
951 >                            }
952                              return true;
953                          }
954 <                        else
955 <                            break; // restart
954 >                        else if ((r = r.completer) == null)
955 >                            break; // not part of root computation
956                      }
975                    if ((r = r.completer) == null)
976                        break outer; // not part of root computation
957                  }
958              }
959              return false;
960          }
961  
962          /**
963 <         * Executes a top-level task and any local tasks remaining
964 <         * after execution.
963 >         * Tries to pop and execute the given task or any other task
964 >         * in its CountedCompleter computation.
965           */
966 <        final void runTask(ForkJoinTask<?> t) {
967 <            if (t != null) {
968 <                (currentSteal = t).doExec();
969 <                currentSteal = null;
970 <                ++nsteals;
971 <                if (base - top < 0) {       // process remaining local tasks
972 <                    if (mode == 0)
973 <                        popAndExecAll();
974 <                    else
975 <                        pollAndExecAll();
966 >        final boolean externalPopAndExecCC(CountedCompleter<?> root) {
967 >            ForkJoinTask<?>[] a; int s; Object o; CountedCompleter<?> t, r;
968 >            if (base - (s = top) < 0 && (a = array) != null) {
969 >                long j = (((a.length - 1) & (s - 1)) << ASHIFT) + ABASE;
970 >                if ((o = U.getObject(a, j)) instanceof CountedCompleter) {
971 >                    for (t = (CountedCompleter<?>)o, r = t;;) {
972 >                        if (r == root) {
973 >                            if (U.compareAndSwapInt(this, QLOCK, 0, 1)) {
974 >                                if (top == s && array == a &&
975 >                                    U.compareAndSwapObject(a, j, t, null)) {
976 >                                    top = s - 1;
977 >                                    qlock = 0;
978 >                                    t.doExec();
979 >                                }
980 >                                else
981 >                                    qlock = 0;
982 >                            }
983 >                            return true;
984 >                        }
985 >                        else if ((r = r.completer) == null)
986 >                            break;
987 >                    }
988                  }
989              }
990 +            return false;
991          }
992  
993          /**
994 <         * Executes a non-top-level (stolen) task.
994 >         * Internal version
995           */
996 <        final void runSubtask(ForkJoinTask<?> t) {
997 <            if (t != null) {
998 <                ForkJoinTask<?> ps = currentSteal;
999 <                (currentSteal = t).doExec();
1000 <                currentSteal = ps;
996 >        final boolean internalPopAndExecCC(CountedCompleter<?> root) {
997 >            ForkJoinTask<?>[] a; int s; Object o; CountedCompleter<?> t, r;
998 >            if (base - (s = top) < 0 && (a = array) != null) {
999 >                long j = (((a.length - 1) & (s - 1)) << ASHIFT) + ABASE;
1000 >                if ((o = U.getObject(a, j)) instanceof CountedCompleter) {
1001 >                    for (t = (CountedCompleter<?>)o, r = t;;) {
1002 >                        if (r == root) {
1003 >                            if (U.compareAndSwapObject(a, j, t, null)) {
1004 >                                top = s - 1;
1005 >                                t.doExec();
1006 >                            }
1007 >                            return true;
1008 >                        }
1009 >                        else if ((r = r.completer) == null)
1010 >                            break;
1011 >                    }
1012 >                }
1013              }
1014 +            return false;
1015          }
1016  
1017          /**
# Line 1022 | Line 1028 | public class ForkJoinPool extends Abstra
1028  
1029          // Unsafe mechanics
1030          private static final sun.misc.Unsafe U;
1031 +        private static final long QBASE;
1032          private static final long QLOCK;
1033          private static final int ABASE;
1034          private static final int ASHIFT;
# Line 1030 | Line 1037 | public class ForkJoinPool extends Abstra
1037                  U = getUnsafe();
1038                  Class<?> k = WorkQueue.class;
1039                  Class<?> ak = ForkJoinTask[].class;
1040 +                QBASE = U.objectFieldOffset
1041 +                    (k.getDeclaredField("base"));
1042                  QLOCK = U.objectFieldOffset
1043                      (k.getDeclaredField("qlock"));
1044                  ABASE = U.arrayBaseOffset(ak);
# Line 1046 | Line 1055 | public class ForkJoinPool extends Abstra
1055      // static fields (initialized in static initializer below)
1056  
1057      /**
1049     * Creates a new ForkJoinWorkerThread. This factory is used unless
1050     * overridden in ForkJoinPool constructors.
1051     */
1052    public static final ForkJoinWorkerThreadFactory
1053        defaultForkJoinWorkerThreadFactory;
1054
1055    /**
1058       * Per-thread submission bookkeeping. Shared across all pools
1059       * to reduce ThreadLocal pollution and because random motion
1060       * to avoid contention in one pool is likely to hold for others.
# Line 1062 | Line 1064 | public class ForkJoinPool extends Abstra
1064      static final ThreadLocal<Submitter> submitters;
1065  
1066      /**
1067 +     * Creates a new ForkJoinWorkerThread. This factory is used unless
1068 +     * overridden in ForkJoinPool constructors.
1069 +     */
1070 +    public static final ForkJoinWorkerThreadFactory
1071 +        defaultForkJoinWorkerThreadFactory;
1072 +
1073 +    /**
1074       * Permission required for callers of methods that may start or
1075       * kill threads.
1076       */
# Line 1076 | Line 1085 | public class ForkJoinPool extends Abstra
1085      static final ForkJoinPool common;
1086  
1087      /**
1088 <     * Common pool parallelism. Must equal common.parallelism.
1088 >     * Common pool parallelism. To allow simpler use and management
1089 >     * when common pool threads are disabled, we allow the underlying
1090 >     * common.parallelism field to be zero, but in that case still report
1091 >     * parallelism as 1 to reflect resulting caller-runs mechanics.
1092       */
1093      static final int commonParallelism;
1094  
# Line 1215 | Line 1227 | public class ForkJoinPool extends Abstra
1227      static final int FIFO_QUEUE          =  1;
1228      static final int SHARED_QUEUE        = -1;
1229  
1218    // bounds for #steps in scan loop -- must be power 2 minus 1
1219    private static final int MIN_SCAN    = 0x1ff;   // cover estimation slop
1220    private static final int MAX_SCAN    = 0x1ffff; // 4 * max workers
1221
1222    // Instance fields
1223
1224    /*
1225     * Field layout of this class tends to matter more than one would
1226     * like. Runtime layout order is only loosely related to
1227     * declaration order and may differ across JVMs, but the following
1228     * empirically works OK on current JVMs.
1229     */
1230
1230      // Heuristic padding to ameliorate unfortunate memory placements
1231      volatile long pad00, pad01, pad02, pad03, pad04, pad05, pad06;
1232  
1233 +    // Instance fields
1234      volatile long stealCount;                  // collects worker counts
1235      volatile long ctl;                         // main pool control
1236      volatile int plock;                        // shutdown status and seqLock
1237      volatile int indexSeed;                    // worker/submitter index seed
1238 <    final int config;                          // mode and parallelism level
1238 >    final short parallelism;                   // parallelism level
1239 >    final short mode;                          // LIFO/FIFO
1240      WorkQueue[] workQueues;                    // main registry
1241      final ForkJoinWorkerThreadFactory factory;
1242 <    final Thread.UncaughtExceptionHandler ueh; // per-worker UEH
1242 >    final UncaughtExceptionHandler ueh;        // per-worker UEH
1243      final String workerNamePrefix;             // to create worker name string
1244  
1245      volatile Object pad10, pad11, pad12, pad13, pad14, pad15, pad16, pad17;
# Line 1253 | Line 1254 | public class ForkJoinPool extends Abstra
1254       * a more conservative alternative to a pure spinlock.
1255       */
1256      private int acquirePlock() {
1257 <        int spins = PL_SPINS, r = 0, ps, nps;
1257 >        int spins = PL_SPINS, ps, nps;
1258          for (;;) {
1259              if (((ps = plock) & PL_LOCK) == 0 &&
1260                  U.compareAndSwapInt(this, PLOCK, ps, nps = ps + PL_LOCK))
1261                  return nps;
1261            else if (r == 0) { // randomize spins if possible
1262                Thread t = Thread.currentThread(); WorkQueue w; Submitter z;
1263                if ((t instanceof ForkJoinWorkerThread) &&
1264                    (w = ((ForkJoinWorkerThread)t).workQueue) != null)
1265                    r = w.seed;
1266                else if ((z = submitters.get()) != null)
1267                    r = z.seed;
1268                else
1269                    r = 1;
1270            }
1262              else if (spins >= 0) {
1263 <                r ^= r << 1; r ^= r >>> 3; r ^= r << 10; // xorshift
1273 <                if (r >= 0)
1263 >                if (ThreadLocalRandom.current().nextInt() >= 0)
1264                      --spins;
1265              }
1266              else if (U.compareAndSwapInt(this, PLOCK, ps, ps | PL_SIGNAL)) {
# Line 1306 | Line 1296 | public class ForkJoinPool extends Abstra
1296       * parallelism level exist. Adjusts counts etc on failure.
1297       */
1298      private void tryAddWorker() {
1299 <        long c; int u;
1299 >        long c; int u, e;
1300          while ((u = (int)((c = ctl) >>> 32)) < 0 &&
1301 <               (u & SHORT_SIGN) != 0 && (int)c == 0) {
1302 <            long nc = (long)(((u + UTC_UNIT) & UTC_MASK) |
1303 <                             ((u + UAC_UNIT) & UAC_MASK)) << 32;
1301 >               (u & SHORT_SIGN) != 0 && (e = (int)c) >= 0) {
1302 >            long nc = ((long)(((u + UTC_UNIT) & UTC_MASK) |
1303 >                              ((u + UAC_UNIT) & UAC_MASK)) << 32) | (long)e;
1304              if (U.compareAndSwapLong(this, CTL, c, nc)) {
1305                  ForkJoinWorkerThreadFactory fac;
1306                  Throwable ex = null;
# Line 1321 | Line 1311 | public class ForkJoinPool extends Abstra
1311                          wt.start();
1312                          break;
1313                      }
1314 <                } catch (Throwable e) {
1315 <                    ex = e;
1314 >                } catch (Throwable rex) {
1315 >                    ex = rex;
1316                  }
1317                  deregisterWorker(wt, ex);
1318                  break;
# Line 1343 | Line 1333 | public class ForkJoinPool extends Abstra
1333       * @return the worker's queue
1334       */
1335      final WorkQueue registerWorker(ForkJoinWorkerThread wt) {
1336 <        Thread.UncaughtExceptionHandler handler; WorkQueue[] ws; int s, ps;
1336 >        UncaughtExceptionHandler handler; WorkQueue[] ws; int s, ps;
1337          wt.setDaemon(true);
1338          if ((handler = ueh) != null)
1339              wt.setUncaughtExceptionHandler(handler);
1340          do {} while (!U.compareAndSwapInt(this, INDEXSEED, s = indexSeed,
1341                                            s += SEED_INCREMENT) ||
1342                       s == 0); // skip 0
1343 <        WorkQueue w = new WorkQueue(this, wt, config >>> 16, s);
1343 >        WorkQueue w = new WorkQueue(this, wt, mode, s);
1344          if (((ps = plock) & PL_LOCK) != 0 ||
1345              !U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1346              ps = acquirePlock();
# Line 1370 | Line 1360 | public class ForkJoinPool extends Abstra
1360                          }
1361                      }
1362                  }
1363 <                w.eventCount = w.poolIndex = r; // volatile write orders
1363 >                w.poolIndex = (short)r;
1364 >                w.eventCount = r; // volatile write orders
1365                  ws[r] = w;
1366              }
1367          } finally {
1368              if (!U.compareAndSwapInt(this, PLOCK, ps, nps))
1369                  releasePlock(nps);
1370          }
1371 <        wt.setName(workerNamePrefix.concat(Integer.toString(w.poolIndex)));
1371 >        wt.setName(workerNamePrefix.concat(Integer.toString(w.poolIndex >>> 1)));
1372          return w;
1373      }
1374  
# Line 1387 | Line 1378 | public class ForkJoinPool extends Abstra
1378       * array, and adjusts counts. If pool is shutting down, tries to
1379       * complete termination.
1380       *
1381 <     * @param wt the worker thread or null if construction failed
1381 >     * @param wt the worker thread, or null if construction failed
1382       * @param ex the exception causing failure, or null if none
1383       */
1384      final void deregisterWorker(ForkJoinWorkerThread wt, Throwable ex) {
1385          WorkQueue w = null;
1386          if (wt != null && (w = wt.workQueue) != null) {
1387 <            int ps;
1387 >            int ps; long sc;
1388              w.qlock = -1;                // ensure set
1389 <            long ns = w.nsteals, sc;     // collect steal count
1390 <            do {} while (!U.compareAndSwapLong(this, STEALCOUNT,
1400 <                                               sc = stealCount, sc + ns));
1389 >            do {} while(!U.compareAndSwapLong(this, STEALCOUNT, sc = stealCount,
1390 >                                              sc + w.nsteals));
1391              if (((ps = plock) & PL_LOCK) != 0 ||
1392                  !U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1393                  ps = acquirePlock();
# Line 1455 | Line 1445 | public class ForkJoinPool extends Abstra
1445      // Submissions
1446  
1447      /**
1448 +     * Per-thread records for threads that submit to pools. Currently
1449 +     * holds only pseudo-random seed / index that is used to choose
1450 +     * submission queues in method externalPush. In the future, this may
1451 +     * also incorporate a means to implement different task rejection
1452 +     * and resubmission policies.
1453 +     *
1454 +     * Seeds for submitters and workers/workQueues work in basically
1455 +     * the same way but are initialized and updated using slightly
1456 +     * different mechanics. Both are initialized using the same
1457 +     * approach as in class ThreadLocal, where successive values are
1458 +     * unlikely to collide with previous values. Seeds are then
1459 +     * randomly modified upon collisions using xorshifts, which
1460 +     * requires a non-zero seed.
1461 +     */
1462 +    static final class Submitter {
1463 +        int seed;
1464 +        Submitter(int s) { seed = s; }
1465 +    }
1466 +
1467 +    /**
1468       * Unless shutting down, adds the given task to a submission queue
1469       * at submitter's current queue index (modulo submission
1470       * range). Only the most common path is directly handled in this
# Line 1463 | Line 1473 | public class ForkJoinPool extends Abstra
1473       * @param task the task. Caller must ensure non-null.
1474       */
1475      final void externalPush(ForkJoinTask<?> task) {
1476 <        WorkQueue[] ws; WorkQueue q; Submitter z; int m; ForkJoinTask<?>[] a;
1477 <        if ((z = submitters.get()) != null && plock > 0 &&
1478 <            (ws = workQueues) != null && (m = (ws.length - 1)) >= 0 &&
1479 <            (q = ws[m & z.seed & SQMASK]) != null &&
1476 >        Submitter z = submitters.get();
1477 >        WorkQueue q; int r, m, s, n, am; ForkJoinTask<?>[] a;
1478 >        int ps = plock;
1479 >        WorkQueue[] ws = workQueues;
1480 >        if (z != null && ps > 0 && ws != null && (m = (ws.length - 1)) >= 0 &&
1481 >            (q = ws[m & (r = z.seed) & SQMASK]) != null && r != 0 &&
1482              U.compareAndSwapInt(q, QLOCK, 0, 1)) { // lock
1483 <            int b = q.base, s = q.top, n, an;
1484 <            if ((a = q.array) != null && (an = a.length) > (n = s + 1 - b)) {
1485 <                int j = (((an - 1) & s) << ASHIFT) + ABASE;
1483 >            if ((a = q.array) != null &&
1484 >                (am = a.length - 1) > (n = (s = q.top) - q.base)) {
1485 >                int j = ((am & s) << ASHIFT) + ABASE;
1486                  U.putOrderedObject(a, j, task);
1487                  q.top = s + 1;                     // push on to deque
1488                  q.qlock = 0;
1489 <                if (n <= 2)
1490 <                    signalWork(q);
1489 >                if (n <= 1)
1490 >                    signalWork(ws, q);
1491                  return;
1492              }
1493              q.qlock = 0;
# Line 1513 | Line 1525 | public class ForkJoinPool extends Abstra
1525                  r = z.seed;
1526                  r ^= r << 13;                   // same xorshift as WorkQueues
1527                  r ^= r >>> 17;
1528 <                z.seed = r ^ (r << 5);
1528 >                z.seed = r ^= (r << 5);
1529              }
1530 <            else if ((ps = plock) < 0)
1530 >            if ((ps = plock) < 0)
1531                  throw new RejectedExecutionException();
1532              else if (ps == 0 || (ws = workQueues) == null ||
1533                       (m = ws.length - 1) < 0) { // initialize workQueues
1534 <                int p = config & SMASK;         // find power of two table size
1534 >                int p = parallelism;            // find power of two table size
1535                  int n = (p > 1) ? p - 1 : 1;    // ensure at least 2 slots
1536                  n |= n >>> 1; n |= n >>> 2;  n |= n >>> 4;
1537                  n |= n >>> 8; n |= n >>> 16; n = (n + 1) << 1;
# Line 1551 | Line 1563 | public class ForkJoinPool extends Abstra
1563                          q.qlock = 0;  // unlock
1564                      }
1565                      if (submitted) {
1566 <                        signalWork(q);
1566 >                        signalWork(ws, q);
1567                          return;
1568                      }
1569                  }
# Line 1559 | Line 1571 | public class ForkJoinPool extends Abstra
1571              }
1572              else if (((ps = plock) & PL_LOCK) == 0) { // create new queue
1573                  q = new WorkQueue(this, null, SHARED_QUEUE, r);
1574 +                q.poolIndex = (short)k;
1575                  if (((ps = plock) & PL_LOCK) != 0 ||
1576                      !U.compareAndSwapInt(this, PLOCK, ps, ps += PL_LOCK))
1577                      ps = acquirePlock();
# Line 1569 | Line 1582 | public class ForkJoinPool extends Abstra
1582                      releasePlock(nps);
1583              }
1584              else
1585 <                r = 0; // try elsewhere while lock held
1585 >                r = 0;
1586          }
1587      }
1588  
# Line 1580 | Line 1593 | public class ForkJoinPool extends Abstra
1593       */
1594      final void incrementActiveCount() {
1595          long c;
1596 <        do {} while (!U.compareAndSwapLong(this, CTL, c = ctl, c + AC_UNIT));
1596 >        do {} while (!U.compareAndSwapLong
1597 >                     (this, CTL, c = ctl, ((c & ~AC_MASK) |
1598 >                                           ((c & AC_MASK) + AC_UNIT))));
1599      }
1600  
1601      /**
1602       * Tries to create or activate a worker if too few are active.
1603       *
1604 <     * @param q the (non-null) queue holding tasks to be signalled
1604 >     * @param ws the worker array to use to find signallees
1605 >     * @param q if non-null, the queue holding tasks to be processed
1606       */
1607 <    final void signalWork(WorkQueue q) {
1608 <        int hint = q.poolIndex;
1609 <        long c; int e, u, i, n; WorkQueue[] ws; WorkQueue w; Thread p;
1610 <        while ((u = (int)((c = ctl) >>> 32)) < 0) {
1611 <            if ((e = (int)c) > 0) {
1612 <                if ((ws = workQueues) != null && ws.length > (i = e & SMASK) &&
1597 <                    (w = ws[i]) != null && w.eventCount == (e | INT_SIGN)) {
1598 <                    long nc = (((long)(w.nextWait & E_MASK)) |
1599 <                               ((long)(u + UAC_UNIT) << 32));
1600 <                    if (U.compareAndSwapLong(this, CTL, c, nc)) {
1601 <                        w.hint = hint;
1602 <                        w.eventCount = (e + E_SEQ) & E_MASK;
1603 <                        if ((p = w.parker) != null)
1604 <                            U.unpark(p);
1605 <                        break;
1606 <                    }
1607 <                    if (q.top - q.base <= 0)
1608 <                        break;
1609 <                }
1610 <                else
1611 <                    break;
1612 <            }
1613 <            else {
1607 >    final void signalWork(WorkQueue[] ws, WorkQueue q) {
1608 >        for (;;) {
1609 >            long c; int e, u, i; WorkQueue w; Thread p;
1610 >            if ((u = (int)((c = ctl) >>> 32)) >= 0)
1611 >                break;
1612 >            if ((e = (int)c) <= 0) {
1613                  if ((short)u < 0)
1614                      tryAddWorker();
1615                  break;
1616              }
1617 +            if (ws == null || ws.length <= (i = e & SMASK) ||
1618 +                (w = ws[i]) == null)
1619 +                break;
1620 +            long nc = (((long)(w.nextWait & E_MASK)) |
1621 +                       ((long)(u + UAC_UNIT)) << 32);
1622 +            int ne = (e + E_SEQ) & E_MASK;
1623 +            if (w.eventCount == (e | INT_SIGN) &&
1624 +                U.compareAndSwapLong(this, CTL, c, nc)) {
1625 +                w.eventCount = ne;
1626 +                if ((p = w.parker) != null)
1627 +                    U.unpark(p);
1628 +                break;
1629 +            }
1630 +            if (q != null && q.base >= q.top)
1631 +                break;
1632          }
1633      }
1634  
# Line 1625 | Line 1639 | public class ForkJoinPool extends Abstra
1639       */
1640      final void runWorker(WorkQueue w) {
1641          w.growArray(); // allocate queue
1642 <        do { w.runTask(scan(w)); } while (w.qlock >= 0);
1642 >        for (int r = w.hint; scan(w, r) == 0; ) {
1643 >            r ^= r << 13; r ^= r >>> 17; r ^= r << 5; // xorshift
1644 >        }
1645      }
1646  
1647      /**
1648 <     * Scans for and, if found, returns one task, else possibly
1648 >     * Scans for and, if found, runs one task, else possibly
1649       * inactivates the worker. This method operates on single reads of
1650       * volatile state and is designed to be re-invoked continuously,
1651       * in part because it returns upon detecting inconsistencies,
1652       * contention, or state changes that indicate possible success on
1653       * re-invocation.
1654       *
1655 <     * The scan searches for tasks across queues (starting at a random
1656 <     * index, and relying on registerWorker to irregularly scatter
1657 <     * them within array to avoid bias), checking each at least twice.
1658 <     * The scan terminates upon either finding a non-empty queue, or
1659 <     * completing the sweep. If the worker is not inactivated, it
1660 <     * takes and returns a task from this queue. Otherwise, if not
1661 <     * activated, it signals workers (that may include itself) and
1662 <     * returns so caller can retry. Also returns for true if the
1663 <     * worker array may have changed during an empty scan.  On failure
1648 <     * to find a task, we take one of the following actions, after
1649 <     * which the caller will retry calling this method unless
1650 <     * terminated.
1651 <     *
1652 <     * * If pool is terminating, terminate the worker.
1653 <     *
1654 <     * * If not already enqueued, try to inactivate and enqueue the
1655 <     * worker on wait queue. Or, if inactivating has caused the pool
1656 <     * to be quiescent, relay to idleAwaitWork to possibly shrink
1657 <     * pool.
1658 <     *
1659 <     * * If already enqueued and none of the above apply, possibly
1660 <     * park awaiting signal, else lingering to help scan and signal.
1661 <     *
1662 <     * * If a non-empty queue discovered or left as a hint,
1663 <     * help wake up other workers before return.
1655 >     * The scan searches for tasks across queues starting at a random
1656 >     * index, checking each at least twice.  The scan terminates upon
1657 >     * either finding a non-empty queue, or completing the sweep. If
1658 >     * the worker is not inactivated, it takes and runs a task from
1659 >     * this queue. Otherwise, if not activated, it tries to activate
1660 >     * itself or some other worker by signalling. On failure to find a
1661 >     * task, returns (for retry) if pool state may have changed during
1662 >     * an empty scan, or tries to inactivate if active, else possibly
1663 >     * blocks or terminates via method awaitWork.
1664       *
1665       * @param w the worker (via its WorkQueue)
1666 <     * @return a task or null if none found
1666 >     * @param r a random seed
1667 >     * @return worker qlock status if would have waited, else 0
1668       */
1669 <    private final ForkJoinTask<?> scan(WorkQueue w) {
1669 >    private final int scan(WorkQueue w, int r) {
1670          WorkQueue[] ws; int m;
1671 <        int ps = plock;                          // read plock before ws
1672 <        if (w != null && (ws = workQueues) != null && (m = ws.length - 1) >= 0) {
1673 <            int ec = w.eventCount;               // ec is negative if inactive
1674 <            int r = w.seed; r ^= r << 13; r ^= r >>> 17; w.seed = r ^= r << 5;
1675 <            w.hint = -1;                         // update seed and clear hint
1676 <            int j = ((m + m + 1) | MIN_SCAN) & MAX_SCAN;
1677 <            do {
1678 <                WorkQueue q; ForkJoinTask<?>[] a; int b;
1679 <                if ((q = ws[(r + j) & m]) != null && (b = q.base) - q.top < 0 &&
1680 <                    (a = q.array) != null) {     // probably nonempty
1681 <                    int i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1682 <                    ForkJoinTask<?> t = (ForkJoinTask<?>)
1683 <                        U.getObjectVolatile(a, i);
1684 <                    if (q.base == b && ec >= 0 && t != null &&
1685 <                        U.compareAndSwapObject(a, i, t, null)) {
1686 <                        if ((q.base = b + 1) - q.top < 0)
1687 <                            signalWork(q);
1688 <                        return t;                // taken
1689 <                    }
1690 <                    else if ((ec < 0 || j < m) && (int)(ctl >> AC_SHIFT) <= 0) {
1691 <                        w.hint = (r + j) & m;    // help signal below
1692 <                        break;                   // cannot take
1693 <                    }
1694 <                }
1695 <            } while (--j >= 0);
1696 <
1697 <            int h, e, ns; long c, sc; WorkQueue q;
1697 <            if ((ns = w.nsteals) != 0) {
1698 <                if (U.compareAndSwapLong(this, STEALCOUNT,
1699 <                                         sc = stealCount, sc + ns))
1700 <                    w.nsteals = 0;               // collect steals and rescan
1701 <            }
1702 <            else if (plock != ps)                // consistency check
1703 <                ;                                // skip
1704 <            else if ((e = (int)(c = ctl)) < 0)
1705 <                w.qlock = -1;                    // pool is terminating
1706 <            else {
1707 <                if ((h = w.hint) < 0) {
1708 <                    if (ec >= 0) {               // try to enqueue/inactivate
1709 <                        long nc = (((long)ec |
1710 <                                    ((c - AC_UNIT) & (AC_MASK|TC_MASK))));
1711 <                        w.nextWait = e;          // link and mark inactive
1671 >        long c = ctl;                            // for consistency check
1672 >        if ((ws = workQueues) != null && (m = ws.length - 1) >= 0 && w != null) {
1673 >            for (int j = m + m + 1, ec = w.eventCount;;) {
1674 >                WorkQueue q; int b, e; ForkJoinTask<?>[] a; ForkJoinTask<?> t;
1675 >                if ((q = ws[(r - j) & m]) != null &&
1676 >                    (b = q.base) - q.top < 0 && (a = q.array) != null) {
1677 >                    long i = (((a.length - 1) & b) << ASHIFT) + ABASE;
1678 >                    if ((t = ((ForkJoinTask<?>)
1679 >                              U.getObjectVolatile(a, i))) != null) {
1680 >                        if (ec < 0)
1681 >                            helpRelease(c, ws, w, q, b);
1682 >                        else if (q.base == b &&
1683 >                                 U.compareAndSwapObject(a, i, t, null)) {
1684 >                            U.putOrderedInt(q, QBASE, b + 1);
1685 >                            if ((b + 1) - q.top < 0)
1686 >                                signalWork(ws, q);
1687 >                            w.runTask(t);
1688 >                        }
1689 >                    }
1690 >                    break;
1691 >                }
1692 >                else if (--j < 0) {
1693 >                    if ((ec | (e = (int)c)) < 0) // inactive or terminating
1694 >                        return awaitWork(w, c, ec);
1695 >                    else if (ctl == c) {         // try to inactivate and enqueue
1696 >                        long nc = (long)ec | ((c - AC_UNIT) & (AC_MASK|TC_MASK));
1697 >                        w.nextWait = e;
1698                          w.eventCount = ec | INT_SIGN;
1699 <                        if (ctl != c || !U.compareAndSwapLong(this, CTL, c, nc))
1700 <                            w.eventCount = ec;   // unmark on CAS failure
1715 <                        else if ((int)(c >> AC_SHIFT) == 1 - (config & SMASK))
1716 <                            idleAwaitWork(w, nc, c);
1717 <                    }
1718 <                    else if (w.eventCount < 0 && ctl == c) {
1719 <                        Thread wt = Thread.currentThread();
1720 <                        Thread.interrupted();    // clear status
1721 <                        U.putObject(wt, PARKBLOCKER, this);
1722 <                        w.parker = wt;           // emulate LockSupport.park
1723 <                        if (w.eventCount < 0)    // recheck
1724 <                            U.park(false, 0L);   // block
1725 <                        w.parker = null;
1726 <                        U.putObject(wt, PARKBLOCKER, null);
1727 <                    }
1728 <                }
1729 <                if ((h >= 0 || (h = w.hint) >= 0) &&
1730 <                    (ws = workQueues) != null && h < ws.length &&
1731 <                    (q = ws[h]) != null) {      // signal others before retry
1732 <                    WorkQueue v; Thread p; int u, i, s;
1733 <                    for (int n = (config & SMASK) - 1;;) {
1734 <                        int idleCount = (w.eventCount < 0) ? 0 : -1;
1735 <                        if (((s = idleCount - q.base + q.top) <= n &&
1736 <                             (n = s) <= 0) ||
1737 <                            (u = (int)((c = ctl) >>> 32)) >= 0 ||
1738 <                            (e = (int)c) <= 0 || m < (i = e & SMASK) ||
1739 <                            (v = ws[i]) == null)
1740 <                            break;
1741 <                        long nc = (((long)(v.nextWait & E_MASK)) |
1742 <                                   ((long)(u + UAC_UNIT) << 32));
1743 <                        if (v.eventCount != (e | INT_SIGN) ||
1744 <                            !U.compareAndSwapLong(this, CTL, c, nc))
1745 <                            break;
1746 <                        v.hint = h;
1747 <                        v.eventCount = (e + E_SEQ) & E_MASK;
1748 <                        if ((p = v.parker) != null)
1749 <                            U.unpark(p);
1750 <                        if (--n <= 0)
1751 <                            break;
1699 >                        if (!U.compareAndSwapLong(this, CTL, c, nc))
1700 >                            w.eventCount = ec;   // back out
1701                      }
1702 +                    break;
1703                  }
1704              }
1705          }
1706 <        return null;
1706 >        return 0;
1707      }
1708  
1709      /**
1710 <     * If inactivating worker w has caused the pool to become
1711 <     * quiescent, checks for pool termination, and, so long as this is
1712 <     * not the only worker, waits for event for up to a given
1713 <     * duration.  On timeout, if ctl has not changed, terminates the
1714 <     * worker, which will in turn wake up another worker to possibly
1715 <     * repeat this process.
1710 >     * A continuation of scan(), possibly blocking or terminating
1711 >     * worker w. Returns without blocking if pool state has apparently
1712 >     * changed since last invocation.  Also, if inactivating w has
1713 >     * caused the pool to become quiescent, checks for pool
1714 >     * termination, and, so long as this is not the only worker, waits
1715 >     * for event for up to a given duration.  On timeout, if ctl has
1716 >     * not changed, terminates the worker, which will in turn wake up
1717 >     * another worker to possibly repeat this process.
1718       *
1719       * @param w the calling worker
1720 <     * @param currentCtl the ctl value triggering possible quiescence
1721 <     * @param prevCtl the ctl value to restore if thread is terminated
1720 >     * @param c the ctl value on entry to scan
1721 >     * @param ec the worker's eventCount on entry to scan
1722       */
1723 <    private void idleAwaitWork(WorkQueue w, long currentCtl, long prevCtl) {
1724 <        if (w != null && w.eventCount < 0 &&
1725 <            !tryTerminate(false, false) && (int)prevCtl != 0 &&
1726 <            ctl == currentCtl) {
1727 <            int dc = -(short)(currentCtl >>> TC_SHIFT);
1728 <            long parkTime = dc < 0 ? FAST_IDLE_TIMEOUT: (dc + 1) * IDLE_TIMEOUT;
1729 <            long deadline = System.nanoTime() + parkTime - TIMEOUT_SLOP;
1730 <            Thread wt = Thread.currentThread();
1731 <            while (ctl == currentCtl) {
1732 <                Thread.interrupted();  // timed variant of version in scan()
1733 <                U.putObject(wt, PARKBLOCKER, this);
1734 <                w.parker = wt;
1735 <                if (ctl == currentCtl)
1736 <                    U.park(false, parkTime);
1737 <                w.parker = null;
1738 <                U.putObject(wt, PARKBLOCKER, null);
1739 <                if (ctl != currentCtl)
1740 <                    break;
1741 <                if (deadline - System.nanoTime() <= 0L &&
1742 <                    U.compareAndSwapLong(this, CTL, currentCtl, prevCtl)) {
1743 <                    w.eventCount = (w.eventCount + E_SEQ) | E_MASK;
1744 <                    w.hint = -1;
1745 <                    w.qlock = -1;   // shrink
1746 <                    break;
1723 >    private final int awaitWork(WorkQueue w, long c, int ec) {
1724 >        int stat, ns; long parkTime, deadline;
1725 >        if ((stat = w.qlock) >= 0 && w.eventCount == ec && ctl == c &&
1726 >            !Thread.interrupted()) {
1727 >            int e = (int)c;
1728 >            int u = (int)(c >>> 32);
1729 >            int d = (u >> UAC_SHIFT) + parallelism; // active count
1730 >
1731 >            if (e < 0 || (d <= 0 && tryTerminate(false, false)))
1732 >                stat = w.qlock = -1;          // pool is terminating
1733 >            else if ((ns = w.nsteals) != 0) { // collect steals and retry
1734 >                long sc;
1735 >                w.nsteals = 0;
1736 >                do {} while(!U.compareAndSwapLong(this, STEALCOUNT,
1737 >                                                  sc = stealCount, sc + ns));
1738 >            }
1739 >            else {
1740 >                long pc = ((d > 0 || ec != (e | INT_SIGN)) ? 0L :
1741 >                           ((long)(w.nextWait & E_MASK)) | // ctl to restore
1742 >                           ((long)(u + UAC_UNIT)) << 32);
1743 >                if (pc != 0L) {               // timed wait if last waiter
1744 >                    int dc = -(short)(c >>> TC_SHIFT);
1745 >                    parkTime = (dc < 0 ? FAST_IDLE_TIMEOUT:
1746 >                                (dc + 1) * IDLE_TIMEOUT);
1747 >                    deadline = System.nanoTime() + parkTime - TIMEOUT_SLOP;
1748 >                }
1749 >                else
1750 >                    parkTime = deadline = 0L;
1751 >                if (w.eventCount == ec && ctl == c) {
1752 >                    Thread wt = Thread.currentThread();
1753 >                    U.putObject(wt, PARKBLOCKER, this);
1754 >                    w.parker = wt;            // emulate LockSupport.park
1755 >                    if (w.eventCount == ec && ctl == c)
1756 >                        U.park(false, parkTime);  // must recheck before park
1757 >                    w.parker = null;
1758 >                    U.putObject(wt, PARKBLOCKER, null);
1759 >                    if (parkTime != 0L && ctl == c &&
1760 >                        deadline - System.nanoTime() <= 0L &&
1761 >                        U.compareAndSwapLong(this, CTL, c, pc))
1762 >                        stat = w.qlock = -1;  // shrink pool
1763                  }
1764              }
1765          }
1766 +        return stat;
1767      }
1768  
1769      /**
1770 <     * Scans through queues looking for work while joining a task; if
1771 <     * any present, signals. May return early if more signalling is
1772 <     * detectably unneeded.
1773 <     *
1774 <     * @param task return early if done
1775 <     * @param origin an index to start scan
1776 <     */
1777 <    private void helpSignal(ForkJoinTask<?> task, int origin) {
1778 <        WorkQueue[] ws; WorkQueue w; Thread p; long c; int m, u, e, i, s;
1779 <        if (task != null && task.status >= 0 &&
1780 <            (u = (int)(ctl >>> 32)) < 0 && (u >> UAC_SHIFT) < 0 &&
1781 <            (ws = workQueues) != null && (m = ws.length - 1) >= 0) {
1782 <            outer: for (int k = origin, j = m; j >= 0; --j) {
1783 <                WorkQueue q = ws[k++ & m];
1784 <                for (int n = m;;) { // limit to at most m signals
1785 <                    if (task.status < 0)
1786 <                        break outer;
1787 <                    if (q == null ||
1788 <                        ((s = -q.base + q.top) <= n && (n = s) <= 0))
1789 <                        break;
1821 <                    if ((u = (int)((c = ctl) >>> 32)) >= 0 ||
1822 <                        (e = (int)c) <= 0 || m < (i = e & SMASK) ||
1823 <                        (w = ws[i]) == null)
1824 <                        break outer;
1825 <                    long nc = (((long)(w.nextWait & E_MASK)) |
1826 <                               ((long)(u + UAC_UNIT) << 32));
1827 <                    if (w.eventCount != (e | INT_SIGN))
1828 <                        break outer;
1829 <                    if (U.compareAndSwapLong(this, CTL, c, nc)) {
1830 <                        w.eventCount = (e + E_SEQ) & E_MASK;
1831 <                        if ((p = w.parker) != null)
1832 <                            U.unpark(p);
1833 <                        if (--n <= 0)
1834 <                            break;
1835 <                    }
1836 <                }
1770 >     * Possibly releases (signals) a worker. Called only from scan()
1771 >     * when a worker with apparently inactive status finds a non-empty
1772 >     * queue. This requires revalidating all of the associated state
1773 >     * from caller.
1774 >     */
1775 >    private final void helpRelease(long c, WorkQueue[] ws, WorkQueue w,
1776 >                                   WorkQueue q, int b) {
1777 >        WorkQueue v; int e, i; Thread p;
1778 >        if (w != null && w.eventCount < 0 && (e = (int)c) > 0 &&
1779 >            ws != null && ws.length > (i = e & SMASK) &&
1780 >            (v = ws[i]) != null && ctl == c) {
1781 >            long nc = (((long)(v.nextWait & E_MASK)) |
1782 >                       ((long)((int)(c >>> 32) + UAC_UNIT)) << 32);
1783 >            int ne = (e + E_SEQ) & E_MASK;
1784 >            if (q != null && q.base == b && w.eventCount < 0 &&
1785 >                v.eventCount == (e | INT_SIGN) &&
1786 >                U.compareAndSwapLong(this, CTL, c, nc)) {
1787 >                v.eventCount = ne;
1788 >                if ((p = v.parker) != null)
1789 >                    U.unpark(p);
1790              }
1791          }
1792      }
# Line 1858 | Line 1811 | public class ForkJoinPool extends Abstra
1811       */
1812      private int tryHelpStealer(WorkQueue joiner, ForkJoinTask<?> task) {
1813          int stat = 0, steps = 0;                    // bound to avoid cycles
1814 <        if (joiner != null && task != null) {       // hoist null checks
1814 >        if (task != null && joiner != null &&
1815 >            joiner.base - joiner.top >= 0) {        // hoist checks
1816              restart: for (;;) {
1817                  ForkJoinTask<?> subtask = task;     // current target
1818                  for (WorkQueue j = joiner, v;;) {   // v is stealer of subtask
# Line 1885 | Line 1839 | public class ForkJoinPool extends Abstra
1839                          }
1840                      }
1841                      for (;;) { // help stealer or descend to its stealer
1842 <                        ForkJoinTask[] a;  int b;
1842 >                        ForkJoinTask[] a; int b;
1843                          if (subtask.status < 0)     // surround probes with
1844                              continue restart;       //   consistency checks
1845                          if ((b = v.base) - v.top < 0 && (a = v.array) != null) {
# Line 1896 | Line 1850 | public class ForkJoinPool extends Abstra
1850                                  v.currentSteal != subtask)
1851                                  continue restart;   // stale
1852                              stat = 1;               // apparent progress
1853 <                            if (t != null && v.base == b &&
1854 <                                U.compareAndSwapObject(a, i, t, null)) {
1855 <                                v.base = b + 1;     // help stealer
1856 <                                joiner.runSubtask(t);
1853 >                            if (v.base == b) {
1854 >                                if (t == null)
1855 >                                    break restart;
1856 >                                if (U.compareAndSwapObject(a, i, t, null)) {
1857 >                                    U.putOrderedInt(v, QBASE, b + 1);
1858 >                                    ForkJoinTask<?> ps = joiner.currentSteal;
1859 >                                    int jt = joiner.top;
1860 >                                    do {
1861 >                                        joiner.currentSteal = t;
1862 >                                        t.doExec(); // clear local tasks too
1863 >                                    } while (task.status >= 0 &&
1864 >                                             joiner.top != jt &&
1865 >                                             (t = joiner.pop()) != null);
1866 >                                    joiner.currentSteal = ps;
1867 >                                    break restart;
1868 >                                }
1869                              }
1904                            else if (v.base == b && ++steps == MAX_HELP)
1905                                break restart;      // v apparently stalled
1870                          }
1871                          else {                      // empty -- try to descend
1872                              ForkJoinTask<?> next = v.currentJoin;
# Line 1929 | Line 1893 | public class ForkJoinPool extends Abstra
1893       * and run tasks within the target's computation.
1894       *
1895       * @param task the task to join
1932     * @param mode if shared, exit upon completing any task
1933     * if all workers are active
1896       */
1897 <    private int helpComplete(ForkJoinTask<?> task, int mode) {
1898 <        WorkQueue[] ws; WorkQueue q; int m, n, s, u;
1899 <        if (task != null && (ws = workQueues) != null &&
1900 <            (m = ws.length - 1) >= 0) {
1901 <            for (int j = 1, origin = j;;) {
1897 >    private int helpComplete(WorkQueue joiner, CountedCompleter<?> task) {
1898 >        WorkQueue[] ws; int m;
1899 >        int s = 0;
1900 >        if ((ws = workQueues) != null && (m = ws.length - 1) >= 0 &&
1901 >            joiner != null && task != null) {
1902 >            int j = joiner.poolIndex;
1903 >            int scans = m + m + 1;
1904 >            long c = 0L;              // for stability check
1905 >            for (int k = scans; ; j += 2) {
1906 >                WorkQueue q;
1907                  if ((s = task.status) < 0)
1908 <                    return s;
1909 <                if ((q = ws[j & m]) != null && q.pollAndExecCC(task)) {
1910 <                    origin = j;
1911 <                    if (mode == SHARED_QUEUE &&
1912 <                        ((u = (int)(ctl >>> 32)) >= 0 || (u >> UAC_SHIFT) >= 0))
1908 >                    break;
1909 >                else if (joiner.internalPopAndExecCC(task))
1910 >                    k = scans;
1911 >                else if ((s = task.status) < 0)
1912 >                    break;
1913 >                else if ((q = ws[j & m]) != null && q.pollAndExecCC(task))
1914 >                    k = scans;
1915 >                else if (--k < 0) {
1916 >                    if (c == (c = ctl))
1917                          break;
1918 +                    k = scans;
1919                  }
1948                else if ((j = (j + 2) & m) == origin)
1949                    break;
1920              }
1921          }
1922 <        return 0;
1922 >        return s;
1923      }
1924  
1925      /**
# Line 1958 | Line 1928 | public class ForkJoinPool extends Abstra
1928       * for blocking. Fails on contention or termination. Otherwise,
1929       * adds a new thread if no idle workers are available and pool
1930       * may become starved.
1931 +     *
1932 +     * @param c the assumed ctl value
1933       */
1934 <    final boolean tryCompensate() {
1935 <        int pc = config & SMASK, e, i, tc; long c;
1936 <        WorkQueue[] ws; WorkQueue w; Thread p;
1937 <        if ((ws = workQueues) != null && (e = (int)(c = ctl)) >= 0) {
1938 <            if (e != 0 && (i = e & SMASK) < ws.length &&
1939 <                (w = ws[i]) != null && w.eventCount == (e | INT_SIGN)) {
1934 >    final boolean tryCompensate(long c) {
1935 >        WorkQueue[] ws = workQueues;
1936 >        int pc = parallelism, e = (int)c, m, tc;
1937 >        if (ws != null && (m = ws.length - 1) >= 0 && e >= 0 && ctl == c) {
1938 >            WorkQueue w = ws[e & m];
1939 >            if (e != 0 && w != null) {
1940 >                Thread p;
1941                  long nc = ((long)(w.nextWait & E_MASK) |
1942                             (c & (AC_MASK|TC_MASK)));
1943 <                if (U.compareAndSwapLong(this, CTL, c, nc)) {
1944 <                    w.eventCount = (e + E_SEQ) & E_MASK;
1943 >                int ne = (e + E_SEQ) & E_MASK;
1944 >                if (w.eventCount == (e | INT_SIGN) &&
1945 >                    U.compareAndSwapLong(this, CTL, c, nc)) {
1946 >                    w.eventCount = ne;
1947                      if ((p = w.parker) != null)
1948                          U.unpark(p);
1949                      return true;   // replace with idle worker
# Line 2011 | Line 1986 | public class ForkJoinPool extends Abstra
1986       */
1987      final int awaitJoin(WorkQueue joiner, ForkJoinTask<?> task) {
1988          int s = 0;
1989 <        if (joiner != null && task != null && (s = task.status) >= 0) {
1989 >        if (task != null && (s = task.status) >= 0 && joiner != null) {
1990              ForkJoinTask<?> prevJoin = joiner.currentJoin;
1991              joiner.currentJoin = task;
1992 <            do {} while ((s = task.status) >= 0 && !joiner.isEmpty() &&
1993 <                         joiner.tryRemoveAndExec(task)); // process local tasks
1994 <            if (s >= 0 && (s = task.status) >= 0) {
1995 <                helpSignal(task, joiner.poolIndex);
1996 <                if ((s = task.status) >= 0 &&
2022 <                    (task instanceof CountedCompleter))
2023 <                    s = helpComplete(task, LIFO_QUEUE);
2024 <            }
1992 >            do {} while (joiner.tryRemoveAndExec(task) && // process local tasks
1993 >                         (s = task.status) >= 0);
1994 >            if (s >= 0 && (task instanceof CountedCompleter))
1995 >                s = helpComplete(joiner, (CountedCompleter<?>)task);
1996 >            long cc = 0;        // for stability checks
1997              while (s >= 0 && (s = task.status) >= 0) {
1998 <                if ((!joiner.isEmpty() ||           // try helping
2027 <                     (s = tryHelpStealer(joiner, task)) == 0) &&
1998 >                if ((s = tryHelpStealer(joiner, task)) == 0 &&
1999                      (s = task.status) >= 0) {
2000 <                    helpSignal(task, joiner.poolIndex);
2001 <                    if ((s = task.status) >= 0 && tryCompensate()) {
2000 >                    if (!tryCompensate(cc))
2001 >                        cc = ctl;
2002 >                    else {
2003                          if (task.trySetSignal() && (s = task.status) >= 0) {
2004                              synchronized (task) {
2005                                  if (task.status >= 0) {
# Line 2040 | Line 2012 | public class ForkJoinPool extends Abstra
2012                                      task.notifyAll();
2013                              }
2014                          }
2015 <                        long c;                          // re-activate
2015 >                        long c; // reactivate
2016                          do {} while (!U.compareAndSwapLong
2017 <                                     (this, CTL, c = ctl, c + AC_UNIT));
2017 >                                     (this, CTL, c = ctl,
2018 >                                      ((c & ~AC_MASK) |
2019 >                                       ((c & AC_MASK) + AC_UNIT))));
2020                      }
2021                  }
2022              }
# Line 2064 | Line 2038 | public class ForkJoinPool extends Abstra
2038          if (joiner != null && task != null && (s = task.status) >= 0) {
2039              ForkJoinTask<?> prevJoin = joiner.currentJoin;
2040              joiner.currentJoin = task;
2041 <            do {} while ((s = task.status) >= 0 && !joiner.isEmpty() &&
2042 <                         joiner.tryRemoveAndExec(task));
2043 <            if (s >= 0 && (s = task.status) >= 0) {
2044 <                helpSignal(task, joiner.poolIndex);
2045 <                if ((s = task.status) >= 0 &&
2072 <                    (task instanceof CountedCompleter))
2073 <                    s = helpComplete(task, LIFO_QUEUE);
2074 <            }
2075 <            if (s >= 0 && joiner.isEmpty()) {
2041 >            do {} while (joiner.tryRemoveAndExec(task) && // process local tasks
2042 >                         (s = task.status) >= 0);
2043 >            if (s >= 0) {
2044 >                if (task instanceof CountedCompleter)
2045 >                    helpComplete(joiner, (CountedCompleter<?>)task);
2046                  do {} while (task.status >= 0 &&
2047                               tryHelpStealer(joiner, task) > 0);
2048              }
# Line 2084 | Line 2054 | public class ForkJoinPool extends Abstra
2054       * Returns a (probably) non-empty steal queue, if one is found
2055       * during a scan, else null.  This method must be retried by
2056       * caller if, by the time it tries to use the queue, it is empty.
2087     * @param r a (random) seed for scanning
2057       */
2058 <    private WorkQueue findNonEmptyStealQueue(int r) {
2058 >    private WorkQueue findNonEmptyStealQueue() {
2059 >        int r = ThreadLocalRandom.current().nextInt();
2060          for (;;) {
2061              int ps = plock, m; WorkQueue[] ws; WorkQueue q;
2062              if ((ws = workQueues) != null && (m = ws.length - 1) >= 0) {
2063                  for (int j = (m + 1) << 2; j >= 0; --j) {
2064 <                    if ((q = ws[(((r + j) << 1) | 1) & m]) != null &&
2064 >                    if ((q = ws[(((r - j) << 1) | 1) & m]) != null &&
2065                          q.base - q.top < 0)
2066                          return q;
2067                  }
# Line 2108 | Line 2078 | public class ForkJoinPool extends Abstra
2078       * find tasks either.
2079       */
2080      final void helpQuiescePool(WorkQueue w) {
2081 +        ForkJoinTask<?> ps = w.currentSteal;
2082          for (boolean active = true;;) {
2083              long c; WorkQueue q; ForkJoinTask<?> t; int b;
2084 <            while ((t = w.nextLocalTask()) != null) {
2114 <                if (w.base - w.top < 0)
2115 <                    signalWork(w);
2084 >            while ((t = w.nextLocalTask()) != null)
2085                  t.doExec();
2086 <            }
2118 <            if ((q = findNonEmptyStealQueue(w.nextSeed())) != null) {
2086 >            if ((q = findNonEmptyStealQueue()) != null) {
2087                  if (!active) {      // re-establish active count
2088                      active = true;
2089                      do {} while (!U.compareAndSwapLong
2090 <                                 (this, CTL, c = ctl, c + AC_UNIT));
2090 >                                 (this, CTL, c = ctl,
2091 >                                  ((c & ~AC_MASK) |
2092 >                                   ((c & AC_MASK) + AC_UNIT))));
2093                  }
2094                  if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null) {
2095 <                    if (q.base - q.top < 0)
2096 <                        signalWork(q);
2127 <                    w.runSubtask(t);
2095 >                    (w.currentSteal = t).doExec();
2096 >                    w.currentSteal = ps;
2097                  }
2098              }
2099              else if (active) {       // decrement active count without queuing
2100 <                long nc = (c = ctl) - AC_UNIT;
2101 <                if ((int)(nc >> AC_SHIFT) + (config & SMASK) == 0)
2102 <                    return;          // bypass decrement-then-increment
2100 >                long nc = ((c = ctl) & ~AC_MASK) | ((c & AC_MASK) - AC_UNIT);
2101 >                if ((int)(nc >> AC_SHIFT) + parallelism == 0)
2102 >                    break;          // bypass decrement-then-increment
2103                  if (U.compareAndSwapLong(this, CTL, c, nc))
2104                      active = false;
2105              }
2106 <            else if ((int)((c = ctl) >> AC_SHIFT) + (config & SMASK) == 0 &&
2107 <                     U.compareAndSwapLong(this, CTL, c, c + AC_UNIT))
2108 <                return;
2106 >            else if ((int)((c = ctl) >> AC_SHIFT) + parallelism <= 0 &&
2107 >                     U.compareAndSwapLong
2108 >                     (this, CTL, c, ((c & ~AC_MASK) |
2109 >                                     ((c & AC_MASK) + AC_UNIT))))
2110 >                break;
2111          }
2112      }
2113  
# Line 2150 | Line 2121 | public class ForkJoinPool extends Abstra
2121              WorkQueue q; int b;
2122              if ((t = w.nextLocalTask()) != null)
2123                  return t;
2124 <            if ((q = findNonEmptyStealQueue(w.nextSeed())) == null)
2124 >            if ((q = findNonEmptyStealQueue()) == null)
2125                  return null;
2126 <            if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null) {
2156 <                if (q.base - q.top < 0)
2157 <                    signalWork(q);
2126 >            if ((b = q.base) - q.top < 0 && (t = q.pollAt(b)) != null)
2127                  return t;
2159            }
2128          }
2129      }
2130  
# Line 2209 | Line 2177 | public class ForkJoinPool extends Abstra
2177      static int getSurplusQueuedTaskCount() {
2178          Thread t; ForkJoinWorkerThread wt; ForkJoinPool pool; WorkQueue q;
2179          if (((t = Thread.currentThread()) instanceof ForkJoinWorkerThread)) {
2180 <            int p = (pool = (wt = (ForkJoinWorkerThread)t).pool).config & SMASK;
2180 >            int p = (pool = (wt = (ForkJoinWorkerThread)t).pool).parallelism;
2181              int n = (q = wt.workQueue).top - q.base;
2182              int a = (int)(pool.ctl >> AC_SHIFT) + p;
2183              return n - (a > (p >>>= 1) ? 0 :
# Line 2239 | Line 2207 | public class ForkJoinPool extends Abstra
2207       */
2208      private boolean tryTerminate(boolean now, boolean enable) {
2209          int ps;
2210 <        if (this == common)                    // cannot shut down
2210 >        if (this == common)                        // cannot shut down
2211              return false;
2212          if ((ps = plock) >= 0) {                   // enable by setting plock
2213              if (!enable)
# Line 2253 | Line 2221 | public class ForkJoinPool extends Abstra
2221          }
2222          for (long c;;) {
2223              if (((c = ctl) & STOP_BIT) != 0) {     // already terminating
2224 <                if ((short)(c >>> TC_SHIFT) == -(config & SMASK)) {
2224 >                if ((short)(c >>> TC_SHIFT) + parallelism <= 0) {
2225                      synchronized (this) {
2226                          notifyAll();               // signal when 0 workers
2227                      }
# Line 2262 | Line 2230 | public class ForkJoinPool extends Abstra
2230              }
2231              if (!now) {                            // check if idle & no tasks
2232                  WorkQueue[] ws; WorkQueue w;
2233 <                if ((int)(c >> AC_SHIFT) != -(config & SMASK))
2233 >                if ((int)(c >> AC_SHIFT) + parallelism > 0)
2234                      return false;
2235                  if ((ws = workQueues) != null) {
2236                      for (int i = 0; i < ws.length; ++i) {
2237 <                        if ((w = ws[i]) != null) {
2238 <                            if (!w.isEmpty()) {    // signal unprocessed tasks
2239 <                                signalWork(w);
2240 <                                return false;
2241 <                            }
2274 <                            if ((i & 1) != 0 && w.eventCount >= 0)
2275 <                                return false;      // unqueued inactive worker
2237 >                        if ((w = ws[i]) != null &&
2238 >                            (!w.isEmpty() ||
2239 >                             ((i & 1) != 0 && w.eventCount >= 0))) {
2240 >                            signalWork(ws, w);
2241 >                            return false;
2242                          }
2243                      }
2244                  }
# Line 2328 | Line 2294 | public class ForkJoinPool extends Abstra
2294       * least one task.
2295       */
2296      static WorkQueue commonSubmitterQueue() {
2297 <        ForkJoinPool p; WorkQueue[] ws; int m; Submitter z;
2297 >        Submitter z; ForkJoinPool p; WorkQueue[] ws; int m, r;
2298          return ((z = submitters.get()) != null &&
2299                  (p = common) != null &&
2300                  (ws = p.workQueues) != null &&
# Line 2339 | Line 2305 | public class ForkJoinPool extends Abstra
2305      /**
2306       * Tries to pop the given task from submitter's queue in common pool.
2307       */
2308 <    static boolean tryExternalUnpush(ForkJoinTask<?> t) {
2309 <        ForkJoinPool p; WorkQueue[] ws; WorkQueue q; Submitter z;
2310 <        ForkJoinTask<?>[] a;  int m, s;
2311 <        if (t != null &&
2312 <            (z = submitters.get()) != null &&
2313 <            (p = common) != null &&
2314 <            (ws = p.workQueues) != null &&
2315 <            (m = ws.length - 1) >= 0 &&
2316 <            (q = ws[m & z.seed & SQMASK]) != null &&
2351 <            (s = q.top) != q.base &&
2352 <            (a = q.array) != null) {
2308 >    final boolean tryExternalUnpush(ForkJoinTask<?> task) {
2309 >        WorkQueue joiner; ForkJoinTask<?>[] a; int m, s;
2310 >        Submitter z = submitters.get();
2311 >        WorkQueue[] ws = workQueues;
2312 >        boolean popped = false;
2313 >        if (z != null && ws != null && (m = ws.length - 1) >= 0 &&
2314 >            (joiner = ws[z.seed & m & SQMASK]) != null &&
2315 >            joiner.base != (s = joiner.top) &&
2316 >            (a = joiner.array) != null) {
2317              long j = (((a.length - 1) & (s - 1)) << ASHIFT) + ABASE;
2318 <            if (U.getObject(a, j) == t &&
2319 <                U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2320 <                if (q.array == a && q.top == s && // recheck
2321 <                    U.compareAndSwapObject(a, j, t, null)) {
2322 <                    q.top = s - 1;
2323 <                    q.qlock = 0;
2360 <                    return true;
2318 >            if (U.getObject(a, j) == task &&
2319 >                U.compareAndSwapInt(joiner, QLOCK, 0, 1)) {
2320 >                if (joiner.top == s && joiner.array == a &&
2321 >                    U.compareAndSwapObject(a, j, task, null)) {
2322 >                    joiner.top = s - 1;
2323 >                    popped = true;
2324                  }
2325 <                q.qlock = 0;
2325 >                joiner.qlock = 0;
2326              }
2327          }
2328 <        return false;
2328 >        return popped;
2329      }
2330  
2331 <    /**
2332 <     * Tries to pop and run local tasks within the same computation
2333 <     * as the given root. On failure, tries to help complete from
2334 <     * other queues via helpComplete.
2335 <     */
2336 <    private void externalHelpComplete(WorkQueue q, ForkJoinTask<?> root) {
2337 <        ForkJoinTask<?>[] a; int m;
2338 <        if (q != null && (a = q.array) != null && (m = (a.length - 1)) >= 0 &&
2339 <            root != null && root.status >= 0) {
2340 <            for (;;) {
2341 <                int s, u; Object o; CountedCompleter<?> task = null;
2342 <                if ((s = q.top) - q.base > 0) {
2343 <                    long j = ((m & (s - 1)) << ASHIFT) + ABASE;
2381 <                    if ((o = U.getObject(a, j)) != null &&
2382 <                        (o instanceof CountedCompleter)) {
2383 <                        CountedCompleter<?> t = (CountedCompleter<?>)o, r = t;
2384 <                        do {
2385 <                            if (r == root) {
2386 <                                if (U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2387 <                                    if (q.array == a && q.top == s &&
2388 <                                        U.compareAndSwapObject(a, j, t, null)) {
2389 <                                        q.top = s - 1;
2390 <                                        task = t;
2391 <                                    }
2392 <                                    q.qlock = 0;
2393 <                                }
2394 <                                break;
2395 <                            }
2396 <                        } while ((r = r.completer) != null);
2397 <                    }
2398 <                }
2399 <                if (task != null)
2400 <                    task.doExec();
2401 <                if (root.status < 0 ||
2402 <                    (u = (int)(ctl >>> 32)) >= 0 || (u >> UAC_SHIFT) >= 0)
2331 >    final int externalHelpComplete(CountedCompleter<?> task) {
2332 >        WorkQueue joiner; int m, j;
2333 >        Submitter z = submitters.get();
2334 >        WorkQueue[] ws = workQueues;
2335 >        int s = 0;
2336 >        if (z != null && ws != null && (m = ws.length - 1) >= 0 &&
2337 >            (joiner = ws[(j = z.seed) & m & SQMASK]) != null && task != null) {
2338 >            int scans = m + m + 1;
2339 >            long c = 0L;             // for stability check
2340 >            j |= 1;                  // poll odd queues
2341 >            for (int k = scans; ; j += 2) {
2342 >                WorkQueue q;
2343 >                if ((s = task.status) < 0)
2344                      break;
2345 <                if (task == null) {
2346 <                    helpSignal(root, q.poolIndex);
2347 <                    if (root.status >= 0)
2407 <                        helpComplete(root, SHARED_QUEUE);
2345 >                else if (joiner.externalPopAndExecCC(task))
2346 >                    k = scans;
2347 >                else if ((s = task.status) < 0)
2348                      break;
2349 +                else if ((q = ws[j & m]) != null && q.pollAndExecCC(task))
2350 +                    k = scans;
2351 +                else if (--k < 0) {
2352 +                    if (c == (c = ctl))
2353 +                        break;
2354 +                    k = scans;
2355                  }
2356              }
2357          }
2358 <    }
2413 <
2414 <    /**
2415 <     * Tries to help execute or signal availability of the given task
2416 <     * from submitter's queue in common pool.
2417 <     */
2418 <    static void externalHelpJoin(ForkJoinTask<?> t) {
2419 <        // Some hard-to-avoid overlap with tryExternalUnpush
2420 <        ForkJoinPool p; WorkQueue[] ws; WorkQueue q, w; Submitter z;
2421 <        ForkJoinTask<?>[] a;  int m, s, n;
2422 <        if (t != null &&
2423 <            (z = submitters.get()) != null &&
2424 <            (p = common) != null &&
2425 <            (ws = p.workQueues) != null &&
2426 <            (m = ws.length - 1) >= 0 &&
2427 <            (q = ws[m & z.seed & SQMASK]) != null &&
2428 <            (a = q.array) != null) {
2429 <            int am = a.length - 1;
2430 <            if ((s = q.top) != q.base) {
2431 <                long j = ((am & (s - 1)) << ASHIFT) + ABASE;
2432 <                if (U.getObject(a, j) == t &&
2433 <                    U.compareAndSwapInt(q, QLOCK, 0, 1)) {
2434 <                    if (q.array == a && q.top == s &&
2435 <                        U.compareAndSwapObject(a, j, t, null)) {
2436 <                        q.top = s - 1;
2437 <                        q.qlock = 0;
2438 <                        t.doExec();
2439 <                    }
2440 <                    else
2441 <                        q.qlock = 0;
2442 <                }
2443 <            }
2444 <            if (t.status >= 0) {
2445 <                if (t instanceof CountedCompleter)
2446 <                    p.externalHelpComplete(q, t);
2447 <                else
2448 <                    p.helpSignal(t, q.poolIndex);
2449 <            }
2450 <        }
2358 >        return s;
2359      }
2360  
2361      // Exported methods
# Line 2514 | Line 2422 | public class ForkJoinPool extends Abstra
2422       */
2423      public ForkJoinPool(int parallelism,
2424                          ForkJoinWorkerThreadFactory factory,
2425 <                        Thread.UncaughtExceptionHandler handler,
2425 >                        UncaughtExceptionHandler handler,
2426                          boolean asyncMode) {
2427 +        this(checkParallelism(parallelism),
2428 +             checkFactory(factory),
2429 +             handler,
2430 +             (asyncMode ? FIFO_QUEUE : LIFO_QUEUE),
2431 +             "ForkJoinPool-" + nextPoolId() + "-worker-");
2432          checkPermission();
2433 <        if (factory == null)
2434 <            throw new NullPointerException();
2433 >    }
2434 >
2435 >    private static int checkParallelism(int parallelism) {
2436          if (parallelism <= 0 || parallelism > MAX_CAP)
2437              throw new IllegalArgumentException();
2438 <        this.factory = factory;
2439 <        this.ueh = handler;
2440 <        this.config = parallelism | (asyncMode ? (FIFO_QUEUE << 16) : 0);
2441 <        long np = (long)(-parallelism); // offset ctl counts
2442 <        this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2443 <        int pn = nextPoolId();
2444 <        StringBuilder sb = new StringBuilder("ForkJoinPool-");
2445 <        sb.append(Integer.toString(pn));
2532 <        sb.append("-worker-");
2533 <        this.workerNamePrefix = sb.toString();
2438 >        return parallelism;
2439 >    }
2440 >
2441 >    private static ForkJoinWorkerThreadFactory checkFactory
2442 >        (ForkJoinWorkerThreadFactory factory) {
2443 >        if (factory == null)
2444 >            throw new NullPointerException();
2445 >        return factory;
2446      }
2447  
2448      /**
2449 <     * Constructor for common pool, suitable only for static initialization.
2450 <     * Basically the same as above, but uses smallest possible initial footprint.
2451 <     */
2452 <    ForkJoinPool(int parallelism, long ctl,
2453 <                 ForkJoinWorkerThreadFactory factory,
2454 <                 Thread.UncaughtExceptionHandler handler) {
2455 <        this.config = parallelism;
2456 <        this.ctl = ctl;
2449 >     * Creates a {@code ForkJoinPool} with the given parameters, without
2450 >     * any security checks or parameter validation.  Invoked directly by
2451 >     * makeCommonPool.
2452 >     */
2453 >    private ForkJoinPool(int parallelism,
2454 >                         ForkJoinWorkerThreadFactory factory,
2455 >                         UncaughtExceptionHandler handler,
2456 >                         int mode,
2457 >                         String workerNamePrefix) {
2458 >        this.workerNamePrefix = workerNamePrefix;
2459          this.factory = factory;
2460          this.ueh = handler;
2461 <        this.workerNamePrefix = "ForkJoinPool.commonPool-worker-";
2461 >        this.mode = (short)mode;
2462 >        this.parallelism = (short)parallelism;
2463 >        long np = (long)(-parallelism); // offset ctl counts
2464 >        this.ctl = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
2465      }
2466  
2467      /**
# Line 2554 | Line 2471 | public class ForkJoinPool extends Abstra
2471       * ongoing processing are automatically terminated upon program
2472       * {@link System#exit}.  Any program that relies on asynchronous
2473       * task processing to complete before program termination should
2474 <     * invoke {@code commonPool().}{@link #awaitQuiescence}, before
2475 <     * exit.
2474 >     * invoke {@code commonPool().}{@link #awaitQuiescence awaitQuiescence},
2475 >     * before exit.
2476       *
2477       * @return the common pool instance
2478       * @since 1.8
# Line 2618 | Line 2535 | public class ForkJoinPool extends Abstra
2535          if (task instanceof ForkJoinTask<?>) // avoid re-wrap
2536              job = (ForkJoinTask<?>) task;
2537          else
2538 <            job = new ForkJoinTask.AdaptedRunnableAction(task);
2538 >            job = new ForkJoinTask.RunnableExecuteAction(task);
2539          externalPush(job);
2540      }
2541  
# Line 2720 | Line 2637 | public class ForkJoinPool extends Abstra
2637       *
2638       * @return the handler, or {@code null} if none
2639       */
2640 <    public Thread.UncaughtExceptionHandler getUncaughtExceptionHandler() {
2640 >    public UncaughtExceptionHandler getUncaughtExceptionHandler() {
2641          return ueh;
2642      }
2643  
# Line 2730 | Line 2647 | public class ForkJoinPool extends Abstra
2647       * @return the targeted parallelism level of this pool
2648       */
2649      public int getParallelism() {
2650 <        return config & SMASK;
2650 >        int par;
2651 >        return ((par = parallelism) > 0) ? par : 1;
2652      }
2653  
2654      /**
# Line 2752 | Line 2670 | public class ForkJoinPool extends Abstra
2670       * @return the number of worker threads
2671       */
2672      public int getPoolSize() {
2673 <        return (config & SMASK) + (short)(ctl >>> TC_SHIFT);
2673 >        return parallelism + (short)(ctl >>> TC_SHIFT);
2674      }
2675  
2676      /**
# Line 2762 | Line 2680 | public class ForkJoinPool extends Abstra
2680       * @return {@code true} if this pool uses async mode
2681       */
2682      public boolean getAsyncMode() {
2683 <        return (config >>> 16) == FIFO_QUEUE;
2683 >        return mode == FIFO_QUEUE;
2684      }
2685  
2686      /**
# Line 2793 | Line 2711 | public class ForkJoinPool extends Abstra
2711       * @return the number of active threads
2712       */
2713      public int getActiveThreadCount() {
2714 <        int r = (config & SMASK) + (int)(ctl >> AC_SHIFT);
2714 >        int r = parallelism + (int)(ctl >> AC_SHIFT);
2715          return (r <= 0) ? 0 : r; // suppress momentarily negative values
2716      }
2717  
# Line 2809 | Line 2727 | public class ForkJoinPool extends Abstra
2727       * @return {@code true} if all threads are currently idle
2728       */
2729      public boolean isQuiescent() {
2730 <        return (int)(ctl >> AC_SHIFT) + (config & SMASK) == 0;
2730 >        return parallelism + (int)(ctl >> AC_SHIFT) <= 0;
2731      }
2732  
2733      /**
# Line 2972 | Line 2890 | public class ForkJoinPool extends Abstra
2890                  }
2891              }
2892          }
2893 <        int pc = (config & SMASK);
2893 >        int pc = parallelism;
2894          int tc = pc + (short)(c >>> TC_SHIFT);
2895          int ac = pc + (int)(c >> AC_SHIFT);
2896          if (ac < 0) // ignore transient negative
# Line 3045 | Line 2963 | public class ForkJoinPool extends Abstra
2963      public boolean isTerminated() {
2964          long c = ctl;
2965          return ((c & STOP_BIT) != 0L &&
2966 <                (short)(c >>> TC_SHIFT) == -(config & SMASK));
2966 >                (short)(c >>> TC_SHIFT) + parallelism <= 0);
2967      }
2968  
2969      /**
# Line 3064 | Line 2982 | public class ForkJoinPool extends Abstra
2982      public boolean isTerminating() {
2983          long c = ctl;
2984          return ((c & STOP_BIT) != 0L &&
2985 <                (short)(c >>> TC_SHIFT) != -(config & SMASK));
2985 >                (short)(c >>> TC_SHIFT) + parallelism > 0);
2986      }
2987  
2988      /**
# Line 3082 | Line 3000 | public class ForkJoinPool extends Abstra
3000       * is interrupted, whichever happens first. Because the {@link
3001       * #commonPool()} never terminates until program shutdown, when
3002       * applied to the common pool, this method is equivalent to {@link
3003 <     * #awaitQuiescence} but always returns {@code false}.
3003 >     * #awaitQuiescence(long, TimeUnit)} but always returns {@code false}.
3004       *
3005       * @param timeout the maximum time to wait
3006       * @param unit the time unit of the timeout argument
# Line 3101 | Line 3019 | public class ForkJoinPool extends Abstra
3019          long nanos = unit.toNanos(timeout);
3020          if (isTerminated())
3021              return true;
3022 <        long startTime = System.nanoTime();
3023 <        boolean terminated = false;
3022 >        if (nanos <= 0L)
3023 >            return false;
3024 >        long deadline = System.nanoTime() + nanos;
3025          synchronized (this) {
3026 <            for (long waitTime = nanos, millis = 0L;;) {
3027 <                if (terminated = isTerminated() ||
3028 <                    waitTime <= 0L ||
3029 <                    (millis = unit.toMillis(waitTime)) <= 0L)
3030 <                    break;
3031 <                wait(millis);
3032 <                waitTime = nanos - (System.nanoTime() - startTime);
3026 >            for (;;) {
3027 >                if (isTerminated())
3028 >                    return true;
3029 >                if (nanos <= 0L)
3030 >                    return false;
3031 >                long millis = TimeUnit.NANOSECONDS.toMillis(nanos);
3032 >                wait(millis > 0L ? millis : 1L);
3033 >                nanos = deadline - System.nanoTime();
3034              }
3035          }
3116        return terminated;
3036      }
3037  
3038      /**
# Line 3152 | Line 3071 | public class ForkJoinPool extends Abstra
3071                  ForkJoinTask<?> t; WorkQueue q; int b;
3072                  if ((q = ws[r++ & m]) != null && (b = q.base) - q.top < 0) {
3073                      found = true;
3074 <                    if ((t = q.pollAt(b)) != null) {
3156 <                        if (q.base - q.top < 0)
3157 <                            signalWork(q);
3074 >                    if ((t = q.pollAt(b)) != null)
3075                          t.doExec();
3159                    }
3076                      break;
3077                  }
3078              }
# Line 3181 | Line 3097 | public class ForkJoinPool extends Abstra
3097       * not necessary. Method {@code block} blocks the current thread
3098       * if necessary (perhaps internally invoking {@code isReleasable}
3099       * before actually blocking). These actions are performed by any
3100 <     * thread invoking {@link ForkJoinPool#managedBlock}.  The
3101 <     * unusual methods in this API accommodate synchronizers that may,
3102 <     * but don't usually, block for long periods. Similarly, they
3100 >     * thread invoking {@link ForkJoinPool#managedBlock(ManagedBlocker)}.
3101 >     * The unusual methods in this API accommodate synchronizers that
3102 >     * may, but don't usually, block for long periods. Similarly, they
3103       * allow more efficient internal handling of cases in which
3104       * additional workers may be, but usually are not, needed to
3105       * ensure sufficient parallelism.  Toward this end,
# Line 3241 | Line 3157 | public class ForkJoinPool extends Abstra
3157  
3158          /**
3159           * Returns {@code true} if blocking is unnecessary.
3160 +         * @return {@code true} if blocking is unnecessary
3161           */
3162          boolean isReleasable();
3163      }
# Line 3270 | Line 3187 | public class ForkJoinPool extends Abstra
3187          Thread t = Thread.currentThread();
3188          if (t instanceof ForkJoinWorkerThread) {
3189              ForkJoinPool p = ((ForkJoinWorkerThread)t).pool;
3190 <            while (!blocker.isReleasable()) { // variant of helpSignal
3191 <                WorkQueue[] ws; WorkQueue q; int m, u;
3275 <                if ((ws = p.workQueues) != null && (m = ws.length - 1) >= 0) {
3276 <                    for (int i = 0; i <= m; ++i) {
3277 <                        if (blocker.isReleasable())
3278 <                            return;
3279 <                        if ((q = ws[i]) != null && q.base - q.top < 0) {
3280 <                            p.signalWork(q);
3281 <                            if ((u = (int)(p.ctl >>> 32)) >= 0 ||
3282 <                                (u >> UAC_SHIFT) >= 0)
3283 <                                break;
3284 <                        }
3285 <                    }
3286 <                }
3287 <                if (p.tryCompensate()) {
3190 >            while (!blocker.isReleasable()) {
3191 >                if (p.tryCompensate(p.ctl)) {
3192                      try {
3193                          do {} while (!blocker.isReleasable() &&
3194                                       !blocker.block());
# Line 3322 | Line 3226 | public class ForkJoinPool extends Abstra
3226      private static final long STEALCOUNT;
3227      private static final long PLOCK;
3228      private static final long INDEXSEED;
3229 +    private static final long QBASE;
3230      private static final long QLOCK;
3231  
3232      static {
# Line 3341 | Line 3246 | public class ForkJoinPool extends Abstra
3246              PARKBLOCKER = U.objectFieldOffset
3247                  (tk.getDeclaredField("parkBlocker"));
3248              Class<?> wk = WorkQueue.class;
3249 +            QBASE = U.objectFieldOffset
3250 +                (wk.getDeclaredField("base"));
3251              QLOCK = U.objectFieldOffset
3252                  (wk.getDeclaredField("qlock"));
3253              Class<?> ak = ForkJoinTask[].class;
# Line 3354 | Line 3261 | public class ForkJoinPool extends Abstra
3261          }
3262  
3263          submitters = new ThreadLocal<Submitter>();
3264 <        ForkJoinWorkerThreadFactory fac = defaultForkJoinWorkerThreadFactory =
3264 >        defaultForkJoinWorkerThreadFactory =
3265              new DefaultForkJoinWorkerThreadFactory();
3266          modifyThreadPermission = new RuntimePermission("modifyThread");
3267  
3268 <        /*
3269 <         * Establish common pool parameters.  For extra caution,
3270 <         * computations to set up common pool state are here; the
3271 <         * constructor just assigns these values to fields.
3272 <         */
3268 >        common = java.security.AccessController.doPrivileged
3269 >            (new java.security.PrivilegedAction<ForkJoinPool>() {
3270 >                public ForkJoinPool run() { return makeCommonPool(); }});
3271 >        int par = common.parallelism; // report 1 even if threads disabled
3272 >        commonParallelism = par > 0 ? par : 1;
3273 >    }
3274  
3275 <        int par = 0;
3276 <        Thread.UncaughtExceptionHandler handler = null;
3277 <        try {  // TBD: limit or report ignored exceptions?
3275 >    /**
3276 >     * Creates and returns the common pool, respecting user settings
3277 >     * specified via system properties.
3278 >     */
3279 >    private static ForkJoinPool makeCommonPool() {
3280 >        int parallelism = -1;
3281 >        ForkJoinWorkerThreadFactory factory
3282 >            = defaultForkJoinWorkerThreadFactory;
3283 >        UncaughtExceptionHandler handler = null;
3284 >        try {  // ignore exceptions in accesing/parsing properties
3285              String pp = System.getProperty
3286                  ("java.util.concurrent.ForkJoinPool.common.parallelism");
3372            String hp = System.getProperty
3373                ("java.util.concurrent.ForkJoinPool.common.exceptionHandler");
3287              String fp = System.getProperty
3288                  ("java.util.concurrent.ForkJoinPool.common.threadFactory");
3289 +            String hp = System.getProperty
3290 +                ("java.util.concurrent.ForkJoinPool.common.exceptionHandler");
3291 +            if (pp != null)
3292 +                parallelism = Integer.parseInt(pp);
3293              if (fp != null)
3294 <                fac = ((ForkJoinWorkerThreadFactory)ClassLoader.
3295 <                       getSystemClassLoader().loadClass(fp).newInstance());
3294 >                factory = ((ForkJoinWorkerThreadFactory)ClassLoader.
3295 >                           getSystemClassLoader().loadClass(fp).newInstance());
3296              if (hp != null)
3297 <                handler = ((Thread.UncaughtExceptionHandler)ClassLoader.
3297 >                handler = ((UncaughtExceptionHandler)ClassLoader.
3298                             getSystemClassLoader().loadClass(hp).newInstance());
3382            if (pp != null)
3383                par = Integer.parseInt(pp);
3299          } catch (Exception ignore) {
3300          }
3301  
3302 <        if (par <= 0)
3303 <            par = Runtime.getRuntime().availableProcessors();
3304 <        if (par > MAX_CAP)
3305 <            par = MAX_CAP;
3306 <        commonParallelism = par;
3307 <        long np = (long)(-par); // precompute initial ctl value
3308 <        long ct = ((np << AC_SHIFT) & AC_MASK) | ((np << TC_SHIFT) & TC_MASK);
3394 <
3395 <        common = new ForkJoinPool(par, ct, fac, handler);
3302 >        if (parallelism < 0 && // default 1 less than #cores
3303 >            (parallelism = Runtime.getRuntime().availableProcessors() - 1) < 0)
3304 >            parallelism = 0;
3305 >        if (parallelism > MAX_CAP)
3306 >            parallelism = MAX_CAP;
3307 >        return new ForkJoinPool(parallelism, factory, handler, LIFO_QUEUE,
3308 >                                "ForkJoinPool.commonPool-worker-");
3309      }
3310  
3311      /**

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines