ViewVC Help
View File | Revision Log | Show Annotations | Download File | Root Listing
root/jsr166/jsr166/src/main/java/util/concurrent/SynchronousQueue.java
Revision: 1.55
Committed: Mon Aug 1 15:26:40 2005 UTC (18 years, 10 months ago) by dl
Branch: MAIN
Changes since 1.54: +755 -463 lines
Log Message:
Replace algorithms

File Contents

# Content
1 /*
2 * Written by Doug Lea, Bill Scherer, and Michael Scott with
3 * assistance from members of JCP JSR-166 Expert Group and released to
4 * the public domain, as explained at
5 * http://creativecommons.org/licenses/publicdomain
6 */
7
8 package java.util.concurrent;
9 import java.util.concurrent.locks.*;
10 import java.util.concurrent.atomic.*;
11 import java.util.*;
12
13 /**
14 * A {@linkplain BlockingQueue blocking queue} in which each insert
15 * operation must wait for a corresponding remove operation by another
16 * thread, and vice versa. A synchronous queue does not have any
17 * internal capacity, not even a capacity of one. You cannot
18 * <tt>peek</tt> at a synchronous queue because an element is only
19 * present when you try to remove it; you cannot insert an element
20 * (using any method) unless another thread is trying to remove it;
21 * you cannot iterate as there is nothing to iterate. The
22 * <em>head</em> of the queue is the element that the first queued
23 * inserting thread is trying to add to the queue; if there is no such
24 * queued thread then no element is available for removal and
25 * <tt>poll()</tt> will return <tt>null</tt>. For purposes of other
26 * <tt>Collection</tt> methods (for example <tt>contains</tt>), a
27 * <tt>SynchronousQueue</tt> acts as an empty collection. This queue
28 * does not permit <tt>null</tt> elements.
29 *
30 * <p>Synchronous queues are similar to rendezvous channels used in
31 * CSP and Ada. They are well suited for handoff designs, in which an
32 * object running in one thread must sync up with an object running
33 * in another thread in order to hand it some information, event, or
34 * task.
35 *
36 * <p> This class supports an optional fairness policy for ordering
37 * waiting producer and consumer threads. By default, this ordering
38 * is not guaranteed. However, a queue constructed with fairness set
39 * to <tt>true</tt> grants threads access in FIFO order.
40 *
41 * <p>This class and its iterator implement all of the
42 * <em>optional</em> methods of the {@link Collection} and {@link
43 * Iterator} interfaces.
44 *
45 * <p>This class is a member of the
46 * <a href="{@docRoot}/../guide/collections/index.html">
47 * Java Collections Framework</a>.
48 *
49 * @since 1.5
50 * @author Doug Lea
51 * @param <E> the type of elements held in this collection
52 */
53 public class SynchronousQueue<E> extends AbstractQueue<E>
54 implements BlockingQueue<E>, java.io.Serializable {
55 private static final long serialVersionUID = -3223113410248163686L;
56
57 /*
58 * This class implements extensions of the dual stack and dual
59 * queue algorithms described in "Nonblocking Concurrent Objects
60 * with Condition Synchronization", by W. N. Scherer III and
61 * M. L. Scott. 18th Annual Conf. on Distributed Computing,
62 * Oct. 2004 (see also
63 * http://www.cs.rochester.edu/u/scott/synchronization/pseudocode/duals.html).
64 * The (Lifo) stack is used for non-fair mode, and the (Fifo)
65 * queue for fair mode. The performance of the two is generally
66 * similar. Fifo usually supports higher throughput under
67 * contention but Lifo maintains higher thread locality in common
68 * applications.
69 *
70 * A dual queue (and similarly stack) is one that at any given
71 * time either holds "data" -- items provided by put operations,
72 * or "requests" -- slots representing take operations, or is
73 * empty. A call to "fulfill" (i.e., a call requesting an item
74 * from a queue holding data or vice versa) dequeues a
75 * complementary node. The most interesting feature of these
76 * queues is that any operation can figure out which mode the
77 * queue is in, and act accordingly without needing locks.
78 *
79 * Both the queue and stack extend abstract class Transferer
80 * defining the single method transfer that does a put or a
81 * take. These are unified into a single method because in dual
82 * data structures, the put and take operations are symmetrical,
83 * so nearly all code can be combined. The resulting transfer
84 * methods are on the long side, but are easier to follow than
85 * they would be if broken up into nearly-duplicated parts.
86 *
87 * The queue and stack data structures share many conceptual
88 * similarities but very few concrete details. For simplicity,
89 * they are kept distinct so that they can later evolve
90 * separately.
91 *
92 * The algorithms here differ from the versions in the above paper
93 * in extending them for use in synchronous queues, as well as
94 * dealing with cancellation. The main differences include:
95 *
96 * 1. The orginal algorithms used bit-marked pointers, but
97 * the ones here use mode bits in nodes, leading to a number
98 * of further adaptations.
99 * 2. SynchronousQueues must block threads waiting to become
100 * fulfilled.
101 * 3. Nodes/threads that have been cancelled due to timeouts
102 * or interruptions are cleaned out of the lists to
103 * avoid garbage retention and memory depletion.
104 *
105 * Blocking is mainly accomplished using LockSupport park/unpark,
106 * except that nodes that appear to be the next ones to become
107 * fulfilled first spin a bit (on multiprocessors only). On very
108 * busy synchronous queues, spinning can dramatically improve
109 * throughput. And on less busy ones, the amount of spinning is
110 * small enough not to be noticeable.
111 *
112 * Cleaning is done in different ways in queues vs stacks. For
113 * queues, we can almost always remove a node immediately in O(1)
114 * time (modulo retries for consistency checks) when it is
115 * cancelled. But if it may be pinned as the current tail, it must
116 * wait until some subsequent cancellation. For stacks, we need a
117 * potentially O(n) traversal to be sure that we can remove the
118 * node, but this can run concurrently with other threads
119 * accessing the stack.
120 *
121 * While garbage collection takes care of most node reclamation
122 * issues that otherwise complicate nonblocking algorithms, care
123 * is made to "forget" references to data, other nodes, and
124 * threads that might be held on to long-term by blocked
125 * threads. In cases where setting to null would otherwise
126 * conflict with main algorithms, this is done by changing a
127 * node's link to now point to the node itself. This doesn't arise
128 * much for Stack nodes (because blocked threads do not hang on to
129 * old head pointers), but references in Queue nodes must be
130 * agressively forgotten to avoid reachability of everything any
131 * node has ever referred to since arrival.
132 */
133
134 /**
135 * Shared internal API for dual stacks and queues.
136 */
137 static abstract class Transferer {
138 /**
139 * Perform a put or take.
140 * @param e if non-null, the item to be handed to a consumer;
141 * if null, requests that transfer return an item offered by
142 * producer.
143 * @param timed if this operation should timeout
144 * @param nanos the timeout, in nanoseconds
145 * @return if nonnull, the item provided or received; if null,
146 * the operation failed due to timeout or interrupt -- the
147 * caller can distinguish which of these occurred by checking
148 * Thread.interrupted.
149 */
150 abstract Object transfer(Object e, boolean timed, long nanos);
151 }
152
153 /** The number of CPUs, for spin control */
154 static final int NCPUS = Runtime.getRuntime().availableProcessors();
155
156 /**
157 * The number of times to spin before blocking in timed waits.
158 * The value is empirically derived -- it works well across a
159 * variety of processors and OSes. Emprically, the best value
160 * seems not to vary with number of CPUs (beyond 2) so is just
161 * a constant.
162 */
163 static final int maxTimedSpins = (NCPUS < 2)? 0 : 32;
164
165 /**
166 * The number of times to spin before blocking in untimed
167 * waits. This is greater than timed value because untimed
168 * waits spin faster since they don't need to check times on
169 * each spin.
170 */
171 static final int maxUntimedSpins = maxTimedSpins * 16;
172
173 /**
174 * The number of nanoseconds for which it is faster to spin
175 * rather than to use timed park. A rough estimate suffices.
176 */
177 static final long spinForTimeoutThreshold = 1000L;
178
179 /** Dual stack */
180 static final class TransferStack extends Transferer {
181 /*
182 * This extends Scherer-Scott dual stack algorithm, differing,
183 * among other ways, by using "covering" nodes rather than
184 * bit-marked pointers: Fulfilling operations push on marker
185 * nodes (with FULFILLING bit set in mode) to reserve a spot
186 * to match a waiting node.
187 */
188
189 /* Modes for SNodes, ORed together in node fields */
190 /** Node represents an unfulfilled consumer */
191 static final int REQUEST = 0;
192 /** Node represents an unfulfilled producer */
193 static final int DATA = 1;
194 /** Node is fulfilling another unfulfilled DATA or REQUEST */
195 static final int FULFILLING = 2;
196
197 /** Return true if m has fulfilling bit set */
198 static boolean isFulfilling(int m) { return (m & FULFILLING) != 0; }
199
200 /** Node class for TransferStacks. */
201 static final class SNode {
202 volatile SNode next; // next node in stack
203 volatile SNode match; // the node matched to this
204 volatile Thread waiter; // to control park/unpark
205 Object item; // data; or null for REQUESTs
206 int mode;
207 // Note: item and mode fields don't need to be volatile
208 // since they are always written before, and read after,
209 // other volatile/atomic operations.
210
211 SNode(Object item) {
212 this.item = item;
213 }
214
215 static final AtomicReferenceFieldUpdater<SNode, SNode>
216 nextUpdater = AtomicReferenceFieldUpdater.newUpdater
217 (SNode.class, SNode.class, "next");
218
219 boolean casNext(SNode cmp, SNode val) {
220 return (cmp == next &&
221 nextUpdater.compareAndSet(this, cmp, val));
222 }
223
224 static final AtomicReferenceFieldUpdater<SNode, SNode>
225 matchUpdater = AtomicReferenceFieldUpdater.newUpdater
226 (SNode.class, SNode.class, "match");
227
228 /**
229 * Try to match node s to this node, if so, waking up
230 * thread. Fulfillers call tryMatch to identify their
231 * waiters. Waiters block until they have been
232 * matched.
233 * @param s the node to match
234 * @return true if successfully matched to s
235 */
236 boolean tryMatch(SNode s) {
237 if (match == null &&
238 matchUpdater.compareAndSet(this, null, s)) {
239 Thread w = waiter;
240 if (w != null) { // waiters need at most one unpark
241 waiter = null;
242 LockSupport.unpark(w);
243 }
244 return true;
245 }
246 return match == s;
247 }
248
249 /**
250 * Try to cancel a wait by matching node to itself.
251 */
252 void tryCancel() {
253 matchUpdater.compareAndSet(this, null, this);
254 }
255
256 boolean isCancelled() {
257 return match == this;
258 }
259 }
260
261 /** The head (top) of the stack */
262 volatile SNode head;
263
264 static final AtomicReferenceFieldUpdater<TransferStack, SNode>
265 headUpdater = AtomicReferenceFieldUpdater.newUpdater
266 (TransferStack.class, SNode.class, "head");
267
268 boolean casHead(SNode h, SNode nh) {
269 return h == head && headUpdater.compareAndSet(this, h, nh);
270 }
271
272 /**
273 * Create or reset fields of a node. Called only from transfer
274 * where the node to push on stack is lazily created and
275 * reused when possible to help reduce intervals between reads
276 * and CASes of head and to avoid surges of garbage when CASes
277 * to push nodes fail due to contention.
278 */
279 static SNode snode(SNode s, Object e, SNode next, int mode) {
280 if (s == null) s = new SNode(e);
281 s.mode = mode;
282 s.next = next;
283 return s;
284 }
285
286 /**
287 * Put or take an item.
288 */
289 Object transfer(Object e, boolean timed, long nanos) {
290 /*
291 * Basic algorithm is to loop trying one of three actions:
292 *
293 * 1. If apparently empty or already containing nodes of same
294 * mode, try to push node on stack and wait for a match,
295 * returning it, or null if cancelled.
296 *
297 * 2. If apparently containing node of complementary mode,
298 * try to push a fulfilling node on to stack, match
299 * with corresponding waiting node, pop both from
300 * stack, and return matched item. The matching or
301 * unlinking might not actually be necessary because of
302 * another threads performing action 3:
303 *
304 * 3. If top of stack already holds another fulfilling node,
305 * help it out by doing its match and/or pop
306 * operations, and then continue. The code for helping
307 * is essentially the same as for fulfilling, except
308 * that it doesn't return the item.
309 */
310
311 SNode s = null; // constructed/reused as needed
312 int mode = (e == null)? REQUEST : DATA;
313
314 for (;;) {
315 SNode h = head;
316 if (h == null || h.mode == mode) { // empty or same-mode
317 if (timed && nanos <= 0) { // can't wait
318 if (h != null && h.isCancelled())
319 casHead(h, h.next); // pop cancelled node
320 else
321 return null;
322 } else if (casHead(h, s = snode(s, e, h, mode))) {
323 SNode m = awaitFulfill(s, timed, nanos);
324 if (m == s) { // wait was cancelled
325 clean(s);
326 return null;
327 }
328 if ((h = head) != null && h.next == s)
329 casHead(h, s.next); // help s's fulfiller
330 return mode == REQUEST? m.item : s.item;
331 }
332 } else if (!isFulfilling(h.mode)) { // try to fulfill
333 if (h.isCancelled()) // already cancelled
334 casHead(h, h.next); // pop and retry
335 else if (casHead(h, s=snode(s, e, h, FULFILLING|mode))) {
336 for (;;) { // loop until matched or waiters disappear
337 SNode m = s.next; // m is s's match
338 if (m == null) { // all waiters are gone
339 casHead(s, null); // pop fulfill node
340 s = null; // use new node next time
341 break; // restart main loop
342 }
343 SNode mn = m.next;
344 if (m.tryMatch(s)) {
345 casHead(s, mn); // pop both s and m
346 return (mode == REQUEST)? m.item : s.item;
347 } else // lost match
348 s.casNext(m, mn); // help unlink
349 }
350 }
351 } else { // help a fulfiller
352 SNode m = h.next; // m is h's match
353 if (m == null) // waiter is gone
354 casHead(h, null); // pop fulfilling node
355 else {
356 SNode mn = m.next;
357 if (m.tryMatch(h)) // help match
358 casHead(h, mn); // pop both h and m
359 else // lost match
360 h.casNext(m, mn); // help unlink
361 }
362 }
363 }
364 }
365
366 /**
367 * Spin/block until node s is matched by a fulfill operation.
368 * @param s the waiting node
369 * @param timed true if timed wait
370 * @param nanos timeout value
371 * @return matched node, or s if cancelled
372 */
373 SNode awaitFulfill(SNode s, boolean timed, long nanos) {
374 /*
375 * When a node/thread is about to block, it sets its waiter
376 * field and then rechecks state at least one more time
377 * before actually parking, thus covering race vs
378 * fulfiller noticing that waiter is nonnull so should be
379 * woken.
380 *
381 * When invoked by nodes that appear at the point of call
382 * to be at the head of the stack, calls to park are
383 * preceded by spins to avoid blocking when producers and
384 * consumers are arriving very close in time. This can
385 * happen enough to bother only on multiprocessors.
386 *
387 * The order of checks for returning out of main loop
388 * reflects fact that interrupts have precedence over
389 * normal returns, which have precedence over
390 * timeouts. (So, on timeout, one last check for match is
391 * done before giving up.) Except that calls from untimed
392 * SynchronousQueue.{poll/offer} don't check interrupts
393 * and don't wait at all, so are trapped in transfer
394 * method rather than calling awaitFulfill.
395 */
396 long lastTime = (timed)? System.nanoTime() : 0;
397 Thread w = Thread.currentThread();
398 SNode h = head;
399 int spins = (shouldSpin(s)?
400 (timed? maxTimedSpins : maxUntimedSpins) : 0);
401 for (;;) {
402 if (w.isInterrupted())
403 s.tryCancel();
404 SNode m = s.match;
405 if (m != null)
406 return m;
407 if (timed) {
408 long now = System.nanoTime();
409 nanos -= now - lastTime;
410 lastTime = now;
411 if (nanos <= 0) {
412 s.tryCancel();
413 continue;
414 }
415 }
416 if (spins > 0)
417 spins = shouldSpin(s)? (spins-1) : 0;
418 else if (s.waiter == null)
419 s.waiter = w; // establish waiter so can park next iter
420 else if (!timed)
421 LockSupport.park(this);
422 else if (nanos > spinForTimeoutThreshold)
423 LockSupport.parkNanos(this, nanos);
424 }
425 }
426
427 /**
428 * Return true if node s is at head or there is an active
429 * fulfiller.
430 */
431 boolean shouldSpin(SNode s) {
432 SNode h = head;
433 return (h == null || h == s || isFulfilling(h.mode));
434 }
435
436 /**
437 * Unlink s from the stack
438 */
439 void clean(SNode s) {
440 s.item = null; // forget item
441 s.waiter = null; // forget thread
442
443 /*
444 * At worst we may need to traverse entire stack to unlink
445 * s. If there are multiple concurrent calls to clean, we
446 * might not see s if another thread has already removed
447 * it. But we can stop when we see any node known to
448 * follow s. We use s.next unless it too is cancelled, in
449 * which case we try the node one past. We don't check any
450 * futher because we don't want to doubly traverse just to
451 * find sentinel.
452 */
453
454 SNode past = s.next;
455 if (past != null && past.isCancelled())
456 past = past.next;
457
458 // Absorb cancelled nodes at head
459 SNode p;
460 while ((p = head) != null && p != past && p.isCancelled())
461 casHead(p, p.next);
462
463 // Unsplice embedded nodes
464 while (p != null && p != past) {
465 SNode n = p.next;
466 if (n != null && n.isCancelled())
467 p.casNext(n, n.next);
468 else
469 p = n;
470 }
471 }
472 }
473
474 /** Dual Queue. */
475 static final class TransferQueue extends Transferer {
476 /*
477 * This extends Scherer-Scott dual queue algorithm, differing,
478 * among other ways, by using modes within nodes rather than
479 * marked pointers. The algorithm is a little simpler than
480 * that for stacks because fulfillers do not need explicit
481 * nodes, and matching is done by CAS'ing QNode.item field
482 * from nonnull to null (for put) or vice versa (for take).
483 */
484
485 /** Node class for TransferQueue. */
486 static final class QNode {
487 volatile QNode next; // next node in queue
488 volatile Object item; // CAS'ed to or from null
489 volatile Thread waiter; // to control park/unpark
490 final boolean isData;
491
492 QNode(Object item, boolean isData) {
493 this.item = item;
494 this.isData = isData;
495 }
496
497 static final AtomicReferenceFieldUpdater<QNode, QNode>
498 nextUpdater = AtomicReferenceFieldUpdater.newUpdater
499 (QNode.class, QNode.class, "next");
500
501 boolean casNext(QNode cmp, QNode val) {
502 return (next == cmp &&
503 nextUpdater.compareAndSet(this, cmp, val));
504 }
505
506 static final AtomicReferenceFieldUpdater<QNode, Object>
507 itemUpdater = AtomicReferenceFieldUpdater.newUpdater
508 (QNode.class, Object.class, "item");
509
510 boolean casItem(Object cmp, Object val) {
511 return (item == cmp &&
512 itemUpdater.compareAndSet(this, cmp, val));
513 }
514
515 /**
516 * Try to cancel by CAS'ing ref to this as item.
517 */
518 void tryCancel(Object cmp) {
519 itemUpdater.compareAndSet(this, cmp, this);
520 }
521
522 boolean isCancelled() {
523 return item == this;
524 }
525 }
526
527 /** Head of queue */
528 transient volatile QNode head;
529 /** Tail of queue */
530 transient volatile QNode tail;
531 /**
532 * Reference to a cancelled node that might not yet have been
533 * unlinked from queue because it was the last inserted node
534 * when it cancelled.
535 */
536 transient volatile QNode cleanMe;
537
538 TransferQueue() {
539 QNode h = new QNode(null, false); // initialize to dummy node.
540 head = h;
541 tail = h;
542 }
543
544 static final AtomicReferenceFieldUpdater<TransferQueue, QNode>
545 headUpdater = AtomicReferenceFieldUpdater.newUpdater
546 (TransferQueue.class, QNode.class, "head");
547
548 /**
549 * Try to cas nh as new head; if successful unlink
550 * old head's next node to avoid garbage retention.
551 */
552 void advanceHead(QNode h, QNode nh) {
553 if (h == head && headUpdater.compareAndSet(this, h, nh))
554 h.next = h; // forget old next
555 }
556
557 static final AtomicReferenceFieldUpdater<TransferQueue, QNode>
558 tailUpdater = AtomicReferenceFieldUpdater.newUpdater
559 (TransferQueue.class, QNode.class, "tail");
560
561 /**
562 * Try to cas nt as new tail.
563 */
564 void advanceTail(QNode t, QNode nt) {
565 if (tail == t)
566 tailUpdater.compareAndSet(this, t, nt);
567 }
568
569 static final AtomicReferenceFieldUpdater<TransferQueue, QNode>
570 cleanMeUpdater = AtomicReferenceFieldUpdater.newUpdater
571 (TransferQueue.class, QNode.class, "cleanMe");
572
573 /**
574 * Try to CAS cleanMe slot
575 */
576 boolean casCleanMe(QNode cmp, QNode val) {
577 return (cleanMe == cmp &&
578 cleanMeUpdater.compareAndSet(this, cmp, val));
579 }
580
581 /**
582 * Put or take an item.
583 */
584 Object transfer(Object e, boolean timed, long nanos) {
585 /* Basic algorithm is to loop trying to take either of
586 * two actions:
587 *
588 * 1. If queue apparently empty or holding same-mode nodes,
589 * try to add node to queue of waiters, wait to be
590 * fulfilled (or cancelled) and return matching item.
591 *
592 * 2. If queue apparently contains waiting items, and this
593 * call is of complementary mode, try to fulfill by CAS'ing
594 * item field of waiting node and dequeuing it, and then
595 * returning matching item.
596 *
597 * In each case, along the way, check for and try to help
598 * advance head and tail on behalf of other stalled/slow
599 * threads.
600 *
601 * The loop starts off with a null check guarding against
602 * seeing uninitialized head or tail values. This never
603 * happens in current SynchronousQueue, but could if
604 * callers held non-volatile/final ref to the
605 * transferer. The check is here anyway because it places
606 * null checks at top of loop, which is usually faster
607 * than having them implicitly interspersed.
608 */
609
610 QNode s = null; // constructed/reused as needed
611 boolean isData = (e != null);
612
613 for (;;) {
614 QNode t = tail;
615 QNode h = head;
616 if (t == null || h == null) // saw unitialized values
617 continue; // spin
618
619 if (h == t || t.isData == isData) { // empty or same-mode
620 QNode tn = t.next;
621 if (t != tail) // inconsistent read
622 continue;
623 if (tn != null) { // lagging tail
624 advanceTail(t, tn);
625 continue;
626 }
627 if (timed && nanos <= 0) // can't wait
628 return null;
629 if (s == null)
630 s = new QNode(e, isData);
631 if (!t.casNext(null, s)) // failed to link in
632 continue;
633
634 advanceTail(t, s); // swing tail and wait
635 Object x = awaitFulfill(s, e, timed, nanos);
636 if (x == s) { // wait was cancelled
637 clean(t, s);
638 return null;
639 }
640
641 if (s.next != s) { // not already unlinked
642 advanceHead(t, s); // unlink
643 if (x != null) // and forget fields
644 s.item = s;
645 s.waiter = null;
646 }
647 return (x != null)? x : e;
648
649 } else { // complementary-mode
650 QNode m = h.next; // node to fulfill
651 if (t != tail || m == null || h != head)
652 continue; // inconsistent read
653
654 Object x = m.item;
655 if (isData == (x != null) || // m already fulfilled
656 x == m || // m cancelled
657 !m.casItem(x, e)) { // lost CAS
658 advanceHead(h, m); // dequeue and retry
659 continue;
660 }
661
662 advanceHead(h, m); // successfully fulfilled
663 LockSupport.unpark(m.waiter);
664 return (x != null)? x : e;
665 }
666 }
667 }
668
669 /**
670 * Spin/block until node s is fulfilled.
671 * @param s the waiting node
672 * @param e the comparison value for checking match
673 * @param timed true if timed wait
674 * @param nanos timeout value
675 * @return matched item, or s if cancelled
676 */
677 Object awaitFulfill(QNode s, Object e, boolean timed, long nanos) {
678 /* Same idea as TransferStack.awaitFulfill */
679 long lastTime = (timed)? System.nanoTime() : 0;
680 Thread w = Thread.currentThread();
681 int spins = ((head.next == s) ?
682 (timed? maxTimedSpins : maxUntimedSpins) : 0);
683 for (;;) {
684 if (w.isInterrupted())
685 s.tryCancel(e);
686 Object x = s.item;
687 if (x != e)
688 return x;
689 if (timed) {
690 long now = System.nanoTime();
691 nanos -= now - lastTime;
692 lastTime = now;
693 if (nanos <= 0) {
694 s.tryCancel(e);
695 continue;
696 }
697 }
698 if (spins > 0)
699 --spins;
700 else if (s.waiter == null)
701 s.waiter = w;
702 else if (!timed)
703 LockSupport.park(this);
704 else if (nanos > spinForTimeoutThreshold)
705 LockSupport.parkNanos(this, nanos);
706 }
707 }
708
709 /**
710 * Get rid of cancelled node s with original predecessor pred.
711 */
712 void clean(QNode pred, QNode s) {
713 s.waiter = null; // forget thread
714 /*
715 * At any given time, exactly one node on list cannot be
716 * deleted -- the last inserted node. To accommodate this,
717 * if we cannot delete s, we save its predecessor as
718 * "cleanMe", deleting the previously saved version
719 * first. At least one of node s or the node previously
720 * saved can always be deleted, so this always terminates.
721 */
722 while (pred.next == s) { // Return early if already unlinked
723 QNode h = head;
724 QNode hn = h.next; // Absorb cancelled first node as head
725 if (hn != null && hn.isCancelled()) {
726 advanceHead(h, hn);
727 continue;
728 }
729 QNode t = tail; // Ensure consistent read for tail
730 if (t == h)
731 return;
732 QNode tn = t.next;
733 if (t != tail)
734 continue;
735 if (tn != null) {
736 advanceTail(t, tn);
737 continue;
738 }
739 if (s != t) { // If not tail, try to unsplice
740 QNode sn = s.next;
741 if (sn == s || pred.casNext(s, sn))
742 return;
743 }
744 QNode dp = cleanMe;
745 if (dp != null) { // Try unlinking previous cancelled node
746 QNode d = dp.next;
747 QNode dn;
748 if (d == null || // d is gone or
749 d == dp || // d is off list or
750 !d.isCancelled() || // d not cancelled or
751 (d != t && // d not tail and
752 (dn = d.next) != null && // has successor
753 dn != d && // that is on list
754 dp.casNext(d, dn))) // d unspliced
755 casCleanMe(dp, null);
756 if (dp == pred)
757 return; // s is already saved node
758 } else if (casCleanMe(null, pred))
759 return; // Postpone cleaning s
760 }
761 }
762 }
763
764 /**
765 * The transferer. Set only in constructor, but cannot be declared
766 * as final without further complicating serialization. Since
767 * this is accessed only once per public method, there isn't a
768 * noticeable performance penalty for using volatile instead of
769 * final here.
770 */
771 private transient volatile Transferer transferer;
772
773 /**
774 * Creates a <tt>SynchronousQueue</tt> with nonfair access policy.
775 */
776 public SynchronousQueue() {
777 this(false);
778 }
779
780 /**
781 * Creates a <tt>SynchronousQueue</tt> with specified fairness policy.
782 * @param fair if true, waiting threads contend in FIFO order for access;
783 * otherwise the order is unspecified.
784 */
785 public SynchronousQueue(boolean fair) {
786 transferer = (fair)? new TransferQueue() : new TransferStack();
787 }
788
789 /**
790 * Adds the specified element to this queue, waiting if necessary for
791 * another thread to receive it.
792 *
793 * @throws InterruptedException {@inheritDoc}
794 * @throws NullPointerException {@inheritDoc}
795 */
796 public void put(E o) throws InterruptedException {
797 if (o == null) throw new NullPointerException();
798 if (transferer.transfer(o, false, 0) == null)
799 throw new InterruptedException();
800 }
801
802 /**
803 * Inserts the specified element into this queue, waiting if necessary
804 * up to the specified wait time for another thread to receive it.
805 *
806 * @return <tt>true</tt> if successful, or <tt>false</tt> if the
807 * specified waiting time elapses before a consumer appears.
808 * @throws InterruptedException {@inheritDoc}
809 * @throws NullPointerException {@inheritDoc}
810 */
811 public boolean offer(E o, long timeout, TimeUnit unit)
812 throws InterruptedException {
813 if (o == null) throw new NullPointerException();
814 if (transferer.transfer(o, true, unit.toNanos(timeout)) != null)
815 return true;
816 if (!Thread.interrupted())
817 return false;
818 throw new InterruptedException();
819 }
820
821 /**
822 * Inserts the specified element into this queue, if another thread is
823 * waiting to receive it.
824 *
825 * @param e the element to add
826 * @return <tt>true</tt> if the element was added to this queue, else
827 * <tt>false</tt>
828 * @throws NullPointerException if the specified element is null
829 */
830 public boolean offer(E e) {
831 if (e == null) throw new NullPointerException();
832 return transferer.transfer(e, true, 0) != null;
833 }
834
835 /**
836 * Retrieves and removes the head of this queue, waiting if necessary
837 * for another thread to insert it.
838 *
839 * @return the head of this queue
840 * @throws InterruptedException {@inheritDoc}
841 */
842 public E take() throws InterruptedException {
843 Object e = transferer.transfer(null, false, 0);
844 if (e != null)
845 return (E)e;
846 throw new InterruptedException();
847 }
848
849 /**
850 * Retrieves and removes the head of this queue, waiting
851 * if necessary up to the specified wait time, for another thread
852 * to insert it.
853 *
854 * @return the head of this queue, or <tt>null</tt> if the
855 * specified waiting time elapses before an element is present.
856 * @throws InterruptedException {@inheritDoc}
857 */
858 public E poll(long timeout, TimeUnit unit) throws InterruptedException {
859 Object e = transferer.transfer(null, true, unit.toNanos(timeout));
860 if (e != null || !Thread.interrupted())
861 return (E)e;
862 throw new InterruptedException();
863 }
864
865 /**
866 * Retrieves and removes the head of this queue, if another thread
867 * is currently making an element available.
868 *
869 * @return the head of this queue, or <tt>null</tt> if no
870 * element is available.
871 */
872 public E poll() {
873 return (E)transferer.transfer(null, true, 0);
874 }
875
876 /**
877 * Always returns <tt>true</tt>.
878 * A <tt>SynchronousQueue</tt> has no internal capacity.
879 * @return <tt>true</tt>
880 */
881 public boolean isEmpty() {
882 return true;
883 }
884
885 /**
886 * Always returns zero.
887 * A <tt>SynchronousQueue</tt> has no internal capacity.
888 * @return zero.
889 */
890 public int size() {
891 return 0;
892 }
893
894 /**
895 * Always returns zero.
896 * A <tt>SynchronousQueue</tt> has no internal capacity.
897 * @return zero.
898 */
899 public int remainingCapacity() {
900 return 0;
901 }
902
903 /**
904 * Does nothing.
905 * A <tt>SynchronousQueue</tt> has no internal capacity.
906 */
907 public void clear() {
908 }
909
910 /**
911 * Always returns <tt>false</tt>.
912 * A <tt>SynchronousQueue</tt> has no internal capacity.
913 * @param o the element
914 * @return <tt>false</tt>
915 */
916 public boolean contains(Object o) {
917 return false;
918 }
919
920 /**
921 * Always returns <tt>false</tt>.
922 * A <tt>SynchronousQueue</tt> has no internal capacity.
923 *
924 * @param o the element to remove
925 * @return <tt>false</tt>
926 */
927 public boolean remove(Object o) {
928 return false;
929 }
930
931 /**
932 * Returns <tt>false</tt> unless given collection is empty.
933 * A <tt>SynchronousQueue</tt> has no internal capacity.
934 * @param c the collection
935 * @return <tt>false</tt> unless given collection is empty
936 */
937 public boolean containsAll(Collection<?> c) {
938 return c.isEmpty();
939 }
940
941 /**
942 * Always returns <tt>false</tt>.
943 * A <tt>SynchronousQueue</tt> has no internal capacity.
944 * @param c the collection
945 * @return <tt>false</tt>
946 */
947 public boolean removeAll(Collection<?> c) {
948 return false;
949 }
950
951 /**
952 * Always returns <tt>false</tt>.
953 * A <tt>SynchronousQueue</tt> has no internal capacity.
954 * @param c the collection
955 * @return <tt>false</tt>
956 */
957 public boolean retainAll(Collection<?> c) {
958 return false;
959 }
960
961 /**
962 * Always returns <tt>null</tt>.
963 * A <tt>SynchronousQueue</tt> does not return elements
964 * unless actively waited on.
965 * @return <tt>null</tt>
966 */
967 public E peek() {
968 return null;
969 }
970
971 static class EmptyIterator<E> implements Iterator<E> {
972 public boolean hasNext() {
973 return false;
974 }
975 public E next() {
976 throw new NoSuchElementException();
977 }
978 public void remove() {
979 throw new IllegalStateException();
980 }
981 }
982
983 /**
984 * Returns an empty iterator in which <tt>hasNext</tt> always returns
985 * <tt>false</tt>.
986 *
987 * @return an empty iterator
988 */
989 public Iterator<E> iterator() {
990 return new EmptyIterator<E>();
991 }
992
993 /**
994 * Returns a zero-length array.
995 * @return a zero-length array
996 */
997 public Object[] toArray() {
998 return new Object[0];
999 }
1000
1001 /**
1002 * Sets the zeroeth element of the specified array to <tt>null</tt>
1003 * (if the array has non-zero length) and returns it.
1004 *
1005 * @param a the array
1006 * @return the specified array
1007 * @throws NullPointerException if the specified array is null
1008 */
1009 public <T> T[] toArray(T[] a) {
1010 if (a.length > 0)
1011 a[0] = null;
1012 return a;
1013 }
1014
1015 /**
1016 * @throws UnsupportedOperationException {@inheritDoc}
1017 * @throws ClassCastException {@inheritDoc}
1018 * @throws NullPointerException {@inheritDoc}
1019 * @throws IllegalArgumentException {@inheritDoc}
1020 */
1021 public int drainTo(Collection<? super E> c) {
1022 if (c == null)
1023 throw new NullPointerException();
1024 if (c == this)
1025 throw new IllegalArgumentException();
1026 int n = 0;
1027 E e;
1028 while ( (e = poll()) != null) {
1029 c.add(e);
1030 ++n;
1031 }
1032 return n;
1033 }
1034
1035 /**
1036 * @throws UnsupportedOperationException {@inheritDoc}
1037 * @throws ClassCastException {@inheritDoc}
1038 * @throws NullPointerException {@inheritDoc}
1039 * @throws IllegalArgumentException {@inheritDoc}
1040 */
1041 public int drainTo(Collection<? super E> c, int maxElements) {
1042 if (c == null)
1043 throw new NullPointerException();
1044 if (c == this)
1045 throw new IllegalArgumentException();
1046 int n = 0;
1047 E e;
1048 while (n < maxElements && (e = poll()) != null) {
1049 c.add(e);
1050 ++n;
1051 }
1052 return n;
1053 }
1054
1055 /*
1056 * To cope with serialization strategy in the 1.5 version of
1057 * SynchronousQueue, we declare some unused classes and fields
1058 * that exist solely to enable serializability across versions.
1059 * These fields are never used, so are initialized only if this
1060 * object is ever serialized or deserialized.
1061 */
1062
1063 static class WaitQueue implements java.io.Serializable { }
1064 static class LifoWaitQueue extends WaitQueue {
1065 private static final long serialVersionUID = -3633113410248163686L;
1066 }
1067 static class FifoWaitQueue extends WaitQueue {
1068 private static final long serialVersionUID = -3623113410248163686L;
1069 }
1070 private ReentrantLock qlock;
1071 private WaitQueue waitingProducers;
1072 private WaitQueue waitingConsumers;
1073
1074 /**
1075 * Save the state to a stream (that is, serialize it).
1076 *
1077 * @param s the stream
1078 */
1079 private void writeObject(java.io.ObjectOutputStream s)
1080 throws java.io.IOException {
1081 boolean fair = transferer instanceof TransferQueue;
1082 if (fair) {
1083 qlock = new ReentrantLock(true);
1084 waitingProducers = new FifoWaitQueue();
1085 waitingConsumers = new FifoWaitQueue();
1086 }
1087 else {
1088 qlock = new ReentrantLock();
1089 waitingProducers = new LifoWaitQueue();
1090 waitingConsumers = new LifoWaitQueue();
1091 }
1092 s.defaultWriteObject();
1093 }
1094
1095 private void readObject(final java.io.ObjectInputStream s)
1096 throws java.io.IOException, ClassNotFoundException {
1097 s.defaultReadObject();
1098 if (waitingProducers instanceof FifoWaitQueue)
1099 transferer = new TransferQueue();
1100 else
1101 transferer = new TransferStack();
1102 }
1103
1104 }