1 /*
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
3 *
4 * This code is free software; you can redistribute it and/or modify it
5 * under the terms of the GNU General Public License version 2 only, as
6 * published by the Free Software Foundation. Oracle designates this
7 * particular file as subject to the "Classpath" exception as provided
8 * by Oracle in the LICENSE file that accompanied this code.
9 *
10 * This code is distributed in the hope that it will be useful, but WITHOUT
11 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
12 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
13 * version 2 for more details (a copy is included in the LICENSE file that
14 * accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License version
17 * 2 along with this work; if not, write to the Free Software Foundation,
18 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
19 *
20 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
21 * or visit www.oracle.com if you need additional information or have any
22 * questions.
23 */
24
25 /*
26 * This file is available under and governed by the GNU General Public
27 * License version 2 only, as published by the Free Software Foundation.
28 * However, the following notice accompanied the original version of this
29 * file:
30 *
31 * Written by Doug Lea with assistance from members of JCP JSR-166
32 * Expert Group and released to the public domain, as explained at
33 * http://creativecommons.org/publicdomain/zero/1.0/
34 */
35
36 package java.util.concurrent;
37
38 import java.io.ObjectStreamField;
39 import java.io.Serializable;
40 import java.lang.reflect.ParameterizedType;
41 import java.lang.reflect.Type;
42 import java.util.AbstractMap;
43 import java.util.Arrays;
44 import java.util.Collection;
45 import java.util.Enumeration;
46 import java.util.HashMap;
47 import java.util.Hashtable;
48 import java.util.Iterator;
49 import java.util.Map;
50 import java.util.NoSuchElementException;
51 import java.util.Set;
52 import java.util.Spliterator;
53 import java.util.concurrent.atomic.AtomicReference;
54 import java.util.concurrent.locks.LockSupport;
55 import java.util.concurrent.locks.ReentrantLock;
56 import java.util.function.BiConsumer;
57 import java.util.function.BiFunction;
58 import java.util.function.Consumer;
59 import java.util.function.DoubleBinaryOperator;
60 import java.util.function.Function;
61 import java.util.function.IntBinaryOperator;
62 import java.util.function.LongBinaryOperator;
63 import java.util.function.Predicate;
64 import java.util.function.ToDoubleBiFunction;
65 import java.util.function.ToDoubleFunction;
66 import java.util.function.ToIntBiFunction;
67 import java.util.function.ToIntFunction;
68 import java.util.function.ToLongBiFunction;
69 import java.util.function.ToLongFunction;
70 import java.util.stream.Stream;
71 import jdk.internal.misc.Unsafe;
72
73 /**
74 * A hash table supporting full concurrency of retrievals and
75 * high expected concurrency for updates. This class obeys the
76 * same functional specification as {@link java.util.Hashtable}, and
77 * includes versions of methods corresponding to each method of
78 * {@code Hashtable}. However, even though all operations are
79 * thread-safe, retrieval operations do <em>not</em> entail locking,
80 * and there is <em>not</em> any support for locking the entire table
81 * in a way that prevents all access. This class is fully
82 * interoperable with {@code Hashtable} in programs that rely on its
83 * thread safety but not on its synchronization details.
84 *
85 * <p>Retrieval operations (including {@code get}) generally do not
86 * block, so may overlap with update operations (including {@code put}
87 * and {@code remove}). Retrievals reflect the results of the most
88 * recently <em>completed</em> update operations holding upon their
89 * onset. (More formally, an update operation for a given key bears a
90 * <em>happens-before</em> relation with any (non-null) retrieval for
91 * that key reporting the updated value.) For aggregate operations
92 * such as {@code putAll} and {@code clear}, concurrent retrievals may
93 * reflect insertion or removal of only some entries. Similarly,
94 * Iterators, Spliterators and Enumerations return elements reflecting the
95 * state of the hash table at some point at or since the creation of the
96 * iterator/enumeration. They do <em>not</em> throw {@link
97 * java.util.ConcurrentModificationException ConcurrentModificationException}.
98 * However, iterators are designed to be used by only one thread at a time.
99 * Bear in mind that the results of aggregate status methods including
100 * {@code size}, {@code isEmpty}, and {@code containsValue} are typically
101 * useful only when a map is not undergoing concurrent updates in other threads.
102 * Otherwise the results of these methods reflect transient states
103 * that may be adequate for monitoring or estimation purposes, but not
104 * for program control.
105 *
106 * <p>The table is dynamically expanded when there are too many
107 * collisions (i.e., keys that have distinct hash codes but fall into
108 * the same slot modulo the table size), with the expected average
109 * effect of maintaining roughly two bins per mapping (corresponding
110 * to a 0.75 load factor threshold for resizing). There may be much
111 * variance around this average as mappings are added and removed, but
112 * overall, this maintains a commonly accepted time/space tradeoff for
113 * hash tables. However, resizing this or any other kind of hash
114 * table may be a relatively slow operation. When possible, it is a
115 * good idea to provide a size estimate as an optional {@code
116 * initialCapacity} constructor argument. An additional optional
117 * {@code loadFactor} constructor argument provides a further means of
118 * customizing initial table capacity by specifying the table density
119 * to be used in calculating the amount of space to allocate for the
120 * given number of elements. Also, for compatibility with previous
121 * versions of this class, constructors may optionally specify an
122 * expected {@code concurrencyLevel} as an additional hint for
123 * internal sizing. Note that using many keys with exactly the same
124 * {@code hashCode()} is a sure way to slow down performance of any
125 * hash table. To ameliorate impact, when keys are {@link Comparable},
126 * this class may use comparison order among keys to help break ties.
127 *
128 * <p>A {@link Set} projection of a ConcurrentHashMap may be created
129 * (using {@link #newKeySet()} or {@link #newKeySet(int)}), or viewed
130 * (using {@link #keySet(Object)} when only keys are of interest, and the
131 * mapped values are (perhaps transiently) not used or all take the
132 * same mapping value.
133 *
134 * <p>A ConcurrentHashMap can be used as a scalable frequency map (a
135 * form of histogram or multiset) by using {@link
136 * java.util.concurrent.atomic.LongAdder} values and initializing via
137 * {@link #computeIfAbsent computeIfAbsent}. For example, to add a count
138 * to a {@code ConcurrentHashMap<String,LongAdder> freqs}, you can use
139 * {@code freqs.computeIfAbsent(key, k -> new LongAdder()).increment();}
140 *
141 * <p>This class and its views and iterators implement all of the
142 * <em>optional</em> methods of the {@link Map} and {@link Iterator}
143 * interfaces.
144 *
145 * <p>Like {@link Hashtable} but unlike {@link HashMap}, this class
146 * does <em>not</em> allow {@code null} to be used as a key or value.
147 *
148 * <p>ConcurrentHashMaps support a set of sequential and parallel bulk
149 * operations that, unlike most {@link Stream} methods, are designed
150 * to be safely, and often sensibly, applied even with maps that are
151 * being concurrently updated by other threads; for example, when
152 * computing a snapshot summary of the values in a shared registry.
153 * There are three kinds of operation, each with four forms, accepting
154 * functions with keys, values, entries, and (key, value) pairs as
155 * arguments and/or return values. Because the elements of a
156 * ConcurrentHashMap are not ordered in any particular way, and may be
157 * processed in different orders in different parallel executions, the
158 * correctness of supplied functions should not depend on any
159 * ordering, or on any other objects or values that may transiently
160 * change while computation is in progress; and except for forEach
161 * actions, should ideally be side-effect-free. Bulk operations on
162 * {@link java.util.Map.Entry} objects do not support method {@code
163 * setValue}.
164 *
165 * <ul>
166 * <li>forEach: Performs a given action on each element.
167 * A variant form applies a given transformation on each element
168 * before performing the action.
169 *
170 * <li>search: Returns the first available non-null result of
171 * applying a given function on each element; skipping further
172 * search when a result is found.
173 *
174 * <li>reduce: Accumulates each element. The supplied reduction
175 * function cannot rely on ordering (more formally, it should be
176 * both associative and commutative). There are five variants:
177 *
178 * <ul>
179 *
180 * <li>Plain reductions. (There is not a form of this method for
181 * (key, value) function arguments since there is no corresponding
182 * return type.)
183 *
184 * <li>Mapped reductions that accumulate the results of a given
185 * function applied to each element.
186 *
187 * <li>Reductions to scalar doubles, longs, and ints, using a
188 * given basis value.
189 *
190 * </ul>
191 * </ul>
192 *
193 * <p>These bulk operations accept a {@code parallelismThreshold}
194 * argument. Methods proceed sequentially if the current map size is
195 * estimated to be less than the given threshold. Using a value of
196 * {@code Long.MAX_VALUE} suppresses all parallelism. Using a value
197 * of {@code 1} results in maximal parallelism by partitioning into
198 * enough subtasks to fully utilize the {@link
199 * ForkJoinPool#commonPool()} that is used for all parallel
200 * computations. Normally, you would initially choose one of these
201 * extreme values, and then measure performance of using in-between
202 * values that trade off overhead versus throughput.
203 *
204 * <p>The concurrency properties of bulk operations follow
205 * from those of ConcurrentHashMap: Any non-null result returned
206 * from {@code get(key)} and related access methods bears a
207 * happens-before relation with the associated insertion or
208 * update. The result of any bulk operation reflects the
209 * composition of these per-element relations (but is not
210 * necessarily atomic with respect to the map as a whole unless it
211 * is somehow known to be quiescent). Conversely, because keys
212 * and values in the map are never null, null serves as a reliable
213 * atomic indicator of the current lack of any result. To
214 * maintain this property, null serves as an implicit basis for
215 * all non-scalar reduction operations. For the double, long, and
216 * int versions, the basis should be one that, when combined with
217 * any other value, returns that other value (more formally, it
218 * should be the identity element for the reduction). Most common
219 * reductions have these properties; for example, computing a sum
220 * with basis 0 or a minimum with basis MAX_VALUE.
221 *
222 * <p>Search and transformation functions provided as arguments
223 * should similarly return null to indicate the lack of any result
224 * (in which case it is not used). In the case of mapped
225 * reductions, this also enables transformations to serve as
226 * filters, returning null (or, in the case of primitive
227 * specializations, the identity basis) if the element should not
228 * be combined. You can create compound transformations and
229 * filterings by composing them yourself under this "null means
230 * there is nothing there now" rule before using them in search or
231 * reduce operations.
232 *
233 * <p>Methods accepting and/or returning Entry arguments maintain
234 * key-value associations. They may be useful for example when
235 * finding the key for the greatest value. Note that "plain" Entry
236 * arguments can be supplied using {@code new
237 * AbstractMap.SimpleEntry(k,v)}.
238 *
239 * <p>Bulk operations may complete abruptly, throwing an
240 * exception encountered in the application of a supplied
241 * function. Bear in mind when handling such exceptions that other
242 * concurrently executing functions could also have thrown
243 * exceptions, or would have done so if the first exception had
244 * not occurred.
245 *
246 * <p>Speedups for parallel compared to sequential forms are common
247 * but not guaranteed. Parallel operations involving brief functions
248 * on small maps may execute more slowly than sequential forms if the
249 * underlying work to parallelize the computation is more expensive
250 * than the computation itself. Similarly, parallelization may not
251 * lead to much actual parallelism if all processors are busy
252 * performing unrelated tasks.
253 *
254 * <p>All arguments to all task methods must be non-null.
255 *
256 * <p>This class is a member of the
257 * <a href="{@docRoot}/java/util/package-summary.html#CollectionsFramework">
258 * Java Collections Framework</a>.
259 *
260 * @since 1.5
261 * @author Doug Lea
262 * @param <K> the type of keys maintained by this map
263 * @param <V> the type of mapped values
264 */
265 public class ConcurrentHashMap<K,V> extends AbstractMap<K,V>
266 implements ConcurrentMap<K,V>, Serializable {
267 private static final long serialVersionUID = 7249069246763182397L;
268
269 /*
270 * Overview:
271 *
272 * The primary design goal of this hash table is to maintain
273 * concurrent readability (typically method get(), but also
274 * iterators and related methods) while minimizing update
275 * contention. Secondary goals are to keep space consumption about
276 * the same or better than java.util.HashMap, and to support high
277 * initial insertion rates on an empty table by many threads.
278 *
279 * This map usually acts as a binned (bucketed) hash table. Each
280 * key-value mapping is held in a Node. Most nodes are instances
281 * of the basic Node class with hash, key, value, and next
282 * fields. However, various subclasses exist: TreeNodes are
283 * arranged in balanced trees, not lists. TreeBins hold the roots
284 * of sets of TreeNodes. ForwardingNodes are placed at the heads
285 * of bins during resizing. ReservationNodes are used as
286 * placeholders while establishing values in computeIfAbsent and
287 * related methods. The types TreeBin, ForwardingNode, and
288 * ReservationNode do not hold normal user keys, values, or
289 * hashes, and are readily distinguishable during search etc
290 * because they have negative hash fields and null key and value
291 * fields. (These special nodes are either uncommon or transient,
292 * so the impact of carrying around some unused fields is
293 * insignificant.)
294 *
295 * The table is lazily initialized to a power-of-two size upon the
296 * first insertion. Each bin in the table normally contains a
297 * list of Nodes (most often, the list has only zero or one Node).
298 * Table accesses require volatile/atomic reads, writes, and
299 * CASes. Because there is no other way to arrange this without
300 * adding further indirections, we use intrinsics
301 * (jdk.internal.misc.Unsafe) operations.
302 *
303 * We use the top (sign) bit of Node hash fields for control
304 * purposes -- it is available anyway because of addressing
305 * constraints. Nodes with negative hash fields are specially
306 * handled or ignored in map methods.
307 *
308 * Insertion (via put or its variants) of the first node in an
309 * empty bin is performed by just CASing it to the bin. This is
310 * by far the most common case for put operations under most
311 * key/hash distributions. Other update operations (insert,
312 * delete, and replace) require locks. We do not want to waste
313 * the space required to associate a distinct lock object with
314 * each bin, so instead use the first node of a bin list itself as
315 * a lock. Locking support for these locks relies on builtin
316 * "synchronized" monitors.
317 *
318 * Using the first node of a list as a lock does not by itself
319 * suffice though: When a node is locked, any update must first
320 * validate that it is still the first node after locking it, and
321 * retry if not. Because new nodes are always appended to lists,
322 * once a node is first in a bin, it remains first until deleted
323 * or the bin becomes invalidated (upon resizing).
324 *
325 * The main disadvantage of per-bin locks is that other update
326 * operations on other nodes in a bin list protected by the same
327 * lock can stall, for example when user equals() or mapping
328 * functions take a long time. However, statistically, under
329 * random hash codes, this is not a common problem. Ideally, the
330 * frequency of nodes in bins follows a Poisson distribution
331 * (http://en.wikipedia.org/wiki/Poisson_distribution) with a
332 * parameter of about 0.5 on average, given the resizing threshold
333 * of 0.75, although with a large variance because of resizing
334 * granularity. Ignoring variance, the expected occurrences of
335 * list size k are (exp(-0.5) * pow(0.5, k) / factorial(k)). The
336 * first values are:
337 *
338 * 0: 0.60653066
339 * 1: 0.30326533
340 * 2: 0.07581633
341 * 3: 0.01263606
342 * 4: 0.00157952
343 * 5: 0.00015795
344 * 6: 0.00001316
345 * 7: 0.00000094
346 * 8: 0.00000006
347 * more: less than 1 in ten million
348 *
349 * Lock contention probability for two threads accessing distinct
350 * elements is roughly 1 / (8 * #elements) under random hashes.
351 *
352 * Actual hash code distributions encountered in practice
353 * sometimes deviate significantly from uniform randomness. This
354 * includes the case when N > (1<<30), so some keys MUST collide.
355 * Similarly for dumb or hostile usages in which multiple keys are
356 * designed to have identical hash codes or ones that differs only
357 * in masked-out high bits. So we use a secondary strategy that
358 * applies when the number of nodes in a bin exceeds a
359 * threshold. These TreeBins use a balanced tree to hold nodes (a
360 * specialized form of red-black trees), bounding search time to
361 * O(log N). Each search step in a TreeBin is at least twice as
362 * slow as in a regular list, but given that N cannot exceed
363 * (1<<64) (before running out of addresses) this bounds search
364 * steps, lock hold times, etc, to reasonable constants (roughly
365 * 100 nodes inspected per operation worst case) so long as keys
366 * are Comparable (which is very common -- String, Long, etc).
367 * TreeBin nodes (TreeNodes) also maintain the same "next"
368 * traversal pointers as regular nodes, so can be traversed in
369 * iterators in the same way.
370 *
371 * The table is resized when occupancy exceeds a percentage
372 * threshold (nominally, 0.75, but see below). Any thread
373 * noticing an overfull bin may assist in resizing after the
374 * initiating thread allocates and sets up the replacement array.
375 * However, rather than stalling, these other threads may proceed
376 * with insertions etc. The use of TreeBins shields us from the
377 * worst case effects of overfilling while resizes are in
378 * progress. Resizing proceeds by transferring bins, one by one,
379 * from the table to the next table. However, threads claim small
380 * blocks of indices to transfer (via field transferIndex) before
381 * doing so, reducing contention. A generation stamp in field
382 * sizeCtl ensures that resizings do not overlap. Because we are
383 * using power-of-two expansion, the elements from each bin must
384 * either stay at same index, or move with a power of two
385 * offset. We eliminate unnecessary node creation by catching
386 * cases where old nodes can be reused because their next fields
387 * won't change. On average, only about one-sixth of them need
388 * cloning when a table doubles. The nodes they replace will be
389 * garbage collectable as soon as they are no longer referenced by
390 * any reader thread that may be in the midst of concurrently
391 * traversing table. Upon transfer, the old table bin contains
392 * only a special forwarding node (with hash field "MOVED") that
393 * contains the next table as its key. On encountering a
394 * forwarding node, access and update operations restart, using
395 * the new table.
396 *
397 * Each bin transfer requires its bin lock, which can stall
398 * waiting for locks while resizing. However, because other
399 * threads can join in and help resize rather than contend for
400 * locks, average aggregate waits become shorter as resizing
401 * progresses. The transfer operation must also ensure that all
402 * accessible bins in both the old and new table are usable by any
403 * traversal. This is arranged in part by proceeding from the
404 * last bin (table.length - 1) up towards the first. Upon seeing
405 * a forwarding node, traversals (see class Traverser) arrange to
406 * move to the new table without revisiting nodes. To ensure that
407 * no intervening nodes are skipped even when moved out of order,
408 * a stack (see class TableStack) is created on first encounter of
409 * a forwarding node during a traversal, to maintain its place if
410 * later processing the current table. The need for these
411 * save/restore mechanics is relatively rare, but when one
412 * forwarding node is encountered, typically many more will be.
413 * So Traversers use a simple caching scheme to avoid creating so
414 * many new TableStack nodes. (Thanks to Peter Levart for
415 * suggesting use of a stack here.)
416 *
417 * The traversal scheme also applies to partial traversals of
418 * ranges of bins (via an alternate Traverser constructor)
419 * to support partitioned aggregate operations. Also, read-only
420 * operations give up if ever forwarded to a null table, which
421 * provides support for shutdown-style clearing, which is also not
422 * currently implemented.
423 *
424 * Lazy table initialization minimizes footprint until first use,
425 * and also avoids resizings when the first operation is from a
426 * putAll, constructor with map argument, or deserialization.
427 * These cases attempt to override the initial capacity settings,
428 * but harmlessly fail to take effect in cases of races.
429 *
430 * The element count is maintained using a specialization of
431 * LongAdder. We need to incorporate a specialization rather than
432 * just use a LongAdder in order to access implicit
433 * contention-sensing that leads to creation of multiple
434 * CounterCells. The counter mechanics avoid contention on
435 * updates but can encounter cache thrashing if read too
436 * frequently during concurrent access. To avoid reading so often,
437 * resizing under contention is attempted only upon adding to a
438 * bin already holding two or more nodes. Under uniform hash
439 * distributions, the probability of this occurring at threshold
440 * is around 13%, meaning that only about 1 in 8 puts check
441 * threshold (and after resizing, many fewer do so).
442 *
443 * TreeBins use a special form of comparison for search and
444 * related operations (which is the main reason we cannot use
445 * existing collections such as TreeMaps). TreeBins contain
446 * Comparable elements, but may contain others, as well as
447 * elements that are Comparable but not necessarily Comparable for
448 * the same T, so we cannot invoke compareTo among them. To handle
449 * this, the tree is ordered primarily by hash value, then by
450 * Comparable.compareTo order if applicable. On lookup at a node,
451 * if elements are not comparable or compare as 0 then both left
452 * and right children may need to be searched in the case of tied
453 * hash values. (This corresponds to the full list search that
454 * would be necessary if all elements were non-Comparable and had
455 * tied hashes.) On insertion, to keep a total ordering (or as
456 * close as is required here) across rebalancings, we compare
457 * classes and identityHashCodes as tie-breakers. The red-black
458 * balancing code is updated from pre-jdk-collections
459 * (http://gee.cs.oswego.edu/dl/classes/collections/RBCell.java)
460 * based in turn on Cormen, Leiserson, and Rivest "Introduction to
461 * Algorithms" (CLR).
462 *
463 * TreeBins also require an additional locking mechanism. While
464 * list traversal is always possible by readers even during
465 * updates, tree traversal is not, mainly because of tree-rotations
466 * that may change the root node and/or its linkages. TreeBins
467 * include a simple read-write lock mechanism parasitic on the
468 * main bin-synchronization strategy: Structural adjustments
469 * associated with an insertion or removal are already bin-locked
470 * (and so cannot conflict with other writers) but must wait for
471 * ongoing readers to finish. Since there can be only one such
472 * waiter, we use a simple scheme using a single "waiter" field to
473 * block writers. However, readers need never block. If the root
474 * lock is held, they proceed along the slow traversal path (via
475 * next-pointers) until the lock becomes available or the list is
476 * exhausted, whichever comes first. These cases are not fast, but
477 * maximize aggregate expected throughput.
478 *
479 * Maintaining API and serialization compatibility with previous
480 * versions of this class introduces several oddities. Mainly: We
481 * leave untouched but unused constructor arguments referring to
482 * concurrencyLevel. We accept a loadFactor constructor argument,
483 * but apply it only to initial table capacity (which is the only
484 * time that we can guarantee to honor it.) We also declare an
485 * unused "Segment" class that is instantiated in minimal form
486 * only when serializing.
487 *
488 * Also, solely for compatibility with previous versions of this
489 * class, it extends AbstractMap, even though all of its methods
490 * are overridden, so it is just useless baggage.
491 *
492 * This file is organized to make things a little easier to follow
493 * while reading than they might otherwise: First the main static
494 * declarations and utilities, then fields, then main public
495 * methods (with a few factorings of multiple public methods into
496 * internal ones), then sizing methods, trees, traversers, and
497 * bulk operations.
498 */
499
500 /* ---------------- Constants -------------- */
501
502 /**
503 * The largest possible table capacity. This value must be
504 * exactly 1<<30 to stay within Java array allocation and indexing
505 * bounds for power of two table sizes, and is further required
506 * because the top two bits of 32bit hash fields are used for
507 * control purposes.
508 */
509 private static final int MAXIMUM_CAPACITY = 1 << 30;
510
511 /**
512 * The default initial table capacity. Must be a power of 2
513 * (i.e., at least 1) and at most MAXIMUM_CAPACITY.
514 */
515 private static final int DEFAULT_CAPACITY = 16;
516
517 /**
518 * The largest possible (non-power of two) array size.
519 * Needed by toArray and related methods.
520 */
521 static final int MAX_ARRAY_SIZE = Integer.MAX_VALUE - 8;
522
523 /**
524 * The default concurrency level for this table. Unused but
525 * defined for compatibility with previous versions of this class.
526 */
527 private static final int DEFAULT_CONCURRENCY_LEVEL = 16;
528
529 /**
530 * The load factor for this table. Overrides of this value in
531 * constructors affect only the initial table capacity. The
532 * actual floating point value isn't normally used -- it is
533 * simpler to use expressions such as {@code n - (n >>> 2)} for
534 * the associated resizing threshold.
535 */
536 private static final float LOAD_FACTOR = 0.75f;
537
538 /**
539 * The bin count threshold for using a tree rather than list for a
540 * bin. Bins are converted to trees when adding an element to a
541 * bin with at least this many nodes. The value must be greater
542 * than 2, and should be at least 8 to mesh with assumptions in
543 * tree removal about conversion back to plain bins upon
544 * shrinkage.
545 */
546 static final int TREEIFY_THRESHOLD = 8;
547
548 /**
549 * The bin count threshold for untreeifying a (split) bin during a
550 * resize operation. Should be less than TREEIFY_THRESHOLD, and at
551 * most 6 to mesh with shrinkage detection under removal.
552 */
553 static final int UNTREEIFY_THRESHOLD = 6;
554
555 /**
556 * The smallest table capacity for which bins may be treeified.
557 * (Otherwise the table is resized if too many nodes in a bin.)
558 * The value should be at least 4 * TREEIFY_THRESHOLD to avoid
559 * conflicts between resizing and treeification thresholds.
560 */
561 static final int MIN_TREEIFY_CAPACITY = 64;
562
563 /**
564 * Minimum number of rebinnings per transfer step. Ranges are
565 * subdivided to allow multiple resizer threads. This value
566 * serves as a lower bound to avoid resizers encountering
567 * excessive memory contention. The value should be at least
568 * DEFAULT_CAPACITY.
569 */
570 private static final int MIN_TRANSFER_STRIDE = 16;
571
572 /**
573 * The number of bits used for generation stamp in sizeCtl.
574 * Must be at least 6 for 32bit arrays.
575 */
576 private static final int RESIZE_STAMP_BITS = 16;
577
578 /**
579 * The maximum number of threads that can help resize.
580 * Must fit in 32 - RESIZE_STAMP_BITS bits.
581 */
582 private static final int MAX_RESIZERS = (1 << (32 - RESIZE_STAMP_BITS)) - 1;
583
584 /**
585 * The bit shift for recording size stamp in sizeCtl.
586 */
587 private static final int RESIZE_STAMP_SHIFT = 32 - RESIZE_STAMP_BITS;
588
589 /*
590 * Encodings for Node hash fields. See above for explanation.
591 */
592 static final int MOVED = -1; // hash for forwarding nodes
593 static final int TREEBIN = -2; // hash for roots of trees
594 static final int RESERVED = -3; // hash for transient reservations
595 static final int HASH_BITS = 0x7fffffff; // usable bits of normal node hash
596
597 /** Number of CPUS, to place bounds on some sizings */
598 static final int NCPU = Runtime.getRuntime().availableProcessors();
599
600 /**
601 * Serialized pseudo-fields, provided only for jdk7 compatibility.
602 * @serialField segments Segment[]
603 * The segments, each of which is a specialized hash table.
604 * @serialField segmentMask int
605 * Mask value for indexing into segments. The upper bits of a
606 * key's hash code are used to choose the segment.
607 * @serialField segmentShift int
608 * Shift value for indexing within segments.
609 */
610 private static final ObjectStreamField[] serialPersistentFields = {
611 new ObjectStreamField("segments", Segment[].class),
612 new ObjectStreamField("segmentMask", Integer.TYPE),
613 new ObjectStreamField("segmentShift", Integer.TYPE),
614 };
615
616 /* ---------------- Nodes -------------- */
617
618 /**
619 * Key-value entry. This class is never exported out as a
620 * user-mutable Map.Entry (i.e., one supporting setValue; see
621 * MapEntry below), but can be used for read-only traversals used
622 * in bulk tasks. Subclasses of Node with a negative hash field
623 * are special, and contain null keys and values (but are never
624 * exported). Otherwise, keys and vals are never null.
625 */
626 static class Node<K,V> implements Map.Entry<K,V> {
627 final int hash;
628 final K key;
629 volatile V val;
630 volatile Node<K,V> next;
631
632 Node(int hash, K key, V val) {
633 this.hash = hash;
634 this.key = key;
635 this.val = val;
636 }
637
638 Node(int hash, K key, V val, Node<K,V> next) {
639 this(hash, key, val);
640 this.next = next;
641 }
642
643 public final K getKey() { return key; }
644 public final V getValue() { return val; }
645 public final int hashCode() { return key.hashCode() ^ val.hashCode(); }
646 public final String toString() {
647 return Helpers.mapEntryToString(key, val);
648 }
649 public final V setValue(V value) {
650 throw new UnsupportedOperationException();
651 }
652
653 public final boolean equals(Object o) {
654 Object k, v, u; Map.Entry<?,?> e;
655 return ((o instanceof Map.Entry) &&
656 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
657 (v = e.getValue()) != null &&
658 (k == key || k.equals(key)) &&
659 (v == (u = val) || v.equals(u)));
660 }
661
662 /**
663 * Virtualized support for map.get(); overridden in subclasses.
664 */
665 Node<K,V> find(int h, Object k) {
666 Node<K,V> e = this;
667 if (k != null) {
668 do {
669 K ek;
670 if (e.hash == h &&
671 ((ek = e.key) == k || (ek != null && k.equals(ek))))
672 return e;
673 } while ((e = e.next) != null);
674 }
675 return null;
676 }
677 }
678
679 /* ---------------- Static utilities -------------- */
680
681 /**
682 * Spreads (XORs) higher bits of hash to lower and also forces top
683 * bit to 0. Because the table uses power-of-two masking, sets of
684 * hashes that vary only in bits above the current mask will
685 * always collide. (Among known examples are sets of Float keys
686 * holding consecutive whole numbers in small tables.) So we
687 * apply a transform that spreads the impact of higher bits
688 * downward. There is a tradeoff between speed, utility, and
689 * quality of bit-spreading. Because many common sets of hashes
690 * are already reasonably distributed (so don't benefit from
691 * spreading), and because we use trees to handle large sets of
692 * collisions in bins, we just XOR some shifted bits in the
693 * cheapest possible way to reduce systematic lossage, as well as
694 * to incorporate impact of the highest bits that would otherwise
695 * never be used in index calculations because of table bounds.
696 */
697 static final int spread(int h) {
698 return (h ^ (h >>> 16)) & HASH_BITS;
699 }
700
701 /**
702 * Returns a power of two table size for the given desired capacity.
703 * See Hackers Delight, sec 3.2
704 */
705 private static final int tableSizeFor(int c) {
706 int n = c - 1;
707 n |= n >>> 1;
708 n |= n >>> 2;
709 n |= n >>> 4;
710 n |= n >>> 8;
711 n |= n >>> 16;
712 return (n < 0) ? 1 : (n >= MAXIMUM_CAPACITY) ? MAXIMUM_CAPACITY : n + 1;
713 }
714
715 /**
716 * Returns x's Class if it is of the form "class C implements
717 * Comparable<C>", else null.
718 */
719 static Class<?> comparableClassFor(Object x) {
720 if (x instanceof Comparable) {
721 Class<?> c; Type[] ts, as; Type t; ParameterizedType p;
722 if ((c = x.getClass()) == String.class) // bypass checks
723 return c;
724 if ((ts = c.getGenericInterfaces()) != null) {
725 for (int i = 0; i < ts.length; ++i) {
726 if (((t = ts[i]) instanceof ParameterizedType) &&
727 ((p = (ParameterizedType)t).getRawType() ==
728 Comparable.class) &&
729 (as = p.getActualTypeArguments()) != null &&
730 as.length == 1 && as[0] == c) // type arg is c
731 return c;
732 }
733 }
734 }
735 return null;
736 }
737
738 /**
739 * Returns k.compareTo(x) if x matches kc (k's screened comparable
740 * class), else 0.
741 */
742 @SuppressWarnings({"rawtypes","unchecked"}) // for cast to Comparable
743 static int compareComparables(Class<?> kc, Object k, Object x) {
744 return (x == null || x.getClass() != kc ? 0 :
745 ((Comparable)k).compareTo(x));
746 }
747
748 /* ---------------- Table element access -------------- */
749
750 /*
751 * Atomic access methods are used for table elements as well as
752 * elements of in-progress next table while resizing. All uses of
753 * the tab arguments must be null checked by callers. All callers
754 * also paranoically precheck that tab's length is not zero (or an
755 * equivalent check), thus ensuring that any index argument taking
756 * the form of a hash value anded with (length - 1) is a valid
757 * index. Note that, to be correct wrt arbitrary concurrency
758 * errors by users, these checks must operate on local variables,
759 * which accounts for some odd-looking inline assignments below.
760 * Note that calls to setTabAt always occur within locked regions,
761 * and so require only release ordering.
762 */
763
764 @SuppressWarnings("unchecked")
765 static final <K,V> Node<K,V> tabAt(Node<K,V>[] tab, int i) {
766 return (Node<K,V>)U.getObjectAcquire(tab, ((long)i << ASHIFT) + ABASE);
767 }
768
769 static final <K,V> boolean casTabAt(Node<K,V>[] tab, int i,
770 Node<K,V> c, Node<K,V> v) {
771 return U.compareAndSetObject(tab, ((long)i << ASHIFT) + ABASE, c, v);
772 }
773
774 static final <K,V> void setTabAt(Node<K,V>[] tab, int i, Node<K,V> v) {
775 U.putObjectRelease(tab, ((long)i << ASHIFT) + ABASE, v);
776 }
777
778 /* ---------------- Fields -------------- */
779
780 /**
781 * The array of bins. Lazily initialized upon first insertion.
782 * Size is always a power of two. Accessed directly by iterators.
783 */
784 transient volatile Node<K,V>[] table;
785
786 /**
787 * The next table to use; non-null only while resizing.
788 */
789 private transient volatile Node<K,V>[] nextTable;
790
791 /**
792 * Base counter value, used mainly when there is no contention,
793 * but also as a fallback during table initialization
794 * races. Updated via CAS.
795 */
796 private transient volatile long baseCount;
797
798 /**
799 * Table initialization and resizing control. When negative, the
800 * table is being initialized or resized: -1 for initialization,
801 * else -(1 + the number of active resizing threads). Otherwise,
802 * when table is null, holds the initial table size to use upon
803 * creation, or 0 for default. After initialization, holds the
804 * next element count value upon which to resize the table.
805 */
806 private transient volatile int sizeCtl;
807
808 /**
809 * The next table index (plus one) to split while resizing.
810 */
811 private transient volatile int transferIndex;
812
813 /**
814 * Spinlock (locked via CAS) used when resizing and/or creating CounterCells.
815 */
816 private transient volatile int cellsBusy;
817
818 /**
819 * Table of counter cells. When non-null, size is a power of 2.
820 */
821 private transient volatile CounterCell[] counterCells;
822
823 // views
824 private transient KeySetView<K,V> keySet;
825 private transient ValuesView<K,V> values;
826 private transient EntrySetView<K,V> entrySet;
827
828
829 /* ---------------- Public operations -------------- */
830
831 /**
832 * Creates a new, empty map with the default initial table size (16).
833 */
834 public ConcurrentHashMap() {
835 }
836
837 /**
838 * Creates a new, empty map with an initial table size
839 * accommodating the specified number of elements without the need
840 * to dynamically resize.
841 *
842 * @param initialCapacity The implementation performs internal
843 * sizing to accommodate this many elements.
844 * @throws IllegalArgumentException if the initial capacity of
845 * elements is negative
846 */
847 public ConcurrentHashMap(int initialCapacity) {
848 if (initialCapacity < 0)
849 throw new IllegalArgumentException();
850 int cap = ((initialCapacity >= (MAXIMUM_CAPACITY >>> 1)) ?
851 MAXIMUM_CAPACITY :
852 tableSizeFor(initialCapacity + (initialCapacity >>> 1) + 1));
853 this.sizeCtl = cap;
854 }
855
856 /**
857 * Creates a new map with the same mappings as the given map.
858 *
859 * @param m the map
860 */
861 public ConcurrentHashMap(Map<? extends K, ? extends V> m) {
862 this.sizeCtl = DEFAULT_CAPACITY;
863 putAll(m);
864 }
865
866 /**
867 * Creates a new, empty map with an initial table size based on
868 * the given number of elements ({@code initialCapacity}) and
869 * initial table density ({@code loadFactor}).
870 *
871 * @param initialCapacity the initial capacity. The implementation
872 * performs internal sizing to accommodate this many elements,
873 * given the specified load factor.
874 * @param loadFactor the load factor (table density) for
875 * establishing the initial table size
876 * @throws IllegalArgumentException if the initial capacity of
877 * elements is negative or the load factor is nonpositive
878 *
879 * @since 1.6
880 */
881 public ConcurrentHashMap(int initialCapacity, float loadFactor) {
882 this(initialCapacity, loadFactor, 1);
883 }
884
885 /**
886 * Creates a new, empty map with an initial table size based on
887 * the given number of elements ({@code initialCapacity}), table
888 * density ({@code loadFactor}), and number of concurrently
889 * updating threads ({@code concurrencyLevel}).
890 *
891 * @param initialCapacity the initial capacity. The implementation
892 * performs internal sizing to accommodate this many elements,
893 * given the specified load factor.
894 * @param loadFactor the load factor (table density) for
895 * establishing the initial table size
896 * @param concurrencyLevel the estimated number of concurrently
897 * updating threads. The implementation may use this value as
898 * a sizing hint.
899 * @throws IllegalArgumentException if the initial capacity is
900 * negative or the load factor or concurrencyLevel are
901 * nonpositive
902 */
903 public ConcurrentHashMap(int initialCapacity,
904 float loadFactor, int concurrencyLevel) {
905 if (!(loadFactor > 0.0f) || initialCapacity < 0 || concurrencyLevel <= 0)
906 throw new IllegalArgumentException();
907 if (initialCapacity < concurrencyLevel) // Use at least as many bins
908 initialCapacity = concurrencyLevel; // as estimated threads
909 long size = (long)(1.0 + (long)initialCapacity / loadFactor);
910 int cap = (size >= (long)MAXIMUM_CAPACITY) ?
911 MAXIMUM_CAPACITY : tableSizeFor((int)size);
912 this.sizeCtl = cap;
913 }
914
915 // Original (since JDK1.2) Map methods
916
917 /**
918 * {@inheritDoc}
919 */
920 public int size() {
921 long n = sumCount();
922 return ((n < 0L) ? 0 :
923 (n > (long)Integer.MAX_VALUE) ? Integer.MAX_VALUE :
924 (int)n);
925 }
926
927 /**
928 * {@inheritDoc}
929 */
930 public boolean isEmpty() {
931 return sumCount() <= 0L; // ignore transient negative values
932 }
933
934 /**
935 * Returns the value to which the specified key is mapped,
936 * or {@code null} if this map contains no mapping for the key.
937 *
938 * <p>More formally, if this map contains a mapping from a key
939 * {@code k} to a value {@code v} such that {@code key.equals(k)},
940 * then this method returns {@code v}; otherwise it returns
941 * {@code null}. (There can be at most one such mapping.)
942 *
943 * @throws NullPointerException if the specified key is null
944 */
945 public V get(Object key) {
946 Node<K,V>[] tab; Node<K,V> e, p; int n, eh; K ek;
947 int h = spread(key.hashCode());
948 if ((tab = table) != null && (n = tab.length) > 0 &&
949 (e = tabAt(tab, (n - 1) & h)) != null) {
950 if ((eh = e.hash) == h) {
951 if ((ek = e.key) == key || (ek != null && key.equals(ek)))
952 return e.val;
953 }
954 else if (eh < 0)
955 return (p = e.find(h, key)) != null ? p.val : null;
956 while ((e = e.next) != null) {
957 if (e.hash == h &&
958 ((ek = e.key) == key || (ek != null && key.equals(ek))))
959 return e.val;
960 }
961 }
962 return null;
963 }
964
965 /**
966 * Tests if the specified object is a key in this table.
967 *
968 * @param key possible key
969 * @return {@code true} if and only if the specified object
970 * is a key in this table, as determined by the
971 * {@code equals} method; {@code false} otherwise
972 * @throws NullPointerException if the specified key is null
973 */
974 public boolean containsKey(Object key) {
975 return get(key) != null;
976 }
977
978 /**
979 * Returns {@code true} if this map maps one or more keys to the
980 * specified value. Note: This method may require a full traversal
981 * of the map, and is much slower than method {@code containsKey}.
982 *
983 * @param value value whose presence in this map is to be tested
984 * @return {@code true} if this map maps one or more keys to the
985 * specified value
986 * @throws NullPointerException if the specified value is null
987 */
988 public boolean containsValue(Object value) {
989 if (value == null)
990 throw new NullPointerException();
991 Node<K,V>[] t;
992 if ((t = table) != null) {
993 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
994 for (Node<K,V> p; (p = it.advance()) != null; ) {
995 V v;
996 if ((v = p.val) == value || (v != null && value.equals(v)))
997 return true;
998 }
999 }
1000 return false;
1001 }
1002
1003 /**
1004 * Maps the specified key to the specified value in this table.
1005 * Neither the key nor the value can be null.
1006 *
1007 * <p>The value can be retrieved by calling the {@code get} method
1008 * with a key that is equal to the original key.
1009 *
1010 * @param key key with which the specified value is to be associated
1011 * @param value value to be associated with the specified key
1012 * @return the previous value associated with {@code key}, or
1013 * {@code null} if there was no mapping for {@code key}
1014 * @throws NullPointerException if the specified key or value is null
1015 */
1016 public V put(K key, V value) {
1017 return putVal(key, value, false);
1018 }
1019
1020 /** Implementation for put and putIfAbsent */
1021 final V putVal(K key, V value, boolean onlyIfAbsent) {
1022 if (key == null || value == null) throw new NullPointerException();
1023 int hash = spread(key.hashCode());
1024 int binCount = 0;
1025 for (Node<K,V>[] tab = table;;) {
1026 Node<K,V> f; int n, i, fh; K fk; V fv;
1027 if (tab == null || (n = tab.length) == 0)
1028 tab = initTable();
1029 else if ((f = tabAt(tab, i = (n - 1) & hash)) == null) {
1030 if (casTabAt(tab, i, null, new Node<K,V>(hash, key, value)))
1031 break; // no lock when adding to empty bin
1032 }
1033 else if ((fh = f.hash) == MOVED)
1034 tab = helpTransfer(tab, f);
1035 else if (onlyIfAbsent // check first node without acquiring lock
1036 && fh == hash
1037 && ((fk = f.key) == key || (fk != null && key.equals(fk)))
1038 && (fv = f.val) != null)
1039 return fv;
1040 else {
1041 V oldVal = null;
1042 synchronized (f) {
1043 if (tabAt(tab, i) == f) {
1044 if (fh >= 0) {
1045 binCount = 1;
1046 for (Node<K,V> e = f;; ++binCount) {
1047 K ek;
1048 if (e.hash == hash &&
1049 ((ek = e.key) == key ||
1050 (ek != null && key.equals(ek)))) {
1051 oldVal = e.val;
1052 if (!onlyIfAbsent)
1053 e.val = value;
1054 break;
1055 }
1056 Node<K,V> pred = e;
1057 if ((e = e.next) == null) {
1058 pred.next = new Node<K,V>(hash, key, value);
1059 break;
1060 }
1061 }
1062 }
1063 else if (f instanceof TreeBin) {
1064 Node<K,V> p;
1065 binCount = 2;
1066 if ((p = ((TreeBin<K,V>)f).putTreeVal(hash, key,
1067 value)) != null) {
1068 oldVal = p.val;
1069 if (!onlyIfAbsent)
1070 p.val = value;
1071 }
1072 }
1073 else if (f instanceof ReservationNode)
1074 throw new IllegalStateException("Recursive update");
1075 }
1076 }
1077 if (binCount != 0) {
1078 if (binCount >= TREEIFY_THRESHOLD)
1079 treeifyBin(tab, i);
1080 if (oldVal != null)
1081 return oldVal;
1082 break;
1083 }
1084 }
1085 }
1086 addCount(1L, binCount);
1087 return null;
1088 }
1089
1090 /**
1091 * Copies all of the mappings from the specified map to this one.
1092 * These mappings replace any mappings that this map had for any of the
1093 * keys currently in the specified map.
1094 *
1095 * @param m mappings to be stored in this map
1096 */
1097 public void putAll(Map<? extends K, ? extends V> m) {
1098 tryPresize(m.size());
1099 for (Map.Entry<? extends K, ? extends V> e : m.entrySet())
1100 putVal(e.getKey(), e.getValue(), false);
1101 }
1102
1103 /**
1104 * Removes the key (and its corresponding value) from this map.
1105 * This method does nothing if the key is not in the map.
1106 *
1107 * @param key the key that needs to be removed
1108 * @return the previous value associated with {@code key}, or
1109 * {@code null} if there was no mapping for {@code key}
1110 * @throws NullPointerException if the specified key is null
1111 */
1112 public V remove(Object key) {
1113 return replaceNode(key, null, null);
1114 }
1115
1116 /**
1117 * Implementation for the four public remove/replace methods:
1118 * Replaces node value with v, conditional upon match of cv if
1119 * non-null. If resulting value is null, delete.
1120 */
1121 final V replaceNode(Object key, V value, Object cv) {
1122 int hash = spread(key.hashCode());
1123 for (Node<K,V>[] tab = table;;) {
1124 Node<K,V> f; int n, i, fh;
1125 if (tab == null || (n = tab.length) == 0 ||
1126 (f = tabAt(tab, i = (n - 1) & hash)) == null)
1127 break;
1128 else if ((fh = f.hash) == MOVED)
1129 tab = helpTransfer(tab, f);
1130 else {
1131 V oldVal = null;
1132 boolean validated = false;
1133 synchronized (f) {
1134 if (tabAt(tab, i) == f) {
1135 if (fh >= 0) {
1136 validated = true;
1137 for (Node<K,V> e = f, pred = null;;) {
1138 K ek;
1139 if (e.hash == hash &&
1140 ((ek = e.key) == key ||
1141 (ek != null && key.equals(ek)))) {
1142 V ev = e.val;
1143 if (cv == null || cv == ev ||
1144 (ev != null && cv.equals(ev))) {
1145 oldVal = ev;
1146 if (value != null)
1147 e.val = value;
1148 else if (pred != null)
1149 pred.next = e.next;
1150 else
1151 setTabAt(tab, i, e.next);
1152 }
1153 break;
1154 }
1155 pred = e;
1156 if ((e = e.next) == null)
1157 break;
1158 }
1159 }
1160 else if (f instanceof TreeBin) {
1161 validated = true;
1162 TreeBin<K,V> t = (TreeBin<K,V>)f;
1163 TreeNode<K,V> r, p;
1164 if ((r = t.root) != null &&
1165 (p = r.findTreeNode(hash, key, null)) != null) {
1166 V pv = p.val;
1167 if (cv == null || cv == pv ||
1168 (pv != null && cv.equals(pv))) {
1169 oldVal = pv;
1170 if (value != null)
1171 p.val = value;
1172 else if (t.removeTreeNode(p))
1173 setTabAt(tab, i, untreeify(t.first));
1174 }
1175 }
1176 }
1177 else if (f instanceof ReservationNode)
1178 throw new IllegalStateException("Recursive update");
1179 }
1180 }
1181 if (validated) {
1182 if (oldVal != null) {
1183 if (value == null)
1184 addCount(-1L, -1);
1185 return oldVal;
1186 }
1187 break;
1188 }
1189 }
1190 }
1191 return null;
1192 }
1193
1194 /**
1195 * Removes all of the mappings from this map.
1196 */
1197 public void clear() {
1198 long delta = 0L; // negative number of deletions
1199 int i = 0;
1200 Node<K,V>[] tab = table;
1201 while (tab != null && i < tab.length) {
1202 int fh;
1203 Node<K,V> f = tabAt(tab, i);
1204 if (f == null)
1205 ++i;
1206 else if ((fh = f.hash) == MOVED) {
1207 tab = helpTransfer(tab, f);
1208 i = 0; // restart
1209 }
1210 else {
1211 synchronized (f) {
1212 if (tabAt(tab, i) == f) {
1213 Node<K,V> p = (fh >= 0 ? f :
1214 (f instanceof TreeBin) ?
1215 ((TreeBin<K,V>)f).first : null);
1216 while (p != null) {
1217 --delta;
1218 p = p.next;
1219 }
1220 setTabAt(tab, i++, null);
1221 }
1222 }
1223 }
1224 }
1225 if (delta != 0L)
1226 addCount(delta, -1);
1227 }
1228
1229 /**
1230 * Returns a {@link Set} view of the keys contained in this map.
1231 * The set is backed by the map, so changes to the map are
1232 * reflected in the set, and vice-versa. The set supports element
1233 * removal, which removes the corresponding mapping from this map,
1234 * via the {@code Iterator.remove}, {@code Set.remove},
1235 * {@code removeAll}, {@code retainAll}, and {@code clear}
1236 * operations. It does not support the {@code add} or
1237 * {@code addAll} operations.
1238 *
1239 * <p>The view's iterators and spliterators are
1240 * <a href="package-summary.html#Weakly"><i>weakly consistent</i></a>.
1241 *
1242 * <p>The view's {@code spliterator} reports {@link Spliterator#CONCURRENT},
1243 * {@link Spliterator#DISTINCT}, and {@link Spliterator#NONNULL}.
1244 *
1245 * @return the set view
1246 */
1247 public KeySetView<K,V> keySet() {
1248 KeySetView<K,V> ks;
1249 if ((ks = keySet) != null) return ks;
1250 return keySet = new KeySetView<K,V>(this, null);
1251 }
1252
1253 /**
1254 * Returns a {@link Collection} view of the values contained in this map.
1255 * The collection is backed by the map, so changes to the map are
1256 * reflected in the collection, and vice-versa. The collection
1257 * supports element removal, which removes the corresponding
1258 * mapping from this map, via the {@code Iterator.remove},
1259 * {@code Collection.remove}, {@code removeAll},
1260 * {@code retainAll}, and {@code clear} operations. It does not
1261 * support the {@code add} or {@code addAll} operations.
1262 *
1263 * <p>The view's iterators and spliterators are
1264 * <a href="package-summary.html#Weakly"><i>weakly consistent</i></a>.
1265 *
1266 * <p>The view's {@code spliterator} reports {@link Spliterator#CONCURRENT}
1267 * and {@link Spliterator#NONNULL}.
1268 *
1269 * @return the collection view
1270 */
1271 public Collection<V> values() {
1272 ValuesView<K,V> vs;
1273 if ((vs = values) != null) return vs;
1274 return values = new ValuesView<K,V>(this);
1275 }
1276
1277 /**
1278 * Returns a {@link Set} view of the mappings contained in this map.
1279 * The set is backed by the map, so changes to the map are
1280 * reflected in the set, and vice-versa. The set supports element
1281 * removal, which removes the corresponding mapping from the map,
1282 * via the {@code Iterator.remove}, {@code Set.remove},
1283 * {@code removeAll}, {@code retainAll}, and {@code clear}
1284 * operations.
1285 *
1286 * <p>The view's iterators and spliterators are
1287 * <a href="package-summary.html#Weakly"><i>weakly consistent</i></a>.
1288 *
1289 * <p>The view's {@code spliterator} reports {@link Spliterator#CONCURRENT},
1290 * {@link Spliterator#DISTINCT}, and {@link Spliterator#NONNULL}.
1291 *
1292 * @return the set view
1293 */
1294 public Set<Map.Entry<K,V>> entrySet() {
1295 EntrySetView<K,V> es;
1296 if ((es = entrySet) != null) return es;
1297 return entrySet = new EntrySetView<K,V>(this);
1298 }
1299
1300 /**
1301 * Returns the hash code value for this {@link Map}, i.e.,
1302 * the sum of, for each key-value pair in the map,
1303 * {@code key.hashCode() ^ value.hashCode()}.
1304 *
1305 * @return the hash code value for this map
1306 */
1307 public int hashCode() {
1308 int h = 0;
1309 Node<K,V>[] t;
1310 if ((t = table) != null) {
1311 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1312 for (Node<K,V> p; (p = it.advance()) != null; )
1313 h += p.key.hashCode() ^ p.val.hashCode();
1314 }
1315 return h;
1316 }
1317
1318 /**
1319 * Returns a string representation of this map. The string
1320 * representation consists of a list of key-value mappings (in no
1321 * particular order) enclosed in braces ("{@code {}}"). Adjacent
1322 * mappings are separated by the characters {@code ", "} (comma
1323 * and space). Each key-value mapping is rendered as the key
1324 * followed by an equals sign ("{@code =}") followed by the
1325 * associated value.
1326 *
1327 * @return a string representation of this map
1328 */
1329 public String toString() {
1330 Node<K,V>[] t;
1331 int f = (t = table) == null ? 0 : t.length;
1332 Traverser<K,V> it = new Traverser<K,V>(t, f, 0, f);
1333 StringBuilder sb = new StringBuilder();
1334 sb.append('{');
1335 Node<K,V> p;
1336 if ((p = it.advance()) != null) {
1337 for (;;) {
1338 K k = p.key;
1339 V v = p.val;
1340 sb.append(k == this ? "(this Map)" : k);
1341 sb.append('=');
1342 sb.append(v == this ? "(this Map)" : v);
1343 if ((p = it.advance()) == null)
1344 break;
1345 sb.append(',').append(' ');
1346 }
1347 }
1348 return sb.append('}').toString();
1349 }
1350
1351 /**
1352 * Compares the specified object with this map for equality.
1353 * Returns {@code true} if the given object is a map with the same
1354 * mappings as this map. This operation may return misleading
1355 * results if either map is concurrently modified during execution
1356 * of this method.
1357 *
1358 * @param o object to be compared for equality with this map
1359 * @return {@code true} if the specified object is equal to this map
1360 */
1361 public boolean equals(Object o) {
1362 if (o != this) {
1363 if (!(o instanceof Map))
1364 return false;
1365 Map<?,?> m = (Map<?,?>) o;
1366 Node<K,V>[] t;
1367 int f = (t = table) == null ? 0 : t.length;
1368 Traverser<K,V> it = new Traverser<K,V>(t, f, 0, f);
1369 for (Node<K,V> p; (p = it.advance()) != null; ) {
1370 V val = p.val;
1371 Object v = m.get(p.key);
1372 if (v == null || (v != val && !v.equals(val)))
1373 return false;
1374 }
1375 for (Map.Entry<?,?> e : m.entrySet()) {
1376 Object mk, mv, v;
1377 if ((mk = e.getKey()) == null ||
1378 (mv = e.getValue()) == null ||
1379 (v = get(mk)) == null ||
1380 (mv != v && !mv.equals(v)))
1381 return false;
1382 }
1383 }
1384 return true;
1385 }
1386
1387 /**
1388 * Stripped-down version of helper class used in previous version,
1389 * declared for the sake of serialization compatibility.
1390 */
1391 static class Segment<K,V> extends ReentrantLock implements Serializable {
1392 private static final long serialVersionUID = 2249069246763182397L;
1393 final float loadFactor;
1394 Segment(float lf) { this.loadFactor = lf; }
1395 }
1396
1397 /**
1398 * Saves the state of the {@code ConcurrentHashMap} instance to a
1399 * stream (i.e., serializes it).
1400 * @param s the stream
1401 * @throws java.io.IOException if an I/O error occurs
1402 * @serialData
1403 * the serialized fields, followed by the key (Object) and value
1404 * (Object) for each key-value mapping, followed by a null pair.
1405 * The key-value mappings are emitted in no particular order.
1406 */
1407 private void writeObject(java.io.ObjectOutputStream s)
1408 throws java.io.IOException {
1409 // For serialization compatibility
1410 // Emulate segment calculation from previous version of this class
1411 int sshift = 0;
1412 int ssize = 1;
1413 while (ssize < DEFAULT_CONCURRENCY_LEVEL) {
1414 ++sshift;
1415 ssize <<= 1;
1416 }
1417 int segmentShift = 32 - sshift;
1418 int segmentMask = ssize - 1;
1419 @SuppressWarnings("unchecked")
1420 Segment<K,V>[] segments = (Segment<K,V>[])
1421 new Segment<?,?>[DEFAULT_CONCURRENCY_LEVEL];
1422 for (int i = 0; i < segments.length; ++i)
1423 segments[i] = new Segment<K,V>(LOAD_FACTOR);
1424 java.io.ObjectOutputStream.PutField streamFields = s.putFields();
1425 streamFields.put("segments", segments);
1426 streamFields.put("segmentShift", segmentShift);
1427 streamFields.put("segmentMask", segmentMask);
1428 s.writeFields();
1429
1430 Node<K,V>[] t;
1431 if ((t = table) != null) {
1432 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1433 for (Node<K,V> p; (p = it.advance()) != null; ) {
1434 s.writeObject(p.key);
1435 s.writeObject(p.val);
1436 }
1437 }
1438 s.writeObject(null);
1439 s.writeObject(null);
1440 }
1441
1442 /**
1443 * Reconstitutes the instance from a stream (that is, deserializes it).
1444 * @param s the stream
1445 * @throws ClassNotFoundException if the class of a serialized object
1446 * could not be found
1447 * @throws java.io.IOException if an I/O error occurs
1448 */
1449 private void readObject(java.io.ObjectInputStream s)
1450 throws java.io.IOException, ClassNotFoundException {
1451 /*
1452 * To improve performance in typical cases, we create nodes
1453 * while reading, then place in table once size is known.
1454 * However, we must also validate uniqueness and deal with
1455 * overpopulated bins while doing so, which requires
1456 * specialized versions of putVal mechanics.
1457 */
1458 sizeCtl = -1; // force exclusion for table construction
1459 s.defaultReadObject();
1460 long size = 0L;
1461 Node<K,V> p = null;
1462 for (;;) {
1463 @SuppressWarnings("unchecked")
1464 K k = (K) s.readObject();
1465 @SuppressWarnings("unchecked")
1466 V v = (V) s.readObject();
1467 if (k != null && v != null) {
1468 p = new Node<K,V>(spread(k.hashCode()), k, v, p);
1469 ++size;
1470 }
1471 else
1472 break;
1473 }
1474 if (size == 0L)
1475 sizeCtl = 0;
1476 else {
1477 int n;
1478 if (size >= (long)(MAXIMUM_CAPACITY >>> 1))
1479 n = MAXIMUM_CAPACITY;
1480 else {
1481 int sz = (int)size;
1482 n = tableSizeFor(sz + (sz >>> 1) + 1);
1483 }
1484 @SuppressWarnings("unchecked")
1485 Node<K,V>[] tab = (Node<K,V>[])new Node<?,?>[n];
1486 int mask = n - 1;
1487 long added = 0L;
1488 while (p != null) {
1489 boolean insertAtFront;
1490 Node<K,V> next = p.next, first;
1491 int h = p.hash, j = h & mask;
1492 if ((first = tabAt(tab, j)) == null)
1493 insertAtFront = true;
1494 else {
1495 K k = p.key;
1496 if (first.hash < 0) {
1497 TreeBin<K,V> t = (TreeBin<K,V>)first;
1498 if (t.putTreeVal(h, k, p.val) == null)
1499 ++added;
1500 insertAtFront = false;
1501 }
1502 else {
1503 int binCount = 0;
1504 insertAtFront = true;
1505 Node<K,V> q; K qk;
1506 for (q = first; q != null; q = q.next) {
1507 if (q.hash == h &&
1508 ((qk = q.key) == k ||
1509 (qk != null && k.equals(qk)))) {
1510 insertAtFront = false;
1511 break;
1512 }
1513 ++binCount;
1514 }
1515 if (insertAtFront && binCount >= TREEIFY_THRESHOLD) {
1516 insertAtFront = false;
1517 ++added;
1518 p.next = first;
1519 TreeNode<K,V> hd = null, tl = null;
1520 for (q = p; q != null; q = q.next) {
1521 TreeNode<K,V> t = new TreeNode<K,V>
1522 (q.hash, q.key, q.val, null, null);
1523 if ((t.prev = tl) == null)
1524 hd = t;
1525 else
1526 tl.next = t;
1527 tl = t;
1528 }
1529 setTabAt(tab, j, new TreeBin<K,V>(hd));
1530 }
1531 }
1532 }
1533 if (insertAtFront) {
1534 ++added;
1535 p.next = first;
1536 setTabAt(tab, j, p);
1537 }
1538 p = next;
1539 }
1540 table = tab;
1541 sizeCtl = n - (n >>> 2);
1542 baseCount = added;
1543 }
1544 }
1545
1546 // ConcurrentMap methods
1547
1548 /**
1549 * {@inheritDoc}
1550 *
1551 * @return the previous value associated with the specified key,
1552 * or {@code null} if there was no mapping for the key
1553 * @throws NullPointerException if the specified key or value is null
1554 */
1555 public V putIfAbsent(K key, V value) {
1556 return putVal(key, value, true);
1557 }
1558
1559 /**
1560 * {@inheritDoc}
1561 *
1562 * @throws NullPointerException if the specified key is null
1563 */
1564 public boolean remove(Object key, Object value) {
1565 if (key == null)
1566 throw new NullPointerException();
1567 return value != null && replaceNode(key, null, value) != null;
1568 }
1569
1570 /**
1571 * {@inheritDoc}
1572 *
1573 * @throws NullPointerException if any of the arguments are null
1574 */
1575 public boolean replace(K key, V oldValue, V newValue) {
1576 if (key == null || oldValue == null || newValue == null)
1577 throw new NullPointerException();
1578 return replaceNode(key, newValue, oldValue) != null;
1579 }
1580
1581 /**
1582 * {@inheritDoc}
1583 *
1584 * @return the previous value associated with the specified key,
1585 * or {@code null} if there was no mapping for the key
1586 * @throws NullPointerException if the specified key or value is null
1587 */
1588 public V replace(K key, V value) {
1589 if (key == null || value == null)
1590 throw new NullPointerException();
1591 return replaceNode(key, value, null);
1592 }
1593
1594 // Overrides of JDK8+ Map extension method defaults
1595
1596 /**
1597 * Returns the value to which the specified key is mapped, or the
1598 * given default value if this map contains no mapping for the
1599 * key.
1600 *
1601 * @param key the key whose associated value is to be returned
1602 * @param defaultValue the value to return if this map contains
1603 * no mapping for the given key
1604 * @return the mapping for the key, if present; else the default value
1605 * @throws NullPointerException if the specified key is null
1606 */
1607 public V getOrDefault(Object key, V defaultValue) {
1608 V v;
1609 return (v = get(key)) == null ? defaultValue : v;
1610 }
1611
1612 public void forEach(BiConsumer<? super K, ? super V> action) {
1613 if (action == null) throw new NullPointerException();
1614 Node<K,V>[] t;
1615 if ((t = table) != null) {
1616 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1617 for (Node<K,V> p; (p = it.advance()) != null; ) {
1618 action.accept(p.key, p.val);
1619 }
1620 }
1621 }
1622
1623 public void replaceAll(BiFunction<? super K, ? super V, ? extends V> function) {
1624 if (function == null) throw new NullPointerException();
1625 Node<K,V>[] t;
1626 if ((t = table) != null) {
1627 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1628 for (Node<K,V> p; (p = it.advance()) != null; ) {
1629 V oldValue = p.val;
1630 for (K key = p.key;;) {
1631 V newValue = function.apply(key, oldValue);
1632 if (newValue == null)
1633 throw new NullPointerException();
1634 if (replaceNode(key, newValue, oldValue) != null ||
1635 (oldValue = get(key)) == null)
1636 break;
1637 }
1638 }
1639 }
1640 }
1641
1642 /**
1643 * Helper method for EntrySetView.removeIf.
1644 */
1645 boolean removeEntryIf(Predicate<? super Entry<K,V>> function) {
1646 if (function == null) throw new NullPointerException();
1647 Node<K,V>[] t;
1648 boolean removed = false;
1649 if ((t = table) != null) {
1650 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1651 for (Node<K,V> p; (p = it.advance()) != null; ) {
1652 K k = p.key;
1653 V v = p.val;
1654 Map.Entry<K,V> e = new AbstractMap.SimpleImmutableEntry<>(k, v);
1655 if (function.test(e) && replaceNode(k, null, v) != null)
1656 removed = true;
1657 }
1658 }
1659 return removed;
1660 }
1661
1662 /**
1663 * Helper method for ValuesView.removeIf.
1664 */
1665 boolean removeValueIf(Predicate<? super V> function) {
1666 if (function == null) throw new NullPointerException();
1667 Node<K,V>[] t;
1668 boolean removed = false;
1669 if ((t = table) != null) {
1670 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1671 for (Node<K,V> p; (p = it.advance()) != null; ) {
1672 K k = p.key;
1673 V v = p.val;
1674 if (function.test(v) && replaceNode(k, null, v) != null)
1675 removed = true;
1676 }
1677 }
1678 return removed;
1679 }
1680
1681 /**
1682 * If the specified key is not already associated with a value,
1683 * attempts to compute its value using the given mapping function
1684 * and enters it into this map unless {@code null}. The entire
1685 * method invocation is performed atomically, so the function is
1686 * applied at most once per key. Some attempted update operations
1687 * on this map by other threads may be blocked while computation
1688 * is in progress, so the computation should be short and simple,
1689 * and must not attempt to update any other mappings of this map.
1690 *
1691 * @param key key with which the specified value is to be associated
1692 * @param mappingFunction the function to compute a value
1693 * @return the current (existing or computed) value associated with
1694 * the specified key, or null if the computed value is null
1695 * @throws NullPointerException if the specified key or mappingFunction
1696 * is null
1697 * @throws IllegalStateException if the computation detectably
1698 * attempts a recursive update to this map that would
1699 * otherwise never complete
1700 * @throws RuntimeException or Error if the mappingFunction does so,
1701 * in which case the mapping is left unestablished
1702 */
1703 public V computeIfAbsent(K key, Function<? super K, ? extends V> mappingFunction) {
1704 if (key == null || mappingFunction == null)
1705 throw new NullPointerException();
1706 int h = spread(key.hashCode());
1707 V val = null;
1708 int binCount = 0;
1709 for (Node<K,V>[] tab = table;;) {
1710 Node<K,V> f; int n, i, fh; K fk; V fv;
1711 if (tab == null || (n = tab.length) == 0)
1712 tab = initTable();
1713 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
1714 Node<K,V> r = new ReservationNode<K,V>();
1715 synchronized (r) {
1716 if (casTabAt(tab, i, null, r)) {
1717 binCount = 1;
1718 Node<K,V> node = null;
1719 try {
1720 if ((val = mappingFunction.apply(key)) != null)
1721 node = new Node<K,V>(h, key, val);
1722 } finally {
1723 setTabAt(tab, i, node);
1724 }
1725 }
1726 }
1727 if (binCount != 0)
1728 break;
1729 }
1730 else if ((fh = f.hash) == MOVED)
1731 tab = helpTransfer(tab, f);
1732 else if (fh == h // check first node without acquiring lock
1733 && ((fk = f.key) == key || (fk != null && key.equals(fk)))
1734 && (fv = f.val) != null)
1735 return fv;
1736 else {
1737 boolean added = false;
1738 synchronized (f) {
1739 if (tabAt(tab, i) == f) {
1740 if (fh >= 0) {
1741 binCount = 1;
1742 for (Node<K,V> e = f;; ++binCount) {
1743 K ek;
1744 if (e.hash == h &&
1745 ((ek = e.key) == key ||
1746 (ek != null && key.equals(ek)))) {
1747 val = e.val;
1748 break;
1749 }
1750 Node<K,V> pred = e;
1751 if ((e = e.next) == null) {
1752 if ((val = mappingFunction.apply(key)) != null) {
1753 if (pred.next != null)
1754 throw new IllegalStateException("Recursive update");
1755 added = true;
1756 pred.next = new Node<K,V>(h, key, val);
1757 }
1758 break;
1759 }
1760 }
1761 }
1762 else if (f instanceof TreeBin) {
1763 binCount = 2;
1764 TreeBin<K,V> t = (TreeBin<K,V>)f;
1765 TreeNode<K,V> r, p;
1766 if ((r = t.root) != null &&
1767 (p = r.findTreeNode(h, key, null)) != null)
1768 val = p.val;
1769 else if ((val = mappingFunction.apply(key)) != null) {
1770 added = true;
1771 t.putTreeVal(h, key, val);
1772 }
1773 }
1774 else if (f instanceof ReservationNode)
1775 throw new IllegalStateException("Recursive update");
1776 }
1777 }
1778 if (binCount != 0) {
1779 if (binCount >= TREEIFY_THRESHOLD)
1780 treeifyBin(tab, i);
1781 if (!added)
1782 return val;
1783 break;
1784 }
1785 }
1786 }
1787 if (val != null)
1788 addCount(1L, binCount);
1789 return val;
1790 }
1791
1792 /**
1793 * If the value for the specified key is present, attempts to
1794 * compute a new mapping given the key and its current mapped
1795 * value. The entire method invocation is performed atomically.
1796 * Some attempted update operations on this map by other threads
1797 * may be blocked while computation is in progress, so the
1798 * computation should be short and simple, and must not attempt to
1799 * update any other mappings of this map.
1800 *
1801 * @param key key with which a value may be associated
1802 * @param remappingFunction the function to compute a value
1803 * @return the new value associated with the specified key, or null if none
1804 * @throws NullPointerException if the specified key or remappingFunction
1805 * is null
1806 * @throws IllegalStateException if the computation detectably
1807 * attempts a recursive update to this map that would
1808 * otherwise never complete
1809 * @throws RuntimeException or Error if the remappingFunction does so,
1810 * in which case the mapping is unchanged
1811 */
1812 public V computeIfPresent(K key, BiFunction<? super K, ? super V, ? extends V> remappingFunction) {
1813 if (key == null || remappingFunction == null)
1814 throw new NullPointerException();
1815 int h = spread(key.hashCode());
1816 V val = null;
1817 int delta = 0;
1818 int binCount = 0;
1819 for (Node<K,V>[] tab = table;;) {
1820 Node<K,V> f; int n, i, fh;
1821 if (tab == null || (n = tab.length) == 0)
1822 tab = initTable();
1823 else if ((f = tabAt(tab, i = (n - 1) & h)) == null)
1824 break;
1825 else if ((fh = f.hash) == MOVED)
1826 tab = helpTransfer(tab, f);
1827 else {
1828 synchronized (f) {
1829 if (tabAt(tab, i) == f) {
1830 if (fh >= 0) {
1831 binCount = 1;
1832 for (Node<K,V> e = f, pred = null;; ++binCount) {
1833 K ek;
1834 if (e.hash == h &&
1835 ((ek = e.key) == key ||
1836 (ek != null && key.equals(ek)))) {
1837 val = remappingFunction.apply(key, e.val);
1838 if (val != null)
1839 e.val = val;
1840 else {
1841 delta = -1;
1842 Node<K,V> en = e.next;
1843 if (pred != null)
1844 pred.next = en;
1845 else
1846 setTabAt(tab, i, en);
1847 }
1848 break;
1849 }
1850 pred = e;
1851 if ((e = e.next) == null)
1852 break;
1853 }
1854 }
1855 else if (f instanceof TreeBin) {
1856 binCount = 2;
1857 TreeBin<K,V> t = (TreeBin<K,V>)f;
1858 TreeNode<K,V> r, p;
1859 if ((r = t.root) != null &&
1860 (p = r.findTreeNode(h, key, null)) != null) {
1861 val = remappingFunction.apply(key, p.val);
1862 if (val != null)
1863 p.val = val;
1864 else {
1865 delta = -1;
1866 if (t.removeTreeNode(p))
1867 setTabAt(tab, i, untreeify(t.first));
1868 }
1869 }
1870 }
1871 else if (f instanceof ReservationNode)
1872 throw new IllegalStateException("Recursive update");
1873 }
1874 }
1875 if (binCount != 0)
1876 break;
1877 }
1878 }
1879 if (delta != 0)
1880 addCount((long)delta, binCount);
1881 return val;
1882 }
1883
1884 /**
1885 * Attempts to compute a mapping for the specified key and its
1886 * current mapped value (or {@code null} if there is no current
1887 * mapping). The entire method invocation is performed atomically.
1888 * Some attempted update operations on this map by other threads
1889 * may be blocked while computation is in progress, so the
1890 * computation should be short and simple, and must not attempt to
1891 * update any other mappings of this Map.
1892 *
1893 * @param key key with which the specified value is to be associated
1894 * @param remappingFunction the function to compute a value
1895 * @return the new value associated with the specified key, or null if none
1896 * @throws NullPointerException if the specified key or remappingFunction
1897 * is null
1898 * @throws IllegalStateException if the computation detectably
1899 * attempts a recursive update to this map that would
1900 * otherwise never complete
1901 * @throws RuntimeException or Error if the remappingFunction does so,
1902 * in which case the mapping is unchanged
1903 */
1904 public V compute(K key,
1905 BiFunction<? super K, ? super V, ? extends V> remappingFunction) {
1906 if (key == null || remappingFunction == null)
1907 throw new NullPointerException();
1908 int h = spread(key.hashCode());
1909 V val = null;
1910 int delta = 0;
1911 int binCount = 0;
1912 for (Node<K,V>[] tab = table;;) {
1913 Node<K,V> f; int n, i, fh;
1914 if (tab == null || (n = tab.length) == 0)
1915 tab = initTable();
1916 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
1917 Node<K,V> r = new ReservationNode<K,V>();
1918 synchronized (r) {
1919 if (casTabAt(tab, i, null, r)) {
1920 binCount = 1;
1921 Node<K,V> node = null;
1922 try {
1923 if ((val = remappingFunction.apply(key, null)) != null) {
1924 delta = 1;
1925 node = new Node<K,V>(h, key, val);
1926 }
1927 } finally {
1928 setTabAt(tab, i, node);
1929 }
1930 }
1931 }
1932 if (binCount != 0)
1933 break;
1934 }
1935 else if ((fh = f.hash) == MOVED)
1936 tab = helpTransfer(tab, f);
1937 else {
1938 synchronized (f) {
1939 if (tabAt(tab, i) == f) {
1940 if (fh >= 0) {
1941 binCount = 1;
1942 for (Node<K,V> e = f, pred = null;; ++binCount) {
1943 K ek;
1944 if (e.hash == h &&
1945 ((ek = e.key) == key ||
1946 (ek != null && key.equals(ek)))) {
1947 val = remappingFunction.apply(key, e.val);
1948 if (val != null)
1949 e.val = val;
1950 else {
1951 delta = -1;
1952 Node<K,V> en = e.next;
1953 if (pred != null)
1954 pred.next = en;
1955 else
1956 setTabAt(tab, i, en);
1957 }
1958 break;
1959 }
1960 pred = e;
1961 if ((e = e.next) == null) {
1962 val = remappingFunction.apply(key, null);
1963 if (val != null) {
1964 if (pred.next != null)
1965 throw new IllegalStateException("Recursive update");
1966 delta = 1;
1967 pred.next = new Node<K,V>(h, key, val);
1968 }
1969 break;
1970 }
1971 }
1972 }
1973 else if (f instanceof TreeBin) {
1974 binCount = 1;
1975 TreeBin<K,V> t = (TreeBin<K,V>)f;
1976 TreeNode<K,V> r, p;
1977 if ((r = t.root) != null)
1978 p = r.findTreeNode(h, key, null);
1979 else
1980 p = null;
1981 V pv = (p == null) ? null : p.val;
1982 val = remappingFunction.apply(key, pv);
1983 if (val != null) {
1984 if (p != null)
1985 p.val = val;
1986 else {
1987 delta = 1;
1988 t.putTreeVal(h, key, val);
1989 }
1990 }
1991 else if (p != null) {
1992 delta = -1;
1993 if (t.removeTreeNode(p))
1994 setTabAt(tab, i, untreeify(t.first));
1995 }
1996 }
1997 else if (f instanceof ReservationNode)
1998 throw new IllegalStateException("Recursive update");
1999 }
2000 }
2001 if (binCount != 0) {
2002 if (binCount >= TREEIFY_THRESHOLD)
2003 treeifyBin(tab, i);
2004 break;
2005 }
2006 }
2007 }
2008 if (delta != 0)
2009 addCount((long)delta, binCount);
2010 return val;
2011 }
2012
2013 /**
2014 * If the specified key is not already associated with a
2015 * (non-null) value, associates it with the given value.
2016 * Otherwise, replaces the value with the results of the given
2017 * remapping function, or removes if {@code null}. The entire
2018 * method invocation is performed atomically. Some attempted
2019 * update operations on this map by other threads may be blocked
2020 * while computation is in progress, so the computation should be
2021 * short and simple, and must not attempt to update any other
2022 * mappings of this Map.
2023 *
2024 * @param key key with which the specified value is to be associated
2025 * @param value the value to use if absent
2026 * @param remappingFunction the function to recompute a value if present
2027 * @return the new value associated with the specified key, or null if none
2028 * @throws NullPointerException if the specified key or the
2029 * remappingFunction is null
2030 * @throws RuntimeException or Error if the remappingFunction does so,
2031 * in which case the mapping is unchanged
2032 */
2033 public V merge(K key, V value, BiFunction<? super V, ? super V, ? extends V> remappingFunction) {
2034 if (key == null || value == null || remappingFunction == null)
2035 throw new NullPointerException();
2036 int h = spread(key.hashCode());
2037 V val = null;
2038 int delta = 0;
2039 int binCount = 0;
2040 for (Node<K,V>[] tab = table;;) {
2041 Node<K,V> f; int n, i, fh;
2042 if (tab == null || (n = tab.length) == 0)
2043 tab = initTable();
2044 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
2045 if (casTabAt(tab, i, null, new Node<K,V>(h, key, value))) {
2046 delta = 1;
2047 val = value;
2048 break;
2049 }
2050 }
2051 else if ((fh = f.hash) == MOVED)
2052 tab = helpTransfer(tab, f);
2053 else {
2054 synchronized (f) {
2055 if (tabAt(tab, i) == f) {
2056 if (fh >= 0) {
2057 binCount = 1;
2058 for (Node<K,V> e = f, pred = null;; ++binCount) {
2059 K ek;
2060 if (e.hash == h &&
2061 ((ek = e.key) == key ||
2062 (ek != null && key.equals(ek)))) {
2063 val = remappingFunction.apply(e.val, value);
2064 if (val != null)
2065 e.val = val;
2066 else {
2067 delta = -1;
2068 Node<K,V> en = e.next;
2069 if (pred != null)
2070 pred.next = en;
2071 else
2072 setTabAt(tab, i, en);
2073 }
2074 break;
2075 }
2076 pred = e;
2077 if ((e = e.next) == null) {
2078 delta = 1;
2079 val = value;
2080 pred.next = new Node<K,V>(h, key, val);
2081 break;
2082 }
2083 }
2084 }
2085 else if (f instanceof TreeBin) {
2086 binCount = 2;
2087 TreeBin<K,V> t = (TreeBin<K,V>)f;
2088 TreeNode<K,V> r = t.root;
2089 TreeNode<K,V> p = (r == null) ? null :
2090 r.findTreeNode(h, key, null);
2091 val = (p == null) ? value :
2092 remappingFunction.apply(p.val, value);
2093 if (val != null) {
2094 if (p != null)
2095 p.val = val;
2096 else {
2097 delta = 1;
2098 t.putTreeVal(h, key, val);
2099 }
2100 }
2101 else if (p != null) {
2102 delta = -1;
2103 if (t.removeTreeNode(p))
2104 setTabAt(tab, i, untreeify(t.first));
2105 }
2106 }
2107 else if (f instanceof ReservationNode)
2108 throw new IllegalStateException("Recursive update");
2109 }
2110 }
2111 if (binCount != 0) {
2112 if (binCount >= TREEIFY_THRESHOLD)
2113 treeifyBin(tab, i);
2114 break;
2115 }
2116 }
2117 }
2118 if (delta != 0)
2119 addCount((long)delta, binCount);
2120 return val;
2121 }
2122
2123 // Hashtable legacy methods
2124
2125 /**
2126 * Tests if some key maps into the specified value in this table.
2127 *
2128 * <p>Note that this method is identical in functionality to
2129 * {@link #containsValue(Object)}, and exists solely to ensure
2130 * full compatibility with class {@link java.util.Hashtable},
2131 * which supported this method prior to introduction of the
2132 * Java Collections Framework.
2133 *
2134 * @param value a value to search for
2135 * @return {@code true} if and only if some key maps to the
2136 * {@code value} argument in this table as
2137 * determined by the {@code equals} method;
2138 * {@code false} otherwise
2139 * @throws NullPointerException if the specified value is null
2140 */
2141 public boolean contains(Object value) {
2142 return containsValue(value);
2143 }
2144
2145 /**
2146 * Returns an enumeration of the keys in this table.
2147 *
2148 * @return an enumeration of the keys in this table
2149 * @see #keySet()
2150 */
2151 public Enumeration<K> keys() {
2152 Node<K,V>[] t;
2153 int f = (t = table) == null ? 0 : t.length;
2154 return new KeyIterator<K,V>(t, f, 0, f, this);
2155 }
2156
2157 /**
2158 * Returns an enumeration of the values in this table.
2159 *
2160 * @return an enumeration of the values in this table
2161 * @see #values()
2162 */
2163 public Enumeration<V> elements() {
2164 Node<K,V>[] t;
2165 int f = (t = table) == null ? 0 : t.length;
2166 return new ValueIterator<K,V>(t, f, 0, f, this);
2167 }
2168
2169 // ConcurrentHashMap-only methods
2170
2171 /**
2172 * Returns the number of mappings. This method should be used
2173 * instead of {@link #size} because a ConcurrentHashMap may
2174 * contain more mappings than can be represented as an int. The
2175 * value returned is an estimate; the actual count may differ if
2176 * there are concurrent insertions or removals.
2177 *
2178 * @return the number of mappings
2179 * @since 1.8
2180 */
2181 public long mappingCount() {
2182 long n = sumCount();
2183 return (n < 0L) ? 0L : n; // ignore transient negative values
2184 }
2185
2186 /**
2187 * Creates a new {@link Set} backed by a ConcurrentHashMap
2188 * from the given type to {@code Boolean.TRUE}.
2189 *
2190 * @param <K> the element type of the returned set
2191 * @return the new set
2192 * @since 1.8
2193 */
2194 public static <K> KeySetView<K,Boolean> newKeySet() {
2195 return new KeySetView<K,Boolean>
2196 (new ConcurrentHashMap<K,Boolean>(), Boolean.TRUE);
2197 }
2198
2199 /**
2200 * Creates a new {@link Set} backed by a ConcurrentHashMap
2201 * from the given type to {@code Boolean.TRUE}.
2202 *
2203 * @param initialCapacity The implementation performs internal
2204 * sizing to accommodate this many elements.
2205 * @param <K> the element type of the returned set
2206 * @return the new set
2207 * @throws IllegalArgumentException if the initial capacity of
2208 * elements is negative
2209 * @since 1.8
2210 */
2211 public static <K> KeySetView<K,Boolean> newKeySet(int initialCapacity) {
2212 return new KeySetView<K,Boolean>
2213 (new ConcurrentHashMap<K,Boolean>(initialCapacity), Boolean.TRUE);
2214 }
2215
2216 /**
2217 * Returns a {@link Set} view of the keys in this map, using the
2218 * given common mapped value for any additions (i.e., {@link
2219 * Collection#add} and {@link Collection#addAll(Collection)}).
2220 * This is of course only appropriate if it is acceptable to use
2221 * the same value for all additions from this view.
2222 *
2223 * @param mappedValue the mapped value to use for any additions
2224 * @return the set view
2225 * @throws NullPointerException if the mappedValue is null
2226 */
2227 public KeySetView<K,V> keySet(V mappedValue) {
2228 if (mappedValue == null)
2229 throw new NullPointerException();
2230 return new KeySetView<K,V>(this, mappedValue);
2231 }
2232
2233 /* ---------------- Special Nodes -------------- */
2234
2235 /**
2236 * A node inserted at head of bins during transfer operations.
2237 */
2238 static final class ForwardingNode<K,V> extends Node<K,V> {
2239 final Node<K,V>[] nextTable;
2240 ForwardingNode(Node<K,V>[] tab) {
2241 super(MOVED, null, null);
2242 this.nextTable = tab;
2243 }
2244
2245 Node<K,V> find(int h, Object k) {
2246 // loop to avoid arbitrarily deep recursion on forwarding nodes
2247 outer: for (Node<K,V>[] tab = nextTable;;) {
2248 Node<K,V> e; int n;
2249 if (k == null || tab == null || (n = tab.length) == 0 ||
2250 (e = tabAt(tab, (n - 1) & h)) == null)
2251 return null;
2252 for (;;) {
2253 int eh; K ek;
2254 if ((eh = e.hash) == h &&
2255 ((ek = e.key) == k || (ek != null && k.equals(ek))))
2256 return e;
2257 if (eh < 0) {
2258 if (e instanceof ForwardingNode) {
2259 tab = ((ForwardingNode<K,V>)e).nextTable;
2260 continue outer;
2261 }
2262 else
2263 return e.find(h, k);
2264 }
2265 if ((e = e.next) == null)
2266 return null;
2267 }
2268 }
2269 }
2270 }
2271
2272 /**
2273 * A place-holder node used in computeIfAbsent and compute.
2274 */
2275 static final class ReservationNode<K,V> extends Node<K,V> {
2276 ReservationNode() {
2277 super(RESERVED, null, null);
2278 }
2279
2280 Node<K,V> find(int h, Object k) {
2281 return null;
2282 }
2283 }
2284
2285 /* ---------------- Table Initialization and Resizing -------------- */
2286
2287 /**
2288 * Returns the stamp bits for resizing a table of size n.
2289 * Must be negative when shifted left by RESIZE_STAMP_SHIFT.
2290 */
2291 static final int resizeStamp(int n) {
2292 return Integer.numberOfLeadingZeros(n) | (1 << (RESIZE_STAMP_BITS - 1));
2293 }
2294
2295 /**
2296 * Initializes table, using the size recorded in sizeCtl.
2297 */
2298 private final Node<K,V>[] initTable() {
2299 Node<K,V>[] tab; int sc;
2300 while ((tab = table) == null || tab.length == 0) {
2301 if ((sc = sizeCtl) < 0)
2302 Thread.yield(); // lost initialization race; just spin
2303 else if (U.compareAndSetInt(this, SIZECTL, sc, -1)) {
2304 try {
2305 if ((tab = table) == null || tab.length == 0) {
2306 int n = (sc > 0) ? sc : DEFAULT_CAPACITY;
2307 @SuppressWarnings("unchecked")
2308 Node<K,V>[] nt = (Node<K,V>[])new Node<?,?>[n];
2309 table = tab = nt;
2310 sc = n - (n >>> 2);
2311 }
2312 } finally {
2313 sizeCtl = sc;
2314 }
2315 break;
2316 }
2317 }
2318 return tab;
2319 }
2320
2321 /**
2322 * Adds to count, and if table is too small and not already
2323 * resizing, initiates transfer. If already resizing, helps
2324 * perform transfer if work is available. Rechecks occupancy
2325 * after a transfer to see if another resize is already needed
2326 * because resizings are lagging additions.
2327 *
2328 * @param x the count to add
2329 * @param check if <0, don't check resize, if <= 1 only check if uncontended
2330 */
2331 private final void addCount(long x, int check) {
2332 CounterCell[] as; long b, s;
2333 if ((as = counterCells) != null ||
2334 !U.compareAndSetLong(this, BASECOUNT, b = baseCount, s = b + x)) {
2335 CounterCell a; long v; int m;
2336 boolean uncontended = true;
2337 if (as == null || (m = as.length - 1) < 0 ||
2338 (a = as[ThreadLocalRandom.getProbe() & m]) == null ||
2339 !(uncontended =
2340 U.compareAndSetLong(a, CELLVALUE, v = a.value, v + x))) {
2341 fullAddCount(x, uncontended);
2342 return;
2343 }
2344 if (check <= 1)
2345 return;
2346 s = sumCount();
2347 }
2348 if (check >= 0) {
2349 Node<K,V>[] tab, nt; int n, sc;
2350 while (s >= (long)(sc = sizeCtl) && (tab = table) != null &&
2351 (n = tab.length) < MAXIMUM_CAPACITY) {
2352 int rs = resizeStamp(n);
2353 if (sc < 0) {
2354 if ((sc >>> RESIZE_STAMP_SHIFT) != rs || sc == rs + 1 ||
2355 sc == rs + MAX_RESIZERS || (nt = nextTable) == null ||
2356 transferIndex <= 0)
2357 break;
2358 if (U.compareAndSetInt(this, SIZECTL, sc, sc + 1))
2359 transfer(tab, nt);
2360 }
2361 else if (U.compareAndSetInt(this, SIZECTL, sc,
2362 (rs << RESIZE_STAMP_SHIFT) + 2))
2363 transfer(tab, null);
2364 s = sumCount();
2365 }
2366 }
2367 }
2368
2369 /**
2370 * Helps transfer if a resize is in progress.
2371 */
2372 final Node<K,V>[] helpTransfer(Node<K,V>[] tab, Node<K,V> f) {
2373 Node<K,V>[] nextTab; int sc;
2374 if (tab != null && (f instanceof ForwardingNode) &&
2375 (nextTab = ((ForwardingNode<K,V>)f).nextTable) != null) {
2376 int rs = resizeStamp(tab.length);
2377 while (nextTab == nextTable && table == tab &&
2378 (sc = sizeCtl) < 0) {
2379 if ((sc >>> RESIZE_STAMP_SHIFT) != rs || sc == rs + 1 ||
2380 sc == rs + MAX_RESIZERS || transferIndex <= 0)
2381 break;
2382 if (U.compareAndSetInt(this, SIZECTL, sc, sc + 1)) {
2383 transfer(tab, nextTab);
2384 break;
2385 }
2386 }
2387 return nextTab;
2388 }
2389 return table;
2390 }
2391
2392 /**
2393 * Tries to presize table to accommodate the given number of elements.
2394 *
2395 * @param size number of elements (doesn't need to be perfectly accurate)
2396 */
2397 private final void tryPresize(int size) {
2398 int c = (size >= (MAXIMUM_CAPACITY >>> 1)) ? MAXIMUM_CAPACITY :
2399 tableSizeFor(size + (size >>> 1) + 1);
2400 int sc;
2401 while ((sc = sizeCtl) >= 0) {
2402 Node<K,V>[] tab = table; int n;
2403 if (tab == null || (n = tab.length) == 0) {
2404 n = (sc > c) ? sc : c;
2405 if (U.compareAndSetInt(this, SIZECTL, sc, -1)) {
2406 try {
2407 if (table == tab) {
2408 @SuppressWarnings("unchecked")
2409 Node<K,V>[] nt = (Node<K,V>[])new Node<?,?>[n];
2410 table = nt;
2411 sc = n - (n >>> 2);
2412 }
2413 } finally {
2414 sizeCtl = sc;
2415 }
2416 }
2417 }
2418 else if (c <= sc || n >= MAXIMUM_CAPACITY)
2419 break;
2420 else if (tab == table) {
2421 int rs = resizeStamp(n);
2422 if (U.compareAndSetInt(this, SIZECTL, sc,
2423 (rs << RESIZE_STAMP_SHIFT) + 2))
2424 transfer(tab, null);
2425 }
2426 }
2427 }
2428
2429 /**
2430 * Moves and/or copies the nodes in each bin to new table. See
2431 * above for explanation.
2432 */
2433 private final void transfer(Node<K,V>[] tab, Node<K,V>[] nextTab) {
2434 int n = tab.length, stride;
2435 if ((stride = (NCPU > 1) ? (n >>> 3) / NCPU : n) < MIN_TRANSFER_STRIDE)
2436 stride = MIN_TRANSFER_STRIDE; // subdivide range
2437 if (nextTab == null) { // initiating
2438 try {
2439 @SuppressWarnings("unchecked")
2440 Node<K,V>[] nt = (Node<K,V>[])new Node<?,?>[n << 1];
2441 nextTab = nt;
2442 } catch (Throwable ex) { // try to cope with OOME
2443 sizeCtl = Integer.MAX_VALUE;
2444 return;
2445 }
2446 nextTable = nextTab;
2447 transferIndex = n;
2448 }
2449 int nextn = nextTab.length;
2450 ForwardingNode<K,V> fwd = new ForwardingNode<K,V>(nextTab);
2451 boolean advance = true;
2452 boolean finishing = false; // to ensure sweep before committing nextTab
2453 for (int i = 0, bound = 0;;) {
2454 Node<K,V> f; int fh;
2455 while (advance) {
2456 int nextIndex, nextBound;
2457 if (--i >= bound || finishing)
2458 advance = false;
2459 else if ((nextIndex = transferIndex) <= 0) {
2460 i = -1;
2461 advance = false;
2462 }
2463 else if (U.compareAndSetInt
2464 (this, TRANSFERINDEX, nextIndex,
2465 nextBound = (nextIndex > stride ?
2466 nextIndex - stride : 0))) {
2467 bound = nextBound;
2468 i = nextIndex - 1;
2469 advance = false;
2470 }
2471 }
2472 if (i < 0 || i >= n || i + n >= nextn) {
2473 int sc;
2474 if (finishing) {
2475 nextTable = null;
2476 table = nextTab;
2477 sizeCtl = (n << 1) - (n >>> 1);
2478 return;
2479 }
2480 if (U.compareAndSetInt(this, SIZECTL, sc = sizeCtl, sc - 1)) {
2481 if ((sc - 2) != resizeStamp(n) << RESIZE_STAMP_SHIFT)
2482 return;
2483 finishing = advance = true;
2484 i = n; // recheck before commit
2485 }
2486 }
2487 else if ((f = tabAt(tab, i)) == null)
2488 advance = casTabAt(tab, i, null, fwd);
2489 else if ((fh = f.hash) == MOVED)
2490 advance = true; // already processed
2491 else {
2492 synchronized (f) {
2493 if (tabAt(tab, i) == f) {
2494 Node<K,V> ln, hn;
2495 if (fh >= 0) {
2496 int runBit = fh & n;
2497 Node<K,V> lastRun = f;
2498 for (Node<K,V> p = f.next; p != null; p = p.next) {
2499 int b = p.hash & n;
2500 if (b != runBit) {
2501 runBit = b;
2502 lastRun = p;
2503 }
2504 }
2505 if (runBit == 0) {
2506 ln = lastRun;
2507 hn = null;
2508 }
2509 else {
2510 hn = lastRun;
2511 ln = null;
2512 }
2513 for (Node<K,V> p = f; p != lastRun; p = p.next) {
2514 int ph = p.hash; K pk = p.key; V pv = p.val;
2515 if ((ph & n) == 0)
2516 ln = new Node<K,V>(ph, pk, pv, ln);
2517 else
2518 hn = new Node<K,V>(ph, pk, pv, hn);
2519 }
2520 setTabAt(nextTab, i, ln);
2521 setTabAt(nextTab, i + n, hn);
2522 setTabAt(tab, i, fwd);
2523 advance = true;
2524 }
2525 else if (f instanceof TreeBin) {
2526 TreeBin<K,V> t = (TreeBin<K,V>)f;
2527 TreeNode<K,V> lo = null, loTail = null;
2528 TreeNode<K,V> hi = null, hiTail = null;
2529 int lc = 0, hc = 0;
2530 for (Node<K,V> e = t.first; e != null; e = e.next) {
2531 int h = e.hash;
2532 TreeNode<K,V> p = new TreeNode<K,V>
2533 (h, e.key, e.val, null, null);
2534 if ((h & n) == 0) {
2535 if ((p.prev = loTail) == null)
2536 lo = p;
2537 else
2538 loTail.next = p;
2539 loTail = p;
2540 ++lc;
2541 }
2542 else {
2543 if ((p.prev = hiTail) == null)
2544 hi = p;
2545 else
2546 hiTail.next = p;
2547 hiTail = p;
2548 ++hc;
2549 }
2550 }
2551 ln = (lc <= UNTREEIFY_THRESHOLD) ? untreeify(lo) :
2552 (hc != 0) ? new TreeBin<K,V>(lo) : t;
2553 hn = (hc <= UNTREEIFY_THRESHOLD) ? untreeify(hi) :
2554 (lc != 0) ? new TreeBin<K,V>(hi) : t;
2555 setTabAt(nextTab, i, ln);
2556 setTabAt(nextTab, i + n, hn);
2557 setTabAt(tab, i, fwd);
2558 advance = true;
2559 }
2560 }
2561 }
2562 }
2563 }
2564 }
2565
2566 /* ---------------- Counter support -------------- */
2567
2568 /**
2569 * A padded cell for distributing counts. Adapted from LongAdder
2570 * and Striped64. See their internal docs for explanation.
2571 */
2572 @jdk.internal.vm.annotation.Contended static final class CounterCell {
2573 volatile long value;
2574 CounterCell(long x) { value = x; }
2575 }
2576
2577 final long sumCount() {
2578 CounterCell[] as = counterCells; CounterCell a;
2579 long sum = baseCount;
2580 if (as != null) {
2581 for (int i = 0; i < as.length; ++i) {
2582 if ((a = as[i]) != null)
2583 sum += a.value;
2584 }
2585 }
2586 return sum;
2587 }
2588
2589 // See LongAdder version for explanation
2590 private final void fullAddCount(long x, boolean wasUncontended) {
2591 int h;
2592 if ((h = ThreadLocalRandom.getProbe()) == 0) {
2593 ThreadLocalRandom.localInit(); // force initialization
2594 h = ThreadLocalRandom.getProbe();
2595 wasUncontended = true;
2596 }
2597 boolean collide = false; // True if last slot nonempty
2598 for (;;) {
2599 CounterCell[] as; CounterCell a; int n; long v;
2600 if ((as = counterCells) != null && (n = as.length) > 0) {
2601 if ((a = as[(n - 1) & h]) == null) {
2602 if (cellsBusy == 0) { // Try to attach new Cell
2603 CounterCell r = new CounterCell(x); // Optimistic create
2604 if (cellsBusy == 0 &&
2605 U.compareAndSetInt(this, CELLSBUSY, 0, 1)) {
2606 boolean created = false;
2607 try { // Recheck under lock
2608 CounterCell[] rs; int m, j;
2609 if ((rs = counterCells) != null &&
2610 (m = rs.length) > 0 &&
2611 rs[j = (m - 1) & h] == null) {
2612 rs[j] = r;
2613 created = true;
2614 }
2615 } finally {
2616 cellsBusy = 0;
2617 }
2618 if (created)
2619 break;
2620 continue; // Slot is now non-empty
2621 }
2622 }
2623 collide = false;
2624 }
2625 else if (!wasUncontended) // CAS already known to fail
2626 wasUncontended = true; // Continue after rehash
2627 else if (U.compareAndSetLong(a, CELLVALUE, v = a.value, v + x))
2628 break;
2629 else if (counterCells != as || n >= NCPU)
2630 collide = false; // At max size or stale
2631 else if (!collide)
2632 collide = true;
2633 else if (cellsBusy == 0 &&
2634 U.compareAndSetInt(this, CELLSBUSY, 0, 1)) {
2635 try {
2636 if (counterCells == as) {// Expand table unless stale
2637 CounterCell[] rs = new CounterCell[n << 1];
2638 for (int i = 0; i < n; ++i)
2639 rs[i] = as[i];
2640 counterCells = rs;
2641 }
2642 } finally {
2643 cellsBusy = 0;
2644 }
2645 collide = false;
2646 continue; // Retry with expanded table
2647 }
2648 h = ThreadLocalRandom.advanceProbe(h);
2649 }
2650 else if (cellsBusy == 0 && counterCells == as &&
2651 U.compareAndSetInt(this, CELLSBUSY, 0, 1)) {
2652 boolean init = false;
2653 try { // Initialize table
2654 if (counterCells == as) {
2655 CounterCell[] rs = new CounterCell[2];
2656 rs[h & 1] = new CounterCell(x);
2657 counterCells = rs;
2658 init = true;
2659 }
2660 } finally {
2661 cellsBusy = 0;
2662 }
2663 if (init)
2664 break;
2665 }
2666 else if (U.compareAndSetLong(this, BASECOUNT, v = baseCount, v + x))
2667 break; // Fall back on using base
2668 }
2669 }
2670
2671 /* ---------------- Conversion from/to TreeBins -------------- */
2672
2673 /**
2674 * Replaces all linked nodes in bin at given index unless table is
2675 * too small, in which case resizes instead.
2676 */
2677 private final void treeifyBin(Node<K,V>[] tab, int index) {
2678 Node<K,V> b; int n;
2679 if (tab != null) {
2680 if ((n = tab.length) < MIN_TREEIFY_CAPACITY)
2681 tryPresize(n << 1);
2682 else if ((b = tabAt(tab, index)) != null && b.hash >= 0) {
2683 synchronized (b) {
2684 if (tabAt(tab, index) == b) {
2685 TreeNode<K,V> hd = null, tl = null;
2686 for (Node<K,V> e = b; e != null; e = e.next) {
2687 TreeNode<K,V> p =
2688 new TreeNode<K,V>(e.hash, e.key, e.val,
2689 null, null);
2690 if ((p.prev = tl) == null)
2691 hd = p;
2692 else
2693 tl.next = p;
2694 tl = p;
2695 }
2696 setTabAt(tab, index, new TreeBin<K,V>(hd));
2697 }
2698 }
2699 }
2700 }
2701 }
2702
2703 /**
2704 * Returns a list of non-TreeNodes replacing those in given list.
2705 */
2706 static <K,V> Node<K,V> untreeify(Node<K,V> b) {
2707 Node<K,V> hd = null, tl = null;
2708 for (Node<K,V> q = b; q != null; q = q.next) {
2709 Node<K,V> p = new Node<K,V>(q.hash, q.key, q.val);
2710 if (tl == null)
2711 hd = p;
2712 else
2713 tl.next = p;
2714 tl = p;
2715 }
2716 return hd;
2717 }
2718
2719 /* ---------------- TreeNodes -------------- */
2720
2721 /**
2722 * Nodes for use in TreeBins.
2723 */
2724 static final class TreeNode<K,V> extends Node<K,V> {
2725 TreeNode<K,V> parent; // red-black tree links
2726 TreeNode<K,V> left;
2727 TreeNode<K,V> right;
2728 TreeNode<K,V> prev; // needed to unlink next upon deletion
2729 boolean red;
2730
2731 TreeNode(int hash, K key, V val, Node<K,V> next,
2732 TreeNode<K,V> parent) {
2733 super(hash, key, val, next);
2734 this.parent = parent;
2735 }
2736
2737 Node<K,V> find(int h, Object k) {
2738 return findTreeNode(h, k, null);
2739 }
2740
2741 /**
2742 * Returns the TreeNode (or null if not found) for the given key
2743 * starting at given root.
2744 */
2745 final TreeNode<K,V> findTreeNode(int h, Object k, Class<?> kc) {
2746 if (k != null) {
2747 TreeNode<K,V> p = this;
2748 do {
2749 int ph, dir; K pk; TreeNode<K,V> q;
2750 TreeNode<K,V> pl = p.left, pr = p.right;
2751 if ((ph = p.hash) > h)
2752 p = pl;
2753 else if (ph < h)
2754 p = pr;
2755 else if ((pk = p.key) == k || (pk != null && k.equals(pk)))
2756 return p;
2757 else if (pl == null)
2758 p = pr;
2759 else if (pr == null)
2760 p = pl;
2761 else if ((kc != null ||
2762 (kc = comparableClassFor(k)) != null) &&
2763 (dir = compareComparables(kc, k, pk)) != 0)
2764 p = (dir < 0) ? pl : pr;
2765 else if ((q = pr.findTreeNode(h, k, kc)) != null)
2766 return q;
2767 else
2768 p = pl;
2769 } while (p != null);
2770 }
2771 return null;
2772 }
2773 }
2774
2775 /* ---------------- TreeBins -------------- */
2776
2777 /**
2778 * TreeNodes used at the heads of bins. TreeBins do not hold user
2779 * keys or values, but instead point to list of TreeNodes and
2780 * their root. They also maintain a parasitic read-write lock
2781 * forcing writers (who hold bin lock) to wait for readers (who do
2782 * not) to complete before tree restructuring operations.
2783 */
2784 static final class TreeBin<K,V> extends Node<K,V> {
2785 TreeNode<K,V> root;
2786 volatile TreeNode<K,V> first;
2787 volatile Thread waiter;
2788 volatile int lockState;
2789 // values for lockState
2790 static final int WRITER = 1; // set while holding write lock
2791 static final int WAITER = 2; // set when waiting for write lock
2792 static final int READER = 4; // increment value for setting read lock
2793
2794 /**
2795 * Tie-breaking utility for ordering insertions when equal
2796 * hashCodes and non-comparable. We don't require a total
2797 * order, just a consistent insertion rule to maintain
2798 * equivalence across rebalancings. Tie-breaking further than
2799 * necessary simplifies testing a bit.
2800 */
2801 static int tieBreakOrder(Object a, Object b) {
2802 int d;
2803 if (a == null || b == null ||
2804 (d = a.getClass().getName().
2805 compareTo(b.getClass().getName())) == 0)
2806 d = (System.identityHashCode(a) <= System.identityHashCode(b) ?
2807 -1 : 1);
2808 return d;
2809 }
2810
2811 /**
2812 * Creates bin with initial set of nodes headed by b.
2813 */
2814 TreeBin(TreeNode<K,V> b) {
2815 super(TREEBIN, null, null);
2816 this.first = b;
2817 TreeNode<K,V> r = null;
2818 for (TreeNode<K,V> x = b, next; x != null; x = next) {
2819 next = (TreeNode<K,V>)x.next;
2820 x.left = x.right = null;
2821 if (r == null) {
2822 x.parent = null;
2823 x.red = false;
2824 r = x;
2825 }
2826 else {
2827 K k = x.key;
2828 int h = x.hash;
2829 Class<?> kc = null;
2830 for (TreeNode<K,V> p = r;;) {
2831 int dir, ph;
2832 K pk = p.key;
2833 if ((ph = p.hash) > h)
2834 dir = -1;
2835 else if (ph < h)
2836 dir = 1;
2837 else if ((kc == null &&
2838 (kc = comparableClassFor(k)) == null) ||
2839 (dir = compareComparables(kc, k, pk)) == 0)
2840 dir = tieBreakOrder(k, pk);
2841 TreeNode<K,V> xp = p;
2842 if ((p = (dir <= 0) ? p.left : p.right) == null) {
2843 x.parent = xp;
2844 if (dir <= 0)
2845 xp.left = x;
2846 else
2847 xp.right = x;
2848 r = balanceInsertion(r, x);
2849 break;
2850 }
2851 }
2852 }
2853 }
2854 this.root = r;
2855 assert checkInvariants(root);
2856 }
2857
2858 /**
2859 * Acquires write lock for tree restructuring.
2860 */
2861 private final void lockRoot() {
2862 if (!U.compareAndSetInt(this, LOCKSTATE, 0, WRITER))
2863 contendedLock(); // offload to separate method
2864 }
2865
2866 /**
2867 * Releases write lock for tree restructuring.
2868 */
2869 private final void unlockRoot() {
2870 lockState = 0;
2871 }
2872
2873 /**
2874 * Possibly blocks awaiting root lock.
2875 */
2876 private final void contendedLock() {
2877 boolean waiting = false;
2878 for (int s;;) {
2879 if (((s = lockState) & ~WAITER) == 0) {
2880 if (U.compareAndSetInt(this, LOCKSTATE, s, WRITER)) {
2881 if (waiting)
2882 waiter = null;
2883 return;
2884 }
2885 }
2886 else if ((s & WAITER) == 0) {
2887 if (U.compareAndSetInt(this, LOCKSTATE, s, s | WAITER)) {
2888 waiting = true;
2889 waiter = Thread.currentThread();
2890 }
2891 }
2892 else if (waiting)
2893 LockSupport.park(this);
2894 }
2895 }
2896
2897 /**
2898 * Returns matching node or null if none. Tries to search
2899 * using tree comparisons from root, but continues linear
2900 * search when lock not available.
2901 */
2902 final Node<K,V> find(int h, Object k) {
2903 if (k != null) {
2904 for (Node<K,V> e = first; e != null; ) {
2905 int s; K ek;
2906 if (((s = lockState) & (WAITER|WRITER)) != 0) {
2907 if (e.hash == h &&
2908 ((ek = e.key) == k || (ek != null && k.equals(ek))))
2909 return e;
2910 e = e.next;
2911 }
2912 else if (U.compareAndSetInt(this, LOCKSTATE, s,
2913 s + READER)) {
2914 TreeNode<K,V> r, p;
2915 try {
2916 p = ((r = root) == null ? null :
2917 r.findTreeNode(h, k, null));
2918 } finally {
2919 Thread w;
2920 if (U.getAndAddInt(this, LOCKSTATE, -READER) ==
2921 (READER|WAITER) && (w = waiter) != null)
2922 LockSupport.unpark(w);
2923 }
2924 return p;
2925 }
2926 }
2927 }
2928 return null;
2929 }
2930
2931 /**
2932 * Finds or adds a node.
2933 * @return null if added
2934 */
2935 final TreeNode<K,V> putTreeVal(int h, K k, V v) {
2936 Class<?> kc = null;
2937 boolean searched = false;
2938 for (TreeNode<K,V> p = root;;) {
2939 int dir, ph; K pk;
2940 if (p == null) {
2941 first = root = new TreeNode<K,V>(h, k, v, null, null);
2942 break;
2943 }
2944 else if ((ph = p.hash) > h)
2945 dir = -1;
2946 else if (ph < h)
2947 dir = 1;
2948 else if ((pk = p.key) == k || (pk != null && k.equals(pk)))
2949 return p;
2950 else if ((kc == null &&
2951 (kc = comparableClassFor(k)) == null) ||
2952 (dir = compareComparables(kc, k, pk)) == 0) {
2953 if (!searched) {
2954 TreeNode<K,V> q, ch;
2955 searched = true;
2956 if (((ch = p.left) != null &&
2957 (q = ch.findTreeNode(h, k, kc)) != null) ||
2958 ((ch = p.right) != null &&
2959 (q = ch.findTreeNode(h, k, kc)) != null))
2960 return q;
2961 }
2962 dir = tieBreakOrder(k, pk);
2963 }
2964
2965 TreeNode<K,V> xp = p;
2966 if ((p = (dir <= 0) ? p.left : p.right) == null) {
2967 TreeNode<K,V> x, f = first;
2968 first = x = new TreeNode<K,V>(h, k, v, f, xp);
2969 if (f != null)
2970 f.prev = x;
2971 if (dir <= 0)
2972 xp.left = x;
2973 else
2974 xp.right = x;
2975 if (!xp.red)
2976 x.red = true;
2977 else {
2978 lockRoot();
2979 try {
2980 root = balanceInsertion(root, x);
2981 } finally {
2982 unlockRoot();
2983 }
2984 }
2985 break;
2986 }
2987 }
2988 assert checkInvariants(root);
2989 return null;
2990 }
2991
2992 /**
2993 * Removes the given node, that must be present before this
2994 * call. This is messier than typical red-black deletion code
2995 * because we cannot swap the contents of an interior node
2996 * with a leaf successor that is pinned by "next" pointers
2997 * that are accessible independently of lock. So instead we
2998 * swap the tree linkages.
2999 *
3000 * @return true if now too small, so should be untreeified
3001 */
3002 final boolean removeTreeNode(TreeNode<K,V> p) {
3003 TreeNode<K,V> next = (TreeNode<K,V>)p.next;
3004 TreeNode<K,V> pred = p.prev; // unlink traversal pointers
3005 TreeNode<K,V> r, rl;
3006 if (pred == null)
3007 first = next;
3008 else
3009 pred.next = next;
3010 if (next != null)
3011 next.prev = pred;
3012 if (first == null) {
3013 root = null;
3014 return true;
3015 }
3016 if ((r = root) == null || r.right == null || // too small
3017 (rl = r.left) == null || rl.left == null)
3018 return true;
3019 lockRoot();
3020 try {
3021 TreeNode<K,V> replacement;
3022 TreeNode<K,V> pl = p.left;
3023 TreeNode<K,V> pr = p.right;
3024 if (pl != null && pr != null) {
3025 TreeNode<K,V> s = pr, sl;
3026 while ((sl = s.left) != null) // find successor
3027 s = sl;
3028 boolean c = s.red; s.red = p.red; p.red = c; // swap colors
3029 TreeNode<K,V> sr = s.right;
3030 TreeNode<K,V> pp = p.parent;
3031 if (s == pr) { // p was s's direct parent
3032 p.parent = s;
3033 s.right = p;
3034 }
3035 else {
3036 TreeNode<K,V> sp = s.parent;
3037 if ((p.parent = sp) != null) {
3038 if (s == sp.left)
3039 sp.left = p;
3040 else
3041 sp.right = p;
3042 }
3043 if ((s.right = pr) != null)
3044 pr.parent = s;
3045 }
3046 p.left = null;
3047 if ((p.right = sr) != null)
3048 sr.parent = p;
3049 if ((s.left = pl) != null)
3050 pl.parent = s;
3051 if ((s.parent = pp) == null)
3052 r = s;
3053 else if (p == pp.left)
3054 pp.left = s;
3055 else
3056 pp.right = s;
3057 if (sr != null)
3058 replacement = sr;
3059 else
3060 replacement = p;
3061 }
3062 else if (pl != null)
3063 replacement = pl;
3064 else if (pr != null)
3065 replacement = pr;
3066 else
3067 replacement = p;
3068 if (replacement != p) {
3069 TreeNode<K,V> pp = replacement.parent = p.parent;
3070 if (pp == null)
3071 r = replacement;
3072 else if (p == pp.left)
3073 pp.left = replacement;
3074 else
3075 pp.right = replacement;
3076 p.left = p.right = p.parent = null;
3077 }
3078
3079 root = (p.red) ? r : balanceDeletion(r, replacement);
3080
3081 if (p == replacement) { // detach pointers
3082 TreeNode<K,V> pp;
3083 if ((pp = p.parent) != null) {
3084 if (p == pp.left)
3085 pp.left = null;
3086 else if (p == pp.right)
3087 pp.right = null;
3088 p.parent = null;
3089 }
3090 }
3091 } finally {
3092 unlockRoot();
3093 }
3094 assert checkInvariants(root);
3095 return false;
3096 }
3097
3098 /* ------------------------------------------------------------ */
3099 // Red-black tree methods, all adapted from CLR
3100
3101 static <K,V> TreeNode<K,V> rotateLeft(TreeNode<K,V> root,
3102 TreeNode<K,V> p) {
3103 TreeNode<K,V> r, pp, rl;
3104 if (p != null && (r = p.right) != null) {
3105 if ((rl = p.right = r.left) != null)
3106 rl.parent = p;
3107 if ((pp = r.parent = p.parent) == null)
3108 (root = r).red = false;
3109 else if (pp.left == p)
3110 pp.left = r;
3111 else
3112 pp.right = r;
3113 r.left = p;
3114 p.parent = r;
3115 }
3116 return root;
3117 }
3118
3119 static <K,V> TreeNode<K,V> rotateRight(TreeNode<K,V> root,
3120 TreeNode<K,V> p) {
3121 TreeNode<K,V> l, pp, lr;
3122 if (p != null && (l = p.left) != null) {
3123 if ((lr = p.left = l.right) != null)
3124 lr.parent = p;
3125 if ((pp = l.parent = p.parent) == null)
3126 (root = l).red = false;
3127 else if (pp.right == p)
3128 pp.right = l;
3129 else
3130 pp.left = l;
3131 l.right = p;
3132 p.parent = l;
3133 }
3134 return root;
3135 }
3136
3137 static <K,V> TreeNode<K,V> balanceInsertion(TreeNode<K,V> root,
3138 TreeNode<K,V> x) {
3139 x.red = true;
3140 for (TreeNode<K,V> xp, xpp, xppl, xppr;;) {
3141 if ((xp = x.parent) == null) {
3142 x.red = false;
3143 return x;
3144 }
3145 else if (!xp.red || (xpp = xp.parent) == null)
3146 return root;
3147 if (xp == (xppl = xpp.left)) {
3148 if ((xppr = xpp.right) != null && xppr.red) {
3149 xppr.red = false;
3150 xp.red = false;
3151 xpp.red = true;
3152 x = xpp;
3153 }
3154 else {
3155 if (x == xp.right) {
3156 root = rotateLeft(root, x = xp);
3157 xpp = (xp = x.parent) == null ? null : xp.parent;
3158 }
3159 if (xp != null) {
3160 xp.red = false;
3161 if (xpp != null) {
3162 xpp.red = true;
3163 root = rotateRight(root, xpp);
3164 }
3165 }
3166 }
3167 }
3168 else {
3169 if (xppl != null && xppl.red) {
3170 xppl.red = false;
3171 xp.red = false;
3172 xpp.red = true;
3173 x = xpp;
3174 }
3175 else {
3176 if (x == xp.left) {
3177 root = rotateRight(root, x = xp);
3178 xpp = (xp = x.parent) == null ? null : xp.parent;
3179 }
3180 if (xp != null) {
3181 xp.red = false;
3182 if (xpp != null) {
3183 xpp.red = true;
3184 root = rotateLeft(root, xpp);
3185 }
3186 }
3187 }
3188 }
3189 }
3190 }
3191
3192 static <K,V> TreeNode<K,V> balanceDeletion(TreeNode<K,V> root,
3193 TreeNode<K,V> x) {
3194 for (TreeNode<K,V> xp, xpl, xpr;;) {
3195 if (x == null || x == root)
3196 return root;
3197 else if ((xp = x.parent) == null) {
3198 x.red = false;
3199 return x;
3200 }
3201 else if (x.red) {
3202 x.red = false;
3203 return root;
3204 }
3205 else if ((xpl = xp.left) == x) {
3206 if ((xpr = xp.right) != null && xpr.red) {
3207 xpr.red = false;
3208 xp.red = true;
3209 root = rotateLeft(root, xp);
3210 xpr = (xp = x.parent) == null ? null : xp.right;
3211 }
3212 if (xpr == null)
3213 x = xp;
3214 else {
3215 TreeNode<K,V> sl = xpr.left, sr = xpr.right;
3216 if ((sr == null || !sr.red) &&
3217 (sl == null || !sl.red)) {
3218 xpr.red = true;
3219 x = xp;
3220 }
3221 else {
3222 if (sr == null || !sr.red) {
3223 if (sl != null)
3224 sl.red = false;
3225 xpr.red = true;
3226 root = rotateRight(root, xpr);
3227 xpr = (xp = x.parent) == null ?
3228 null : xp.right;
3229 }
3230 if (xpr != null) {
3231 xpr.red = (xp == null) ? false : xp.red;
3232 if ((sr = xpr.right) != null)
3233 sr.red = false;
3234 }
3235 if (xp != null) {
3236 xp.red = false;
3237 root = rotateLeft(root, xp);
3238 }
3239 x = root;
3240 }
3241 }
3242 }
3243 else { // symmetric
3244 if (xpl != null && xpl.red) {
3245 xpl.red = false;
3246 xp.red = true;
3247 root = rotateRight(root, xp);
3248 xpl = (xp = x.parent) == null ? null : xp.left;
3249 }
3250 if (xpl == null)
3251 x = xp;
3252 else {
3253 TreeNode<K,V> sl = xpl.left, sr = xpl.right;
3254 if ((sl == null || !sl.red) &&
3255 (sr == null || !sr.red)) {
3256 xpl.red = true;
3257 x = xp;
3258 }
3259 else {
3260 if (sl == null || !sl.red) {
3261 if (sr != null)
3262 sr.red = false;
3263 xpl.red = true;
3264 root = rotateLeft(root, xpl);
3265 xpl = (xp = x.parent) == null ?
3266 null : xp.left;
3267 }
3268 if (xpl != null) {
3269 xpl.red = (xp == null) ? false : xp.red;
3270 if ((sl = xpl.left) != null)
3271 sl.red = false;
3272 }
3273 if (xp != null) {
3274 xp.red = false;
3275 root = rotateRight(root, xp);
3276 }
3277 x = root;
3278 }
3279 }
3280 }
3281 }
3282 }
3283
3284 /**
3285 * Checks invariants recursively for the tree of Nodes rooted at t.
3286 */
3287 static <K,V> boolean checkInvariants(TreeNode<K,V> t) {
3288 TreeNode<K,V> tp = t.parent, tl = t.left, tr = t.right,
3289 tb = t.prev, tn = (TreeNode<K,V>)t.next;
3290 if (tb != null && tb.next != t)
3291 return false;
3292 if (tn != null && tn.prev != t)
3293 return false;
3294 if (tp != null && t != tp.left && t != tp.right)
3295 return false;
3296 if (tl != null && (tl.parent != t || tl.hash > t.hash))
3297 return false;
3298 if (tr != null && (tr.parent != t || tr.hash < t.hash))
3299 return false;
3300 if (t.red && tl != null && tl.red && tr != null && tr.red)
3301 return false;
3302 if (tl != null && !checkInvariants(tl))
3303 return false;
3304 if (tr != null && !checkInvariants(tr))
3305 return false;
3306 return true;
3307 }
3308
3309 private static final Unsafe U = Unsafe.getUnsafe();
3310 private static final long LOCKSTATE
3311 = U.objectFieldOffset(TreeBin.class, "lockState");
3312 }
3313
3314 /* ----------------Table Traversal -------------- */
3315
3316 /**
3317 * Records the table, its length, and current traversal index for a
3318 * traverser that must process a region of a forwarded table before
3319 * proceeding with current table.
3320 */
3321 static final class TableStack<K,V> {
3322 int length;
3323 int index;
3324 Node<K,V>[] tab;
3325 TableStack<K,V> next;
3326 }
3327
3328 /**
3329 * Encapsulates traversal for methods such as containsValue; also
3330 * serves as a base class for other iterators and spliterators.
3331 *
3332 * Method advance visits once each still-valid node that was
3333 * reachable upon iterator construction. It might miss some that
3334 * were added to a bin after the bin was visited, which is OK wrt
3335 * consistency guarantees. Maintaining this property in the face
3336 * of possible ongoing resizes requires a fair amount of
3337 * bookkeeping state that is difficult to optimize away amidst
3338 * volatile accesses. Even so, traversal maintains reasonable
3339 * throughput.
3340 *
3341 * Normally, iteration proceeds bin-by-bin traversing lists.
3342 * However, if the table has been resized, then all future steps
3343 * must traverse both the bin at the current index as well as at
3344 * (index + baseSize); and so on for further resizings. To
3345 * paranoically cope with potential sharing by users of iterators
3346 * across threads, iteration terminates if a bounds checks fails
3347 * for a table read.
3348 */
3349 static class Traverser<K,V> {
3350 Node<K,V>[] tab; // current table; updated if resized
3351 Node<K,V> next; // the next entry to use
3352 TableStack<K,V> stack, spare; // to save/restore on ForwardingNodes
3353 int index; // index of bin to use next
3354 int baseIndex; // current index of initial table
3355 int baseLimit; // index bound for initial table
3356 final int baseSize; // initial table size
3357
3358 Traverser(Node<K,V>[] tab, int size, int index, int limit) {
3359 this.tab = tab;
3360 this.baseSize = size;
3361 this.baseIndex = this.index = index;
3362 this.baseLimit = limit;
3363 this.next = null;
3364 }
3365
3366 /**
3367 * Advances if possible, returning next valid node, or null if none.
3368 */
3369 final Node<K,V> advance() {
3370 Node<K,V> e;
3371 if ((e = next) != null)
3372 e = e.next;
3373 for (;;) {
3374 Node<K,V>[] t; int i, n; // must use locals in checks
3375 if (e != null)
3376 return next = e;
3377 if (baseIndex >= baseLimit || (t = tab) == null ||
3378 (n = t.length) <= (i = index) || i < 0)
3379 return next = null;
3380 if ((e = tabAt(t, i)) != null && e.hash < 0) {
3381 if (e instanceof ForwardingNode) {
3382 tab = ((ForwardingNode<K,V>)e).nextTable;
3383 e = null;
3384 pushState(t, i, n);
3385 continue;
3386 }
3387 else if (e instanceof TreeBin)
3388 e = ((TreeBin<K,V>)e).first;
3389 else
3390 e = null;
3391 }
3392 if (stack != null)
3393 recoverState(n);
3394 else if ((index = i + baseSize) >= n)
3395 index = ++baseIndex; // visit upper slots if present
3396 }
3397 }
3398
3399 /**
3400 * Saves traversal state upon encountering a forwarding node.
3401 */
3402 private void pushState(Node<K,V>[] t, int i, int n) {
3403 TableStack<K,V> s = spare; // reuse if possible
3404 if (s != null)
3405 spare = s.next;
3406 else
3407 s = new TableStack<K,V>();
3408 s.tab = t;
3409 s.length = n;
3410 s.index = i;
3411 s.next = stack;
3412 stack = s;
3413 }
3414
3415 /**
3416 * Possibly pops traversal state.
3417 *
3418 * @param n length of current table
3419 */
3420 private void recoverState(int n) {
3421 TableStack<K,V> s; int len;
3422 while ((s = stack) != null && (index += (len = s.length)) >= n) {
3423 n = len;
3424 index = s.index;
3425 tab = s.tab;
3426 s.tab = null;
3427 TableStack<K,V> next = s.next;
3428 s.next = spare; // save for reuse
3429 stack = next;
3430 spare = s;
3431 }
3432 if (s == null && (index += baseSize) >= n)
3433 index = ++baseIndex;
3434 }
3435 }
3436
3437 /**
3438 * Base of key, value, and entry Iterators. Adds fields to
3439 * Traverser to support iterator.remove.
3440 */
3441 static class BaseIterator<K,V> extends Traverser<K,V> {
3442 final ConcurrentHashMap<K,V> map;
3443 Node<K,V> lastReturned;
3444 BaseIterator(Node<K,V>[] tab, int size, int index, int limit,
3445 ConcurrentHashMap<K,V> map) {
3446 super(tab, size, index, limit);
3447 this.map = map;
3448 advance();
3449 }
3450
3451 public final boolean hasNext() { return next != null; }
3452 public final boolean hasMoreElements() { return next != null; }
3453
3454 public final void remove() {
3455 Node<K,V> p;
3456 if ((p = lastReturned) == null)
3457 throw new IllegalStateException();
3458 lastReturned = null;
3459 map.replaceNode(p.key, null, null);
3460 }
3461 }
3462
3463 static final class KeyIterator<K,V> extends BaseIterator<K,V>
3464 implements Iterator<K>, Enumeration<K> {
3465 KeyIterator(Node<K,V>[] tab, int size, int index, int limit,
3466 ConcurrentHashMap<K,V> map) {
3467 super(tab, size, index, limit, map);
3468 }
3469
3470 public final K next() {
3471 Node<K,V> p;
3472 if ((p = next) == null)
3473 throw new NoSuchElementException();
3474 K k = p.key;
3475 lastReturned = p;
3476 advance();
3477 return k;
3478 }
3479
3480 public final K nextElement() { return next(); }
3481 }
3482
3483 static final class ValueIterator<K,V> extends BaseIterator<K,V>
3484 implements Iterator<V>, Enumeration<V> {
3485 ValueIterator(Node<K,V>[] tab, int size, int index, int limit,
3486 ConcurrentHashMap<K,V> map) {
3487 super(tab, size, index, limit, map);
3488 }
3489
3490 public final V next() {
3491 Node<K,V> p;
3492 if ((p = next) == null)
3493 throw new NoSuchElementException();
3494 V v = p.val;
3495 lastReturned = p;
3496 advance();
3497 return v;
3498 }
3499
3500 public final V nextElement() { return next(); }
3501 }
3502
3503 static final class EntryIterator<K,V> extends BaseIterator<K,V>
3504 implements Iterator<Map.Entry<K,V>> {
3505 EntryIterator(Node<K,V>[] tab, int size, int index, int limit,
3506 ConcurrentHashMap<K,V> map) {
3507 super(tab, size, index, limit, map);
3508 }
3509
3510 public final Map.Entry<K,V> next() {
3511 Node<K,V> p;
3512 if ((p = next) == null)
3513 throw new NoSuchElementException();
3514 K k = p.key;
3515 V v = p.val;
3516 lastReturned = p;
3517 advance();
3518 return new MapEntry<K,V>(k, v, map);
3519 }
3520 }
3521
3522 /**
3523 * Exported Entry for EntryIterator.
3524 */
3525 static final class MapEntry<K,V> implements Map.Entry<K,V> {
3526 final K key; // non-null
3527 V val; // non-null
3528 final ConcurrentHashMap<K,V> map;
3529 MapEntry(K key, V val, ConcurrentHashMap<K,V> map) {
3530 this.key = key;
3531 this.val = val;
3532 this.map = map;
3533 }
3534 public K getKey() { return key; }
3535 public V getValue() { return val; }
3536 public int hashCode() { return key.hashCode() ^ val.hashCode(); }
3537 public String toString() {
3538 return Helpers.mapEntryToString(key, val);
3539 }
3540
3541 public boolean equals(Object o) {
3542 Object k, v; Map.Entry<?,?> e;
3543 return ((o instanceof Map.Entry) &&
3544 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
3545 (v = e.getValue()) != null &&
3546 (k == key || k.equals(key)) &&
3547 (v == val || v.equals(val)));
3548 }
3549
3550 /**
3551 * Sets our entry's value and writes through to the map. The
3552 * value to return is somewhat arbitrary here. Since we do not
3553 * necessarily track asynchronous changes, the most recent
3554 * "previous" value could be different from what we return (or
3555 * could even have been removed, in which case the put will
3556 * re-establish). We do not and cannot guarantee more.
3557 */
3558 public V setValue(V value) {
3559 if (value == null) throw new NullPointerException();
3560 V v = val;
3561 val = value;
3562 map.put(key, value);
3563 return v;
3564 }
3565 }
3566
3567 static final class KeySpliterator<K,V> extends Traverser<K,V>
3568 implements Spliterator<K> {
3569 long est; // size estimate
3570 KeySpliterator(Node<K,V>[] tab, int size, int index, int limit,
3571 long est) {
3572 super(tab, size, index, limit);
3573 this.est = est;
3574 }
3575
3576 public KeySpliterator<K,V> trySplit() {
3577 int i, f, h;
3578 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3579 new KeySpliterator<K,V>(tab, baseSize, baseLimit = h,
3580 f, est >>>= 1);
3581 }
3582
3583 public void forEachRemaining(Consumer<? super K> action) {
3584 if (action == null) throw new NullPointerException();
3585 for (Node<K,V> p; (p = advance()) != null;)
3586 action.accept(p.key);
3587 }
3588
3589 public boolean tryAdvance(Consumer<? super K> action) {
3590 if (action == null) throw new NullPointerException();
3591 Node<K,V> p;
3592 if ((p = advance()) == null)
3593 return false;
3594 action.accept(p.key);
3595 return true;
3596 }
3597
3598 public long estimateSize() { return est; }
3599
3600 public int characteristics() {
3601 return Spliterator.DISTINCT | Spliterator.CONCURRENT |
3602 Spliterator.NONNULL;
3603 }
3604 }
3605
3606 static final class ValueSpliterator<K,V> extends Traverser<K,V>
3607 implements Spliterator<V> {
3608 long est; // size estimate
3609 ValueSpliterator(Node<K,V>[] tab, int size, int index, int limit,
3610 long est) {
3611 super(tab, size, index, limit);
3612 this.est = est;
3613 }
3614
3615 public ValueSpliterator<K,V> trySplit() {
3616 int i, f, h;
3617 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3618 new ValueSpliterator<K,V>(tab, baseSize, baseLimit = h,
3619 f, est >>>= 1);
3620 }
3621
3622 public void forEachRemaining(Consumer<? super V> action) {
3623 if (action == null) throw new NullPointerException();
3624 for (Node<K,V> p; (p = advance()) != null;)
3625 action.accept(p.val);
3626 }
3627
3628 public boolean tryAdvance(Consumer<? super V> action) {
3629 if (action == null) throw new NullPointerException();
3630 Node<K,V> p;
3631 if ((p = advance()) == null)
3632 return false;
3633 action.accept(p.val);
3634 return true;
3635 }
3636
3637 public long estimateSize() { return est; }
3638
3639 public int characteristics() {
3640 return Spliterator.CONCURRENT | Spliterator.NONNULL;
3641 }
3642 }
3643
3644 static final class EntrySpliterator<K,V> extends Traverser<K,V>
3645 implements Spliterator<Map.Entry<K,V>> {
3646 final ConcurrentHashMap<K,V> map; // To export MapEntry
3647 long est; // size estimate
3648 EntrySpliterator(Node<K,V>[] tab, int size, int index, int limit,
3649 long est, ConcurrentHashMap<K,V> map) {
3650 super(tab, size, index, limit);
3651 this.map = map;
3652 this.est = est;
3653 }
3654
3655 public EntrySpliterator<K,V> trySplit() {
3656 int i, f, h;
3657 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3658 new EntrySpliterator<K,V>(tab, baseSize, baseLimit = h,
3659 f, est >>>= 1, map);
3660 }
3661
3662 public void forEachRemaining(Consumer<? super Map.Entry<K,V>> action) {
3663 if (action == null) throw new NullPointerException();
3664 for (Node<K,V> p; (p = advance()) != null; )
3665 action.accept(new MapEntry<K,V>(p.key, p.val, map));
3666 }
3667
3668 public boolean tryAdvance(Consumer<? super Map.Entry<K,V>> action) {
3669 if (action == null) throw new NullPointerException();
3670 Node<K,V> p;
3671 if ((p = advance()) == null)
3672 return false;
3673 action.accept(new MapEntry<K,V>(p.key, p.val, map));
3674 return true;
3675 }
3676
3677 public long estimateSize() { return est; }
3678
3679 public int characteristics() {
3680 return Spliterator.DISTINCT | Spliterator.CONCURRENT |
3681 Spliterator.NONNULL;
3682 }
3683 }
3684
3685 // Parallel bulk operations
3686
3687 /**
3688 * Computes initial batch value for bulk tasks. The returned value
3689 * is approximately exp2 of the number of times (minus one) to
3690 * split task by two before executing leaf action. This value is
3691 * faster to compute and more convenient to use as a guide to
3692 * splitting than is the depth, since it is used while dividing by
3693 * two anyway.
3694 */
3695 final int batchFor(long b) {
3696 long n;
3697 if (b == Long.MAX_VALUE || (n = sumCount()) <= 1L || n < b)
3698 return 0;
3699 int sp = ForkJoinPool.getCommonPoolParallelism() << 2; // slack of 4
3700 return (b <= 0L || (n /= b) >= sp) ? sp : (int)n;
3701 }
3702
3703 /**
3704 * Performs the given action for each (key, value).
3705 *
3706 * @param parallelismThreshold the (estimated) number of elements
3707 * needed for this operation to be executed in parallel
3708 * @param action the action
3709 * @since 1.8
3710 */
3711 public void forEach(long parallelismThreshold,
3712 BiConsumer<? super K,? super V> action) {
3713 if (action == null) throw new NullPointerException();
3714 new ForEachMappingTask<K,V>
3715 (null, batchFor(parallelismThreshold), 0, 0, table,
3716 action).invoke();
3717 }
3718
3719 /**
3720 * Performs the given action for each non-null transformation
3721 * of each (key, value).
3722 *
3723 * @param parallelismThreshold the (estimated) number of elements
3724 * needed for this operation to be executed in parallel
3725 * @param transformer a function returning the transformation
3726 * for an element, or null if there is no transformation (in
3727 * which case the action is not applied)
3728 * @param action the action
3729 * @param <U> the return type of the transformer
3730 * @since 1.8
3731 */
3732 public <U> void forEach(long parallelismThreshold,
3733 BiFunction<? super K, ? super V, ? extends U> transformer,
3734 Consumer<? super U> action) {
3735 if (transformer == null || action == null)
3736 throw new NullPointerException();
3737 new ForEachTransformedMappingTask<K,V,U>
3738 (null, batchFor(parallelismThreshold), 0, 0, table,
3739 transformer, action).invoke();
3740 }
3741
3742 /**
3743 * Returns a non-null result from applying the given search
3744 * function on each (key, value), or null if none. Upon
3745 * success, further element processing is suppressed and the
3746 * results of any other parallel invocations of the search
3747 * function are ignored.
3748 *
3749 * @param parallelismThreshold the (estimated) number of elements
3750 * needed for this operation to be executed in parallel
3751 * @param searchFunction a function returning a non-null
3752 * result on success, else null
3753 * @param <U> the return type of the search function
3754 * @return a non-null result from applying the given search
3755 * function on each (key, value), or null if none
3756 * @since 1.8
3757 */
3758 public <U> U search(long parallelismThreshold,
3759 BiFunction<? super K, ? super V, ? extends U> searchFunction) {
3760 if (searchFunction == null) throw new NullPointerException();
3761 return new SearchMappingsTask<K,V,U>
3762 (null, batchFor(parallelismThreshold), 0, 0, table,
3763 searchFunction, new AtomicReference<U>()).invoke();
3764 }
3765
3766 /**
3767 * Returns the result of accumulating the given transformation
3768 * of all (key, value) pairs using the given reducer to
3769 * combine values, or null if none.
3770 *
3771 * @param parallelismThreshold the (estimated) number of elements
3772 * needed for this operation to be executed in parallel
3773 * @param transformer a function returning the transformation
3774 * for an element, or null if there is no transformation (in
3775 * which case it is not combined)
3776 * @param reducer a commutative associative combining function
3777 * @param <U> the return type of the transformer
3778 * @return the result of accumulating the given transformation
3779 * of all (key, value) pairs
3780 * @since 1.8
3781 */
3782 public <U> U reduce(long parallelismThreshold,
3783 BiFunction<? super K, ? super V, ? extends U> transformer,
3784 BiFunction<? super U, ? super U, ? extends U> reducer) {
3785 if (transformer == null || reducer == null)
3786 throw new NullPointerException();
3787 return new MapReduceMappingsTask<K,V,U>
3788 (null, batchFor(parallelismThreshold), 0, 0, table,
3789 null, transformer, reducer).invoke();
3790 }
3791
3792 /**
3793 * Returns the result of accumulating the given transformation
3794 * of all (key, value) pairs using the given reducer to
3795 * combine values, and the given basis as an identity value.
3796 *
3797 * @param parallelismThreshold the (estimated) number of elements
3798 * needed for this operation to be executed in parallel
3799 * @param transformer a function returning the transformation
3800 * for an element
3801 * @param basis the identity (initial default value) for the reduction
3802 * @param reducer a commutative associative combining function
3803 * @return the result of accumulating the given transformation
3804 * of all (key, value) pairs
3805 * @since 1.8
3806 */
3807 public double reduceToDouble(long parallelismThreshold,
3808 ToDoubleBiFunction<? super K, ? super V> transformer,
3809 double basis,
3810 DoubleBinaryOperator reducer) {
3811 if (transformer == null || reducer == null)
3812 throw new NullPointerException();
3813 return new MapReduceMappingsToDoubleTask<K,V>
3814 (null, batchFor(parallelismThreshold), 0, 0, table,
3815 null, transformer, basis, reducer).invoke();
3816 }
3817
3818 /**
3819 * Returns the result of accumulating the given transformation
3820 * of all (key, value) pairs using the given reducer to
3821 * combine values, and the given basis as an identity value.
3822 *
3823 * @param parallelismThreshold the (estimated) number of elements
3824 * needed for this operation to be executed in parallel
3825 * @param transformer a function returning the transformation
3826 * for an element
3827 * @param basis the identity (initial default value) for the reduction
3828 * @param reducer a commutative associative combining function
3829 * @return the result of accumulating the given transformation
3830 * of all (key, value) pairs
3831 * @since 1.8
3832 */
3833 public long reduceToLong(long parallelismThreshold,
3834 ToLongBiFunction<? super K, ? super V> transformer,
3835 long basis,
3836 LongBinaryOperator reducer) {
3837 if (transformer == null || reducer == null)
3838 throw new NullPointerException();
3839 return new MapReduceMappingsToLongTask<K,V>
3840 (null, batchFor(parallelismThreshold), 0, 0, table,
3841 null, transformer, basis, reducer).invoke();
3842 }
3843
3844 /**
3845 * Returns the result of accumulating the given transformation
3846 * of all (key, value) pairs using the given reducer to
3847 * combine values, and the given basis as an identity value.
3848 *
3849 * @param parallelismThreshold the (estimated) number of elements
3850 * needed for this operation to be executed in parallel
3851 * @param transformer a function returning the transformation
3852 * for an element
3853 * @param basis the identity (initial default value) for the reduction
3854 * @param reducer a commutative associative combining function
3855 * @return the result of accumulating the given transformation
3856 * of all (key, value) pairs
3857 * @since 1.8
3858 */
3859 public int reduceToInt(long parallelismThreshold,
3860 ToIntBiFunction<? super K, ? super V> transformer,
3861 int basis,
3862 IntBinaryOperator reducer) {
3863 if (transformer == null || reducer == null)
3864 throw new NullPointerException();
3865 return new MapReduceMappingsToIntTask<K,V>
3866 (null, batchFor(parallelismThreshold), 0, 0, table,
3867 null, transformer, basis, reducer).invoke();
3868 }
3869
3870 /**
3871 * Performs the given action for each key.
3872 *
3873 * @param parallelismThreshold the (estimated) number of elements
3874 * needed for this operation to be executed in parallel
3875 * @param action the action
3876 * @since 1.8
3877 */
3878 public void forEachKey(long parallelismThreshold,
3879 Consumer<? super K> action) {
3880 if (action == null) throw new NullPointerException();
3881 new ForEachKeyTask<K,V>
3882 (null, batchFor(parallelismThreshold), 0, 0, table,
3883 action).invoke();
3884 }
3885
3886 /**
3887 * Performs the given action for each non-null transformation
3888 * of each key.
3889 *
3890 * @param parallelismThreshold the (estimated) number of elements
3891 * needed for this operation to be executed in parallel
3892 * @param transformer a function returning the transformation
3893 * for an element, or null if there is no transformation (in
3894 * which case the action is not applied)
3895 * @param action the action
3896 * @param <U> the return type of the transformer
3897 * @since 1.8
3898 */
3899 public <U> void forEachKey(long parallelismThreshold,
3900 Function<? super K, ? extends U> transformer,
3901 Consumer<? super U> action) {
3902 if (transformer == null || action == null)
3903 throw new NullPointerException();
3904 new ForEachTransformedKeyTask<K,V,U>
3905 (null, batchFor(parallelismThreshold), 0, 0, table,
3906 transformer, action).invoke();
3907 }
3908
3909 /**
3910 * Returns a non-null result from applying the given search
3911 * function on each key, or null if none. Upon success,
3912 * further element processing is suppressed and the results of
3913 * any other parallel invocations of the search function are
3914 * ignored.
3915 *
3916 * @param parallelismThreshold the (estimated) number of elements
3917 * needed for this operation to be executed in parallel
3918 * @param searchFunction a function returning a non-null
3919 * result on success, else null
3920 * @param <U> the return type of the search function
3921 * @return a non-null result from applying the given search
3922 * function on each key, or null if none
3923 * @since 1.8
3924 */
3925 public <U> U searchKeys(long parallelismThreshold,
3926 Function<? super K, ? extends U> searchFunction) {
3927 if (searchFunction == null) throw new NullPointerException();
3928 return new SearchKeysTask<K,V,U>
3929 (null, batchFor(parallelismThreshold), 0, 0, table,
3930 searchFunction, new AtomicReference<U>()).invoke();
3931 }
3932
3933 /**
3934 * Returns the result of accumulating all keys using the given
3935 * reducer to combine values, or null if none.
3936 *
3937 * @param parallelismThreshold the (estimated) number of elements
3938 * needed for this operation to be executed in parallel
3939 * @param reducer a commutative associative combining function
3940 * @return the result of accumulating all keys using the given
3941 * reducer to combine values, or null if none
3942 * @since 1.8
3943 */
3944 public K reduceKeys(long parallelismThreshold,
3945 BiFunction<? super K, ? super K, ? extends K> reducer) {
3946 if (reducer == null) throw new NullPointerException();
3947 return new ReduceKeysTask<K,V>
3948 (null, batchFor(parallelismThreshold), 0, 0, table,
3949 null, reducer).invoke();
3950 }
3951
3952 /**
3953 * Returns the result of accumulating the given transformation
3954 * of all keys using the given reducer to combine values, or
3955 * null if none.
3956 *
3957 * @param parallelismThreshold the (estimated) number of elements
3958 * needed for this operation to be executed in parallel
3959 * @param transformer a function returning the transformation
3960 * for an element, or null if there is no transformation (in
3961 * which case it is not combined)
3962 * @param reducer a commutative associative combining function
3963 * @param <U> the return type of the transformer
3964 * @return the result of accumulating the given transformation
3965 * of all keys
3966 * @since 1.8
3967 */
3968 public <U> U reduceKeys(long parallelismThreshold,
3969 Function<? super K, ? extends U> transformer,
3970 BiFunction<? super U, ? super U, ? extends U> reducer) {
3971 if (transformer == null || reducer == null)
3972 throw new NullPointerException();
3973 return new MapReduceKeysTask<K,V,U>
3974 (null, batchFor(parallelismThreshold), 0, 0, table,
3975 null, transformer, reducer).invoke();
3976 }
3977
3978 /**
3979 * Returns the result of accumulating the given transformation
3980 * of all keys using the given reducer to combine values, and
3981 * the given basis as an identity value.
3982 *
3983 * @param parallelismThreshold the (estimated) number of elements
3984 * needed for this operation to be executed in parallel
3985 * @param transformer a function returning the transformation
3986 * for an element
3987 * @param basis the identity (initial default value) for the reduction
3988 * @param reducer a commutative associative combining function
3989 * @return the result of accumulating the given transformation
3990 * of all keys
3991 * @since 1.8
3992 */
3993 public double reduceKeysToDouble(long parallelismThreshold,
3994 ToDoubleFunction<? super K> transformer,
3995 double basis,
3996 DoubleBinaryOperator reducer) {
3997 if (transformer == null || reducer == null)
3998 throw new NullPointerException();
3999 return new MapReduceKeysToDoubleTask<K,V>
4000 (null, batchFor(parallelismThreshold), 0, 0, table,
4001 null, transformer, basis, reducer).invoke();
4002 }
4003
4004 /**
4005 * Returns the result of accumulating the given transformation
4006 * of all keys using the given reducer to combine values, and
4007 * the given basis as an identity value.
4008 *
4009 * @param parallelismThreshold the (estimated) number of elements
4010 * needed for this operation to be executed in parallel
4011 * @param transformer a function returning the transformation
4012 * for an element
4013 * @param basis the identity (initial default value) for the reduction
4014 * @param reducer a commutative associative combining function
4015 * @return the result of accumulating the given transformation
4016 * of all keys
4017 * @since 1.8
4018 */
4019 public long reduceKeysToLong(long parallelismThreshold,
4020 ToLongFunction<? super K> transformer,
4021 long basis,
4022 LongBinaryOperator reducer) {
4023 if (transformer == null || reducer == null)
4024 throw new NullPointerException();
4025 return new MapReduceKeysToLongTask<K,V>
4026 (null, batchFor(parallelismThreshold), 0, 0, table,
4027 null, transformer, basis, reducer).invoke();
4028 }
4029
4030 /**
4031 * Returns the result of accumulating the given transformation
4032 * of all keys using the given reducer to combine values, and
4033 * the given basis as an identity value.
4034 *
4035 * @param parallelismThreshold the (estimated) number of elements
4036 * needed for this operation to be executed in parallel
4037 * @param transformer a function returning the transformation
4038 * for an element
4039 * @param basis the identity (initial default value) for the reduction
4040 * @param reducer a commutative associative combining function
4041 * @return the result of accumulating the given transformation
4042 * of all keys
4043 * @since 1.8
4044 */
4045 public int reduceKeysToInt(long parallelismThreshold,
4046 ToIntFunction<? super K> transformer,
4047 int basis,
4048 IntBinaryOperator reducer) {
4049 if (transformer == null || reducer == null)
4050 throw new NullPointerException();
4051 return new MapReduceKeysToIntTask<K,V>
4052 (null, batchFor(parallelismThreshold), 0, 0, table,
4053 null, transformer, basis, reducer).invoke();
4054 }
4055
4056 /**
4057 * Performs the given action for each value.
4058 *
4059 * @param parallelismThreshold the (estimated) number of elements
4060 * needed for this operation to be executed in parallel
4061 * @param action the action
4062 * @since 1.8
4063 */
4064 public void forEachValue(long parallelismThreshold,
4065 Consumer<? super V> action) {
4066 if (action == null)
4067 throw new NullPointerException();
4068 new ForEachValueTask<K,V>
4069 (null, batchFor(parallelismThreshold), 0, 0, table,
4070 action).invoke();
4071 }
4072
4073 /**
4074 * Performs the given action for each non-null transformation
4075 * of each value.
4076 *
4077 * @param parallelismThreshold the (estimated) number of elements
4078 * needed for this operation to be executed in parallel
4079 * @param transformer a function returning the transformation
4080 * for an element, or null if there is no transformation (in
4081 * which case the action is not applied)
4082 * @param action the action
4083 * @param <U> the return type of the transformer
4084 * @since 1.8
4085 */
4086 public <U> void forEachValue(long parallelismThreshold,
4087 Function<? super V, ? extends U> transformer,
4088 Consumer<? super U> action) {
4089 if (transformer == null || action == null)
4090 throw new NullPointerException();
4091 new ForEachTransformedValueTask<K,V,U>
4092 (null, batchFor(parallelismThreshold), 0, 0, table,
4093 transformer, action).invoke();
4094 }
4095
4096 /**
4097 * Returns a non-null result from applying the given search
4098 * function on each value, or null if none. Upon success,
4099 * further element processing is suppressed and the results of
4100 * any other parallel invocations of the search function are
4101 * ignored.
4102 *
4103 * @param parallelismThreshold the (estimated) number of elements
4104 * needed for this operation to be executed in parallel
4105 * @param searchFunction a function returning a non-null
4106 * result on success, else null
4107 * @param <U> the return type of the search function
4108 * @return a non-null result from applying the given search
4109 * function on each value, or null if none
4110 * @since 1.8
4111 */
4112 public <U> U searchValues(long parallelismThreshold,
4113 Function<? super V, ? extends U> searchFunction) {
4114 if (searchFunction == null) throw new NullPointerException();
4115 return new SearchValuesTask<K,V,U>
4116 (null, batchFor(parallelismThreshold), 0, 0, table,
4117 searchFunction, new AtomicReference<U>()).invoke();
4118 }
4119
4120 /**
4121 * Returns the result of accumulating all values using the
4122 * given reducer to combine values, or null if none.
4123 *
4124 * @param parallelismThreshold the (estimated) number of elements
4125 * needed for this operation to be executed in parallel
4126 * @param reducer a commutative associative combining function
4127 * @return the result of accumulating all values
4128 * @since 1.8
4129 */
4130 public V reduceValues(long parallelismThreshold,
4131 BiFunction<? super V, ? super V, ? extends V> reducer) {
4132 if (reducer == null) throw new NullPointerException();
4133 return new ReduceValuesTask<K,V>
4134 (null, batchFor(parallelismThreshold), 0, 0, table,
4135 null, reducer).invoke();
4136 }
4137
4138 /**
4139 * Returns the result of accumulating the given transformation
4140 * of all values using the given reducer to combine values, or
4141 * null if none.
4142 *
4143 * @param parallelismThreshold the (estimated) number of elements
4144 * needed for this operation to be executed in parallel
4145 * @param transformer a function returning the transformation
4146 * for an element, or null if there is no transformation (in
4147 * which case it is not combined)
4148 * @param reducer a commutative associative combining function
4149 * @param <U> the return type of the transformer
4150 * @return the result of accumulating the given transformation
4151 * of all values
4152 * @since 1.8
4153 */
4154 public <U> U reduceValues(long parallelismThreshold,
4155 Function<? super V, ? extends U> transformer,
4156 BiFunction<? super U, ? super U, ? extends U> reducer) {
4157 if (transformer == null || reducer == null)
4158 throw new NullPointerException();
4159 return new MapReduceValuesTask<K,V,U>
4160 (null, batchFor(parallelismThreshold), 0, 0, table,
4161 null, transformer, reducer).invoke();
4162 }
4163
4164 /**
4165 * Returns the result of accumulating the given transformation
4166 * of all values using the given reducer to combine values,
4167 * and the given basis as an identity value.
4168 *
4169 * @param parallelismThreshold the (estimated) number of elements
4170 * needed for this operation to be executed in parallel
4171 * @param transformer a function returning the transformation
4172 * for an element
4173 * @param basis the identity (initial default value) for the reduction
4174 * @param reducer a commutative associative combining function
4175 * @return the result of accumulating the given transformation
4176 * of all values
4177 * @since 1.8
4178 */
4179 public double reduceValuesToDouble(long parallelismThreshold,
4180 ToDoubleFunction<? super V> transformer,
4181 double basis,
4182 DoubleBinaryOperator reducer) {
4183 if (transformer == null || reducer == null)
4184 throw new NullPointerException();
4185 return new MapReduceValuesToDoubleTask<K,V>
4186 (null, batchFor(parallelismThreshold), 0, 0, table,
4187 null, transformer, basis, reducer).invoke();
4188 }
4189
4190 /**
4191 * Returns the result of accumulating the given transformation
4192 * of all values using the given reducer to combine values,
4193 * and the given basis as an identity value.
4194 *
4195 * @param parallelismThreshold the (estimated) number of elements
4196 * needed for this operation to be executed in parallel
4197 * @param transformer a function returning the transformation
4198 * for an element
4199 * @param basis the identity (initial default value) for the reduction
4200 * @param reducer a commutative associative combining function
4201 * @return the result of accumulating the given transformation
4202 * of all values
4203 * @since 1.8
4204 */
4205 public long reduceValuesToLong(long parallelismThreshold,
4206 ToLongFunction<? super V> transformer,
4207 long basis,
4208 LongBinaryOperator reducer) {
4209 if (transformer == null || reducer == null)
4210 throw new NullPointerException();
4211 return new MapReduceValuesToLongTask<K,V>
4212 (null, batchFor(parallelismThreshold), 0, 0, table,
4213 null, transformer, basis, reducer).invoke();
4214 }
4215
4216 /**
4217 * Returns the result of accumulating the given transformation
4218 * of all values using the given reducer to combine values,
4219 * and the given basis as an identity value.
4220 *
4221 * @param parallelismThreshold the (estimated) number of elements
4222 * needed for this operation to be executed in parallel
4223 * @param transformer a function returning the transformation
4224 * for an element
4225 * @param basis the identity (initial default value) for the reduction
4226 * @param reducer a commutative associative combining function
4227 * @return the result of accumulating the given transformation
4228 * of all values
4229 * @since 1.8
4230 */
4231 public int reduceValuesToInt(long parallelismThreshold,
4232 ToIntFunction<? super V> transformer,
4233 int basis,
4234 IntBinaryOperator reducer) {
4235 if (transformer == null || reducer == null)
4236 throw new NullPointerException();
4237 return new MapReduceValuesToIntTask<K,V>
4238 (null, batchFor(parallelismThreshold), 0, 0, table,
4239 null, transformer, basis, reducer).invoke();
4240 }
4241
4242 /**
4243 * Performs the given action for each entry.
4244 *
4245 * @param parallelismThreshold the (estimated) number of elements
4246 * needed for this operation to be executed in parallel
4247 * @param action the action
4248 * @since 1.8
4249 */
4250 public void forEachEntry(long parallelismThreshold,
4251 Consumer<? super Map.Entry<K,V>> action) {
4252 if (action == null) throw new NullPointerException();
4253 new ForEachEntryTask<K,V>(null, batchFor(parallelismThreshold), 0, 0, table,
4254 action).invoke();
4255 }
4256
4257 /**
4258 * Performs the given action for each non-null transformation
4259 * of each entry.
4260 *
4261 * @param parallelismThreshold the (estimated) number of elements
4262 * needed for this operation to be executed in parallel
4263 * @param transformer a function returning the transformation
4264 * for an element, or null if there is no transformation (in
4265 * which case the action is not applied)
4266 * @param action the action
4267 * @param <U> the return type of the transformer
4268 * @since 1.8
4269 */
4270 public <U> void forEachEntry(long parallelismThreshold,
4271 Function<Map.Entry<K,V>, ? extends U> transformer,
4272 Consumer<? super U> action) {
4273 if (transformer == null || action == null)
4274 throw new NullPointerException();
4275 new ForEachTransformedEntryTask<K,V,U>
4276 (null, batchFor(parallelismThreshold), 0, 0, table,
4277 transformer, action).invoke();
4278 }
4279
4280 /**
4281 * Returns a non-null result from applying the given search
4282 * function on each entry, or null if none. Upon success,
4283 * further element processing is suppressed and the results of
4284 * any other parallel invocations of the search function are
4285 * ignored.
4286 *
4287 * @param parallelismThreshold the (estimated) number of elements
4288 * needed for this operation to be executed in parallel
4289 * @param searchFunction a function returning a non-null
4290 * result on success, else null
4291 * @param <U> the return type of the search function
4292 * @return a non-null result from applying the given search
4293 * function on each entry, or null if none
4294 * @since 1.8
4295 */
4296 public <U> U searchEntries(long parallelismThreshold,
4297 Function<Map.Entry<K,V>, ? extends U> searchFunction) {
4298 if (searchFunction == null) throw new NullPointerException();
4299 return new SearchEntriesTask<K,V,U>
4300 (null, batchFor(parallelismThreshold), 0, 0, table,
4301 searchFunction, new AtomicReference<U>()).invoke();
4302 }
4303
4304 /**
4305 * Returns the result of accumulating all entries using the
4306 * given reducer to combine values, or null if none.
4307 *
4308 * @param parallelismThreshold the (estimated) number of elements
4309 * needed for this operation to be executed in parallel
4310 * @param reducer a commutative associative combining function
4311 * @return the result of accumulating all entries
4312 * @since 1.8
4313 */
4314 public Map.Entry<K,V> reduceEntries(long parallelismThreshold,
4315 BiFunction<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer) {
4316 if (reducer == null) throw new NullPointerException();
4317 return new ReduceEntriesTask<K,V>
4318 (null, batchFor(parallelismThreshold), 0, 0, table,
4319 null, reducer).invoke();
4320 }
4321
4322 /**
4323 * Returns the result of accumulating the given transformation
4324 * of all entries using the given reducer to combine values,
4325 * or null if none.
4326 *
4327 * @param parallelismThreshold the (estimated) number of elements
4328 * needed for this operation to be executed in parallel
4329 * @param transformer a function returning the transformation
4330 * for an element, or null if there is no transformation (in
4331 * which case it is not combined)
4332 * @param reducer a commutative associative combining function
4333 * @param <U> the return type of the transformer
4334 * @return the result of accumulating the given transformation
4335 * of all entries
4336 * @since 1.8
4337 */
4338 public <U> U reduceEntries(long parallelismThreshold,
4339 Function<Map.Entry<K,V>, ? extends U> transformer,
4340 BiFunction<? super U, ? super U, ? extends U> reducer) {
4341 if (transformer == null || reducer == null)
4342 throw new NullPointerException();
4343 return new MapReduceEntriesTask<K,V,U>
4344 (null, batchFor(parallelismThreshold), 0, 0, table,
4345 null, transformer, reducer).invoke();
4346 }
4347
4348 /**
4349 * Returns the result of accumulating the given transformation
4350 * of all entries using the given reducer to combine values,
4351 * and the given basis as an identity value.
4352 *
4353 * @param parallelismThreshold the (estimated) number of elements
4354 * needed for this operation to be executed in parallel
4355 * @param transformer a function returning the transformation
4356 * for an element
4357 * @param basis the identity (initial default value) for the reduction
4358 * @param reducer a commutative associative combining function
4359 * @return the result of accumulating the given transformation
4360 * of all entries
4361 * @since 1.8
4362 */
4363 public double reduceEntriesToDouble(long parallelismThreshold,
4364 ToDoubleFunction<Map.Entry<K,V>> transformer,
4365 double basis,
4366 DoubleBinaryOperator reducer) {
4367 if (transformer == null || reducer == null)
4368 throw new NullPointerException();
4369 return new MapReduceEntriesToDoubleTask<K,V>
4370 (null, batchFor(parallelismThreshold), 0, 0, table,
4371 null, transformer, basis, reducer).invoke();
4372 }
4373
4374 /**
4375 * Returns the result of accumulating the given transformation
4376 * of all entries using the given reducer to combine values,
4377 * and the given basis as an identity value.
4378 *
4379 * @param parallelismThreshold the (estimated) number of elements
4380 * needed for this operation to be executed in parallel
4381 * @param transformer a function returning the transformation
4382 * for an element
4383 * @param basis the identity (initial default value) for the reduction
4384 * @param reducer a commutative associative combining function
4385 * @return the result of accumulating the given transformation
4386 * of all entries
4387 * @since 1.8
4388 */
4389 public long reduceEntriesToLong(long parallelismThreshold,
4390 ToLongFunction<Map.Entry<K,V>> transformer,
4391 long basis,
4392 LongBinaryOperator reducer) {
4393 if (transformer == null || reducer == null)
4394 throw new NullPointerException();
4395 return new MapReduceEntriesToLongTask<K,V>
4396 (null, batchFor(parallelismThreshold), 0, 0, table,
4397 null, transformer, basis, reducer).invoke();
4398 }
4399
4400 /**
4401 * Returns the result of accumulating the given transformation
4402 * of all entries using the given reducer to combine values,
4403 * and the given basis as an identity value.
4404 *
4405 * @param parallelismThreshold the (estimated) number of elements
4406 * needed for this operation to be executed in parallel
4407 * @param transformer a function returning the transformation
4408 * for an element
4409 * @param basis the identity (initial default value) for the reduction
4410 * @param reducer a commutative associative combining function
4411 * @return the result of accumulating the given transformation
4412 * of all entries
4413 * @since 1.8
4414 */
4415 public int reduceEntriesToInt(long parallelismThreshold,
4416 ToIntFunction<Map.Entry<K,V>> transformer,
4417 int basis,
4418 IntBinaryOperator reducer) {
4419 if (transformer == null || reducer == null)
4420 throw new NullPointerException();
4421 return new MapReduceEntriesToIntTask<K,V>
4422 (null, batchFor(parallelismThreshold), 0, 0, table,
4423 null, transformer, basis, reducer).invoke();
4424 }
4425
4426
4427 /* ----------------Views -------------- */
4428
4429 /**
4430 * Base class for views.
4431 */
4432 abstract static class CollectionView<K,V,E>
4433 implements Collection<E>, java.io.Serializable {
4434 private static final long serialVersionUID = 7249069246763182397L;
4435 final ConcurrentHashMap<K,V> map;
4436 CollectionView(ConcurrentHashMap<K,V> map) { this.map = map; }
4437
4438 /**
4439 * Returns the map backing this view.
4440 *
4441 * @return the map backing this view
4442 */
4443 public ConcurrentHashMap<K,V> getMap() { return map; }
4444
4445 /**
4446 * Removes all of the elements from this view, by removing all
4447 * the mappings from the map backing this view.
4448 */
4449 public final void clear() { map.clear(); }
4450 public final int size() { return map.size(); }
4451 public final boolean isEmpty() { return map.isEmpty(); }
4452
4453 // implementations below rely on concrete classes supplying these
4454 // abstract methods
4455 /**
4456 * Returns an iterator over the elements in this collection.
4457 *
4458 * <p>The returned iterator is
4459 * <a href="package-summary.html#Weakly"><i>weakly consistent</i></a>.
4460 *
4461 * @return an iterator over the elements in this collection
4462 */
4463 public abstract Iterator<E> iterator();
4464 public abstract boolean contains(Object o);
4465 public abstract boolean remove(Object o);
4466
4467 private static final String OOME_MSG = "Required array size too large";
4468
4469 public final Object[] toArray() {
4470 long sz = map.mappingCount();
4471 if (sz > MAX_ARRAY_SIZE)
4472 throw new OutOfMemoryError(OOME_MSG);
4473 int n = (int)sz;
4474 Object[] r = new Object[n];
4475 int i = 0;
4476 for (E e : this) {
4477 if (i == n) {
4478 if (n >= MAX_ARRAY_SIZE)
4479 throw new OutOfMemoryError(OOME_MSG);
4480 if (n >= MAX_ARRAY_SIZE - (MAX_ARRAY_SIZE >>> 1) - 1)
4481 n = MAX_ARRAY_SIZE;
4482 else
4483 n += (n >>> 1) + 1;
4484 r = Arrays.copyOf(r, n);
4485 }
4486 r[i++] = e;
4487 }
4488 return (i == n) ? r : Arrays.copyOf(r, i);
4489 }
4490
4491 @SuppressWarnings("unchecked")
4492 public final <T> T[] toArray(T[] a) {
4493 long sz = map.mappingCount();
4494 if (sz > MAX_ARRAY_SIZE)
4495 throw new OutOfMemoryError(OOME_MSG);
4496 int m = (int)sz;
4497 T[] r = (a.length >= m) ? a :
4498 (T[])java.lang.reflect.Array
4499 .newInstance(a.getClass().getComponentType(), m);
4500 int n = r.length;
4501 int i = 0;
4502 for (E e : this) {
4503 if (i == n) {
4504 if (n >= MAX_ARRAY_SIZE)
4505 throw new OutOfMemoryError(OOME_MSG);
4506 if (n >= MAX_ARRAY_SIZE - (MAX_ARRAY_SIZE >>> 1) - 1)
4507 n = MAX_ARRAY_SIZE;
4508 else
4509 n += (n >>> 1) + 1;
4510 r = Arrays.copyOf(r, n);
4511 }
4512 r[i++] = (T)e;
4513 }
4514 if (a == r && i < n) {
4515 r[i] = null; // null-terminate
4516 return r;
4517 }
4518 return (i == n) ? r : Arrays.copyOf(r, i);
4519 }
4520
4521 /**
4522 * Returns a string representation of this collection.
4523 * The string representation consists of the string representations
4524 * of the collection's elements in the order they are returned by
4525 * its iterator, enclosed in square brackets ({@code "[]"}).
4526 * Adjacent elements are separated by the characters {@code ", "}
4527 * (comma and space). Elements are converted to strings as by
4528 * {@link String#valueOf(Object)}.
4529 *
4530 * @return a string representation of this collection
4531 */
4532 public final String toString() {
4533 StringBuilder sb = new StringBuilder();
4534 sb.append('[');
4535 Iterator<E> it = iterator();
4536 if (it.hasNext()) {
4537 for (;;) {
4538 Object e = it.next();
4539 sb.append(e == this ? "(this Collection)" : e);
4540 if (!it.hasNext())
4541 break;
4542 sb.append(',').append(' ');
4543 }
4544 }
4545 return sb.append(']').toString();
4546 }
4547
4548 public final boolean containsAll(Collection<?> c) {
4549 if (c != this) {
4550 for (Object e : c) {
4551 if (e == null || !contains(e))
4552 return false;
4553 }
4554 }
4555 return true;
4556 }
4557
4558 public boolean removeAll(Collection<?> c) {
4559 if (c == null) throw new NullPointerException();
4560 boolean modified = false;
4561 // Use (c instanceof Set) as a hint that lookup in c is as
4562 // efficient as this view
4563 Node<K,V>[] t;
4564 if ((t = map.table) == null) {
4565 return false;
4566 } else if (c instanceof Set<?> && c.size() > t.length) {
4567 for (Iterator<?> it = iterator(); it.hasNext(); ) {
4568 if (c.contains(it.next())) {
4569 it.remove();
4570 modified = true;
4571 }
4572 }
4573 } else {
4574 for (Object e : c)
4575 modified |= remove(e);
4576 }
4577 return modified;
4578 }
4579
4580 public final boolean retainAll(Collection<?> c) {
4581 if (c == null) throw new NullPointerException();
4582 boolean modified = false;
4583 for (Iterator<E> it = iterator(); it.hasNext();) {
4584 if (!c.contains(it.next())) {
4585 it.remove();
4586 modified = true;
4587 }
4588 }
4589 return modified;
4590 }
4591
4592 }
4593
4594 /**
4595 * A view of a ConcurrentHashMap as a {@link Set} of keys, in
4596 * which additions may optionally be enabled by mapping to a
4597 * common value. This class cannot be directly instantiated.
4598 * See {@link #keySet() keySet()},
4599 * {@link #keySet(Object) keySet(V)},
4600 * {@link #newKeySet() newKeySet()},
4601 * {@link #newKeySet(int) newKeySet(int)}.
4602 *
4603 * @since 1.8
4604 */
4605 public static class KeySetView<K,V> extends CollectionView<K,V,K>
4606 implements Set<K>, java.io.Serializable {
4607 private static final long serialVersionUID = 7249069246763182397L;
4608 private final V value;
4609 KeySetView(ConcurrentHashMap<K,V> map, V value) { // non-public
4610 super(map);
4611 this.value = value;
4612 }
4613
4614 /**
4615 * Returns the default mapped value for additions,
4616 * or {@code null} if additions are not supported.
4617 *
4618 * @return the default mapped value for additions, or {@code null}
4619 * if not supported
4620 */
4621 public V getMappedValue() { return value; }
4622
4623 /**
4624 * {@inheritDoc}
4625 * @throws NullPointerException if the specified key is null
4626 */
4627 public boolean contains(Object o) { return map.containsKey(o); }
4628
4629 /**
4630 * Removes the key from this map view, by removing the key (and its
4631 * corresponding value) from the backing map. This method does
4632 * nothing if the key is not in the map.
4633 *
4634 * @param o the key to be removed from the backing map
4635 * @return {@code true} if the backing map contained the specified key
4636 * @throws NullPointerException if the specified key is null
4637 */
4638 public boolean remove(Object o) { return map.remove(o) != null; }
4639
4640 /**
4641 * @return an iterator over the keys of the backing map
4642 */
4643 public Iterator<K> iterator() {
4644 Node<K,V>[] t;
4645 ConcurrentHashMap<K,V> m = map;
4646 int f = (t = m.table) == null ? 0 : t.length;
4647 return new KeyIterator<K,V>(t, f, 0, f, m);
4648 }
4649
4650 /**
4651 * Adds the specified key to this set view by mapping the key to
4652 * the default mapped value in the backing map, if defined.
4653 *
4654 * @param e key to be added
4655 * @return {@code true} if this set changed as a result of the call
4656 * @throws NullPointerException if the specified key is null
4657 * @throws UnsupportedOperationException if no default mapped value
4658 * for additions was provided
4659 */
4660 public boolean add(K e) {
4661 V v;
4662 if ((v = value) == null)
4663 throw new UnsupportedOperationException();
4664 return map.putVal(e, v, true) == null;
4665 }
4666
4667 /**
4668 * Adds all of the elements in the specified collection to this set,
4669 * as if by calling {@link #add} on each one.
4670 *
4671 * @param c the elements to be inserted into this set
4672 * @return {@code true} if this set changed as a result of the call
4673 * @throws NullPointerException if the collection or any of its
4674 * elements are {@code null}
4675 * @throws UnsupportedOperationException if no default mapped value
4676 * for additions was provided
4677 */
4678 public boolean addAll(Collection<? extends K> c) {
4679 boolean added = false;
4680 V v;
4681 if ((v = value) == null)
4682 throw new UnsupportedOperationException();
4683 for (K e : c) {
4684 if (map.putVal(e, v, true) == null)
4685 added = true;
4686 }
4687 return added;
4688 }
4689
4690 public int hashCode() {
4691 int h = 0;
4692 for (K e : this)
4693 h += e.hashCode();
4694 return h;
4695 }
4696
4697 public boolean equals(Object o) {
4698 Set<?> c;
4699 return ((o instanceof Set) &&
4700 ((c = (Set<?>)o) == this ||
4701 (containsAll(c) && c.containsAll(this))));
4702 }
4703
4704 public Spliterator<K> spliterator() {
4705 Node<K,V>[] t;
4706 ConcurrentHashMap<K,V> m = map;
4707 long n = m.sumCount();
4708 int f = (t = m.table) == null ? 0 : t.length;
4709 return new KeySpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n);
4710 }
4711
4712 public void forEach(Consumer<? super K> action) {
4713 if (action == null) throw new NullPointerException();
4714 Node<K,V>[] t;
4715 if ((t = map.table) != null) {
4716 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4717 for (Node<K,V> p; (p = it.advance()) != null; )
4718 action.accept(p.key);
4719 }
4720 }
4721 }
4722
4723 /**
4724 * A view of a ConcurrentHashMap as a {@link Collection} of
4725 * values, in which additions are disabled. This class cannot be
4726 * directly instantiated. See {@link #values()}.
4727 */
4728 static final class ValuesView<K,V> extends CollectionView<K,V,V>
4729 implements Collection<V>, java.io.Serializable {
4730 private static final long serialVersionUID = 2249069246763182397L;
4731 ValuesView(ConcurrentHashMap<K,V> map) { super(map); }
4732 public final boolean contains(Object o) {
4733 return map.containsValue(o);
4734 }
4735
4736 public final boolean remove(Object o) {
4737 if (o != null) {
4738 for (Iterator<V> it = iterator(); it.hasNext();) {
4739 if (o.equals(it.next())) {
4740 it.remove();
4741 return true;
4742 }
4743 }
4744 }
4745 return false;
4746 }
4747
4748 public final Iterator<V> iterator() {
4749 ConcurrentHashMap<K,V> m = map;
4750 Node<K,V>[] t;
4751 int f = (t = m.table) == null ? 0 : t.length;
4752 return new ValueIterator<K,V>(t, f, 0, f, m);
4753 }
4754
4755 public final boolean add(V e) {
4756 throw new UnsupportedOperationException();
4757 }
4758 public final boolean addAll(Collection<? extends V> c) {
4759 throw new UnsupportedOperationException();
4760 }
4761
4762 @Override public boolean removeAll(Collection<?> c) {
4763 if (c == null) throw new NullPointerException();
4764 boolean modified = false;
4765 for (Iterator<V> it = iterator(); it.hasNext();) {
4766 if (c.contains(it.next())) {
4767 it.remove();
4768 modified = true;
4769 }
4770 }
4771 return modified;
4772 }
4773
4774 public boolean removeIf(Predicate<? super V> filter) {
4775 return map.removeValueIf(filter);
4776 }
4777
4778 public Spliterator<V> spliterator() {
4779 Node<K,V>[] t;
4780 ConcurrentHashMap<K,V> m = map;
4781 long n = m.sumCount();
4782 int f = (t = m.table) == null ? 0 : t.length;
4783 return new ValueSpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n);
4784 }
4785
4786 public void forEach(Consumer<? super V> action) {
4787 if (action == null) throw new NullPointerException();
4788 Node<K,V>[] t;
4789 if ((t = map.table) != null) {
4790 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4791 for (Node<K,V> p; (p = it.advance()) != null; )
4792 action.accept(p.val);
4793 }
4794 }
4795 }
4796
4797 /**
4798 * A view of a ConcurrentHashMap as a {@link Set} of (key, value)
4799 * entries. This class cannot be directly instantiated. See
4800 * {@link #entrySet()}.
4801 */
4802 static final class EntrySetView<K,V> extends CollectionView<K,V,Map.Entry<K,V>>
4803 implements Set<Map.Entry<K,V>>, java.io.Serializable {
4804 private static final long serialVersionUID = 2249069246763182397L;
4805 EntrySetView(ConcurrentHashMap<K,V> map) { super(map); }
4806
4807 public boolean contains(Object o) {
4808 Object k, v, r; Map.Entry<?,?> e;
4809 return ((o instanceof Map.Entry) &&
4810 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
4811 (r = map.get(k)) != null &&
4812 (v = e.getValue()) != null &&
4813 (v == r || v.equals(r)));
4814 }
4815
4816 public boolean remove(Object o) {
4817 Object k, v; Map.Entry<?,?> e;
4818 return ((o instanceof Map.Entry) &&
4819 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
4820 (v = e.getValue()) != null &&
4821 map.remove(k, v));
4822 }
4823
4824 /**
4825 * @return an iterator over the entries of the backing map
4826 */
4827 public Iterator<Map.Entry<K,V>> iterator() {
4828 ConcurrentHashMap<K,V> m = map;
4829 Node<K,V>[] t;
4830 int f = (t = m.table) == null ? 0 : t.length;
4831 return new EntryIterator<K,V>(t, f, 0, f, m);
4832 }
4833
4834 public boolean add(Entry<K,V> e) {
4835 return map.putVal(e.getKey(), e.getValue(), false) == null;
4836 }
4837
4838 public boolean addAll(Collection<? extends Entry<K,V>> c) {
4839 boolean added = false;
4840 for (Entry<K,V> e : c) {
4841 if (add(e))
4842 added = true;
4843 }
4844 return added;
4845 }
4846
4847 public boolean removeIf(Predicate<? super Entry<K,V>> filter) {
4848 return map.removeEntryIf(filter);
4849 }
4850
4851 public final int hashCode() {
4852 int h = 0;
4853 Node<K,V>[] t;
4854 if ((t = map.table) != null) {
4855 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4856 for (Node<K,V> p; (p = it.advance()) != null; ) {
4857 h += p.hashCode();
4858 }
4859 }
4860 return h;
4861 }
4862
4863 public final boolean equals(Object o) {
4864 Set<?> c;
4865 return ((o instanceof Set) &&
4866 ((c = (Set<?>)o) == this ||
4867 (containsAll(c) && c.containsAll(this))));
4868 }
4869
4870 public Spliterator<Map.Entry<K,V>> spliterator() {
4871 Node<K,V>[] t;
4872 ConcurrentHashMap<K,V> m = map;
4873 long n = m.sumCount();
4874 int f = (t = m.table) == null ? 0 : t.length;
4875 return new EntrySpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n, m);
4876 }
4877
4878 public void forEach(Consumer<? super Map.Entry<K,V>> action) {
4879 if (action == null) throw new NullPointerException();
4880 Node<K,V>[] t;
4881 if ((t = map.table) != null) {
4882 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4883 for (Node<K,V> p; (p = it.advance()) != null; )
4884 action.accept(new MapEntry<K,V>(p.key, p.val, map));
4885 }
4886 }
4887
4888 }
4889
4890 // -------------------------------------------------------
4891
4892 /**
4893 * Base class for bulk tasks. Repeats some fields and code from
4894 * class Traverser, because we need to subclass CountedCompleter.
4895 */
4896 @SuppressWarnings("serial")
4897 abstract static class BulkTask<K,V,R> extends CountedCompleter<R> {
4898 Node<K,V>[] tab; // same as Traverser
4899 Node<K,V> next;
4900 TableStack<K,V> stack, spare;
4901 int index;
4902 int baseIndex;
4903 int baseLimit;
4904 final int baseSize;
4905 int batch; // split control
4906
4907 BulkTask(BulkTask<K,V,?> par, int b, int i, int f, Node<K,V>[] t) {
4908 super(par);
4909 this.batch = b;
4910 this.index = this.baseIndex = i;
4911 if ((this.tab = t) == null)
4912 this.baseSize = this.baseLimit = 0;
4913 else if (par == null)
4914 this.baseSize = this.baseLimit = t.length;
4915 else {
4916 this.baseLimit = f;
4917 this.baseSize = par.baseSize;
4918 }
4919 }
4920
4921 /**
4922 * Same as Traverser version.
4923 */
4924 final Node<K,V> advance() {
4925 Node<K,V> e;
4926 if ((e = next) != null)
4927 e = e.next;
4928 for (;;) {
4929 Node<K,V>[] t; int i, n;
4930 if (e != null)
4931 return next = e;
4932 if (baseIndex >= baseLimit || (t = tab) == null ||
4933 (n = t.length) <= (i = index) || i < 0)
4934 return next = null;
4935 if ((e = tabAt(t, i)) != null && e.hash < 0) {
4936 if (e instanceof ForwardingNode) {
4937 tab = ((ForwardingNode<K,V>)e).nextTable;
4938 e = null;
4939 pushState(t, i, n);
4940 continue;
4941 }
4942 else if (e instanceof TreeBin)
4943 e = ((TreeBin<K,V>)e).first;
4944 else
4945 e = null;
4946 }
4947 if (stack != null)
4948 recoverState(n);
4949 else if ((index = i + baseSize) >= n)
4950 index = ++baseIndex;
4951 }
4952 }
4953
4954 private void pushState(Node<K,V>[] t, int i, int n) {
4955 TableStack<K,V> s = spare;
4956 if (s != null)
4957 spare = s.next;
4958 else
4959 s = new TableStack<K,V>();
4960 s.tab = t;
4961 s.length = n;
4962 s.index = i;
4963 s.next = stack;
4964 stack = s;
4965 }
4966
4967 private void recoverState(int n) {
4968 TableStack<K,V> s; int len;
4969 while ((s = stack) != null && (index += (len = s.length)) >= n) {
4970 n = len;
4971 index = s.index;
4972 tab = s.tab;
4973 s.tab = null;
4974 TableStack<K,V> next = s.next;
4975 s.next = spare; // save for reuse
4976 stack = next;
4977 spare = s;
4978 }
4979 if (s == null && (index += baseSize) >= n)
4980 index = ++baseIndex;
4981 }
4982 }
4983
4984 /*
4985 * Task classes. Coded in a regular but ugly format/style to
4986 * simplify checks that each variant differs in the right way from
4987 * others. The null screenings exist because compilers cannot tell
4988 * that we've already null-checked task arguments, so we force
4989 * simplest hoisted bypass to help avoid convoluted traps.
4990 */
4991 @SuppressWarnings("serial")
4992 static final class ForEachKeyTask<K,V>
4993 extends BulkTask<K,V,Void> {
4994 final Consumer<? super K> action;
4995 ForEachKeyTask
4996 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4997 Consumer<? super K> action) {
4998 super(p, b, i, f, t);
4999 this.action = action;
5000 }
5001 public final void compute() {
5002 final Consumer<? super K> action;
5003 if ((action = this.action) != null) {
5004 for (int i = baseIndex, f, h; batch > 0 &&
5005 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5006 addToPendingCount(1);
5007 new ForEachKeyTask<K,V>
5008 (this, batch >>>= 1, baseLimit = h, f, tab,
5009 action).fork();
5010 }
5011 for (Node<K,V> p; (p = advance()) != null;)
5012 action.accept(p.key);
5013 propagateCompletion();
5014 }
5015 }
5016 }
5017
5018 @SuppressWarnings("serial")
5019 static final class ForEachValueTask<K,V>
5020 extends BulkTask<K,V,Void> {
5021 final Consumer<? super V> action;
5022 ForEachValueTask
5023 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5024 Consumer<? super V> action) {
5025 super(p, b, i, f, t);
5026 this.action = action;
5027 }
5028 public final void compute() {
5029 final Consumer<? super V> action;
5030 if ((action = this.action) != null) {
5031 for (int i = baseIndex, f, h; batch > 0 &&
5032 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5033 addToPendingCount(1);
5034 new ForEachValueTask<K,V>
5035 (this, batch >>>= 1, baseLimit = h, f, tab,
5036 action).fork();
5037 }
5038 for (Node<K,V> p; (p = advance()) != null;)
5039 action.accept(p.val);
5040 propagateCompletion();
5041 }
5042 }
5043 }
5044
5045 @SuppressWarnings("serial")
5046 static final class ForEachEntryTask<K,V>
5047 extends BulkTask<K,V,Void> {
5048 final Consumer<? super Entry<K,V>> action;
5049 ForEachEntryTask
5050 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5051 Consumer<? super Entry<K,V>> action) {
5052 super(p, b, i, f, t);
5053 this.action = action;
5054 }
5055 public final void compute() {
5056 final Consumer<? super Entry<K,V>> action;
5057 if ((action = this.action) != null) {
5058 for (int i = baseIndex, f, h; batch > 0 &&
5059 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5060 addToPendingCount(1);
5061 new ForEachEntryTask<K,V>
5062 (this, batch >>>= 1, baseLimit = h, f, tab,
5063 action).fork();
5064 }
5065 for (Node<K,V> p; (p = advance()) != null; )
5066 action.accept(p);
5067 propagateCompletion();
5068 }
5069 }
5070 }
5071
5072 @SuppressWarnings("serial")
5073 static final class ForEachMappingTask<K,V>
5074 extends BulkTask<K,V,Void> {
5075 final BiConsumer<? super K, ? super V> action;
5076 ForEachMappingTask
5077 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5078 BiConsumer<? super K,? super V> action) {
5079 super(p, b, i, f, t);
5080 this.action = action;
5081 }
5082 public final void compute() {
5083 final BiConsumer<? super K, ? super V> action;
5084 if ((action = this.action) != null) {
5085 for (int i = baseIndex, f, h; batch > 0 &&
5086 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5087 addToPendingCount(1);
5088 new ForEachMappingTask<K,V>
5089 (this, batch >>>= 1, baseLimit = h, f, tab,
5090 action).fork();
5091 }
5092 for (Node<K,V> p; (p = advance()) != null; )
5093 action.accept(p.key, p.val);
5094 propagateCompletion();
5095 }
5096 }
5097 }
5098
5099 @SuppressWarnings("serial")
5100 static final class ForEachTransformedKeyTask<K,V,U>
5101 extends BulkTask<K,V,Void> {
5102 final Function<? super K, ? extends U> transformer;
5103 final Consumer<? super U> action;
5104 ForEachTransformedKeyTask
5105 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5106 Function<? super K, ? extends U> transformer, Consumer<? super U> action) {
5107 super(p, b, i, f, t);
5108 this.transformer = transformer; this.action = action;
5109 }
5110 public final void compute() {
5111 final Function<? super K, ? extends U> transformer;
5112 final Consumer<? super U> action;
5113 if ((transformer = this.transformer) != null &&
5114 (action = this.action) != null) {
5115 for (int i = baseIndex, f, h; batch > 0 &&
5116 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5117 addToPendingCount(1);
5118 new ForEachTransformedKeyTask<K,V,U>
5119 (this, batch >>>= 1, baseLimit = h, f, tab,
5120 transformer, action).fork();
5121 }
5122 for (Node<K,V> p; (p = advance()) != null; ) {
5123 U u;
5124 if ((u = transformer.apply(p.key)) != null)
5125 action.accept(u);
5126 }
5127 propagateCompletion();
5128 }
5129 }
5130 }
5131
5132 @SuppressWarnings("serial")
5133 static final class ForEachTransformedValueTask<K,V,U>
5134 extends BulkTask<K,V,Void> {
5135 final Function<? super V, ? extends U> transformer;
5136 final Consumer<? super U> action;
5137 ForEachTransformedValueTask
5138 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5139 Function<? super V, ? extends U> transformer, Consumer<? super U> action) {
5140 super(p, b, i, f, t);
5141 this.transformer = transformer; this.action = action;
5142 }
5143 public final void compute() {
5144 final Function<? super V, ? extends U> transformer;
5145 final Consumer<? super U> action;
5146 if ((transformer = this.transformer) != null &&
5147 (action = this.action) != null) {
5148 for (int i = baseIndex, f, h; batch > 0 &&
5149 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5150 addToPendingCount(1);
5151 new ForEachTransformedValueTask<K,V,U>
5152 (this, batch >>>= 1, baseLimit = h, f, tab,
5153 transformer, action).fork();
5154 }
5155 for (Node<K,V> p; (p = advance()) != null; ) {
5156 U u;
5157 if ((u = transformer.apply(p.val)) != null)
5158 action.accept(u);
5159 }
5160 propagateCompletion();
5161 }
5162 }
5163 }
5164
5165 @SuppressWarnings("serial")
5166 static final class ForEachTransformedEntryTask<K,V,U>
5167 extends BulkTask<K,V,Void> {
5168 final Function<Map.Entry<K,V>, ? extends U> transformer;
5169 final Consumer<? super U> action;
5170 ForEachTransformedEntryTask
5171 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5172 Function<Map.Entry<K,V>, ? extends U> transformer, Consumer<? super U> action) {
5173 super(p, b, i, f, t);
5174 this.transformer = transformer; this.action = action;
5175 }
5176 public final void compute() {
5177 final Function<Map.Entry<K,V>, ? extends U> transformer;
5178 final Consumer<? super U> action;
5179 if ((transformer = this.transformer) != null &&
5180 (action = this.action) != null) {
5181 for (int i = baseIndex, f, h; batch > 0 &&
5182 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5183 addToPendingCount(1);
5184 new ForEachTransformedEntryTask<K,V,U>
5185 (this, batch >>>= 1, baseLimit = h, f, tab,
5186 transformer, action).fork();
5187 }
5188 for (Node<K,V> p; (p = advance()) != null; ) {
5189 U u;
5190 if ((u = transformer.apply(p)) != null)
5191 action.accept(u);
5192 }
5193 propagateCompletion();
5194 }
5195 }
5196 }
5197
5198 @SuppressWarnings("serial")
5199 static final class ForEachTransformedMappingTask<K,V,U>
5200 extends BulkTask<K,V,Void> {
5201 final BiFunction<? super K, ? super V, ? extends U> transformer;
5202 final Consumer<? super U> action;
5203 ForEachTransformedMappingTask
5204 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5205 BiFunction<? super K, ? super V, ? extends U> transformer,
5206 Consumer<? super U> action) {
5207 super(p, b, i, f, t);
5208 this.transformer = transformer; this.action = action;
5209 }
5210 public final void compute() {
5211 final BiFunction<? super K, ? super V, ? extends U> transformer;
5212 final Consumer<? super U> action;
5213 if ((transformer = this.transformer) != null &&
5214 (action = this.action) != null) {
5215 for (int i = baseIndex, f, h; batch > 0 &&
5216 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5217 addToPendingCount(1);
5218 new ForEachTransformedMappingTask<K,V,U>
5219 (this, batch >>>= 1, baseLimit = h, f, tab,
5220 transformer, action).fork();
5221 }
5222 for (Node<K,V> p; (p = advance()) != null; ) {
5223 U u;
5224 if ((u = transformer.apply(p.key, p.val)) != null)
5225 action.accept(u);
5226 }
5227 propagateCompletion();
5228 }
5229 }
5230 }
5231
5232 @SuppressWarnings("serial")
5233 static final class SearchKeysTask<K,V,U>
5234 extends BulkTask<K,V,U> {
5235 final Function<? super K, ? extends U> searchFunction;
5236 final AtomicReference<U> result;
5237 SearchKeysTask
5238 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5239 Function<? super K, ? extends U> searchFunction,
5240 AtomicReference<U> result) {
5241 super(p, b, i, f, t);
5242 this.searchFunction = searchFunction; this.result = result;
5243 }
5244 public final U getRawResult() { return result.get(); }
5245 public final void compute() {
5246 final Function<? super K, ? extends U> searchFunction;
5247 final AtomicReference<U> result;
5248 if ((searchFunction = this.searchFunction) != null &&
5249 (result = this.result) != null) {
5250 for (int i = baseIndex, f, h; batch > 0 &&
5251 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5252 if (result.get() != null)
5253 return;
5254 addToPendingCount(1);
5255 new SearchKeysTask<K,V,U>
5256 (this, batch >>>= 1, baseLimit = h, f, tab,
5257 searchFunction, result).fork();
5258 }
5259 while (result.get() == null) {
5260 U u;
5261 Node<K,V> p;
5262 if ((p = advance()) == null) {
5263 propagateCompletion();
5264 break;
5265 }
5266 if ((u = searchFunction.apply(p.key)) != null) {
5267 if (result.compareAndSet(null, u))
5268 quietlyCompleteRoot();
5269 break;
5270 }
5271 }
5272 }
5273 }
5274 }
5275
5276 @SuppressWarnings("serial")
5277 static final class SearchValuesTask<K,V,U>
5278 extends BulkTask<K,V,U> {
5279 final Function<? super V, ? extends U> searchFunction;
5280 final AtomicReference<U> result;
5281 SearchValuesTask
5282 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5283 Function<? super V, ? extends U> searchFunction,
5284 AtomicReference<U> result) {
5285 super(p, b, i, f, t);
5286 this.searchFunction = searchFunction; this.result = result;
5287 }
5288 public final U getRawResult() { return result.get(); }
5289 public final void compute() {
5290 final Function<? super V, ? extends U> searchFunction;
5291 final AtomicReference<U> result;
5292 if ((searchFunction = this.searchFunction) != null &&
5293 (result = this.result) != null) {
5294 for (int i = baseIndex, f, h; batch > 0 &&
5295 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5296 if (result.get() != null)
5297 return;
5298 addToPendingCount(1);
5299 new SearchValuesTask<K,V,U>
5300 (this, batch >>>= 1, baseLimit = h, f, tab,
5301 searchFunction, result).fork();
5302 }
5303 while (result.get() == null) {
5304 U u;
5305 Node<K,V> p;
5306 if ((p = advance()) == null) {
5307 propagateCompletion();
5308 break;
5309 }
5310 if ((u = searchFunction.apply(p.val)) != null) {
5311 if (result.compareAndSet(null, u))
5312 quietlyCompleteRoot();
5313 break;
5314 }
5315 }
5316 }
5317 }
5318 }
5319
5320 @SuppressWarnings("serial")
5321 static final class SearchEntriesTask<K,V,U>
5322 extends BulkTask<K,V,U> {
5323 final Function<Entry<K,V>, ? extends U> searchFunction;
5324 final AtomicReference<U> result;
5325 SearchEntriesTask
5326 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5327 Function<Entry<K,V>, ? extends U> searchFunction,
5328 AtomicReference<U> result) {
5329 super(p, b, i, f, t);
5330 this.searchFunction = searchFunction; this.result = result;
5331 }
5332 public final U getRawResult() { return result.get(); }
5333 public final void compute() {
5334 final Function<Entry<K,V>, ? extends U> searchFunction;
5335 final AtomicReference<U> result;
5336 if ((searchFunction = this.searchFunction) != null &&
5337 (result = this.result) != null) {
5338 for (int i = baseIndex, f, h; batch > 0 &&
5339 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5340 if (result.get() != null)
5341 return;
5342 addToPendingCount(1);
5343 new SearchEntriesTask<K,V,U>
5344 (this, batch >>>= 1, baseLimit = h, f, tab,
5345 searchFunction, result).fork();
5346 }
5347 while (result.get() == null) {
5348 U u;
5349 Node<K,V> p;
5350 if ((p = advance()) == null) {
5351 propagateCompletion();
5352 break;
5353 }
5354 if ((u = searchFunction.apply(p)) != null) {
5355 if (result.compareAndSet(null, u))
5356 quietlyCompleteRoot();
5357 return;
5358 }
5359 }
5360 }
5361 }
5362 }
5363
5364 @SuppressWarnings("serial")
5365 static final class SearchMappingsTask<K,V,U>
5366 extends BulkTask<K,V,U> {
5367 final BiFunction<? super K, ? super V, ? extends U> searchFunction;
5368 final AtomicReference<U> result;
5369 SearchMappingsTask
5370 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5371 BiFunction<? super K, ? super V, ? extends U> searchFunction,
5372 AtomicReference<U> result) {
5373 super(p, b, i, f, t);
5374 this.searchFunction = searchFunction; this.result = result;
5375 }
5376 public final U getRawResult() { return result.get(); }
5377 public final void compute() {
5378 final BiFunction<? super K, ? super V, ? extends U> searchFunction;
5379 final AtomicReference<U> result;
5380 if ((searchFunction = this.searchFunction) != null &&
5381 (result = this.result) != null) {
5382 for (int i = baseIndex, f, h; batch > 0 &&
5383 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5384 if (result.get() != null)
5385 return;
5386 addToPendingCount(1);
5387 new SearchMappingsTask<K,V,U>
5388 (this, batch >>>= 1, baseLimit = h, f, tab,
5389 searchFunction, result).fork();
5390 }
5391 while (result.get() == null) {
5392 U u;
5393 Node<K,V> p;
5394 if ((p = advance()) == null) {
5395 propagateCompletion();
5396 break;
5397 }
5398 if ((u = searchFunction.apply(p.key, p.val)) != null) {
5399 if (result.compareAndSet(null, u))
5400 quietlyCompleteRoot();
5401 break;
5402 }
5403 }
5404 }
5405 }
5406 }
5407
5408 @SuppressWarnings("serial")
5409 static final class ReduceKeysTask<K,V>
5410 extends BulkTask<K,V,K> {
5411 final BiFunction<? super K, ? super K, ? extends K> reducer;
5412 K result;
5413 ReduceKeysTask<K,V> rights, nextRight;
5414 ReduceKeysTask
5415 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5416 ReduceKeysTask<K,V> nextRight,
5417 BiFunction<? super K, ? super K, ? extends K> reducer) {
5418 super(p, b, i, f, t); this.nextRight = nextRight;
5419 this.reducer = reducer;
5420 }
5421 public final K getRawResult() { return result; }
5422 public final void compute() {
5423 final BiFunction<? super K, ? super K, ? extends K> reducer;
5424 if ((reducer = this.reducer) != null) {
5425 for (int i = baseIndex, f, h; batch > 0 &&
5426 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5427 addToPendingCount(1);
5428 (rights = new ReduceKeysTask<K,V>
5429 (this, batch >>>= 1, baseLimit = h, f, tab,
5430 rights, reducer)).fork();
5431 }
5432 K r = null;
5433 for (Node<K,V> p; (p = advance()) != null; ) {
5434 K u = p.key;
5435 r = (r == null) ? u : u == null ? r : reducer.apply(r, u);
5436 }
5437 result = r;
5438 CountedCompleter<?> c;
5439 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5440 @SuppressWarnings("unchecked")
5441 ReduceKeysTask<K,V>
5442 t = (ReduceKeysTask<K,V>)c,
5443 s = t.rights;
5444 while (s != null) {
5445 K tr, sr;
5446 if ((sr = s.result) != null)
5447 t.result = (((tr = t.result) == null) ? sr :
5448 reducer.apply(tr, sr));
5449 s = t.rights = s.nextRight;
5450 }
5451 }
5452 }
5453 }
5454 }
5455
5456 @SuppressWarnings("serial")
5457 static final class ReduceValuesTask<K,V>
5458 extends BulkTask<K,V,V> {
5459 final BiFunction<? super V, ? super V, ? extends V> reducer;
5460 V result;
5461 ReduceValuesTask<K,V> rights, nextRight;
5462 ReduceValuesTask
5463 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5464 ReduceValuesTask<K,V> nextRight,
5465 BiFunction<? super V, ? super V, ? extends V> reducer) {
5466 super(p, b, i, f, t); this.nextRight = nextRight;
5467 this.reducer = reducer;
5468 }
5469 public final V getRawResult() { return result; }
5470 public final void compute() {
5471 final BiFunction<? super V, ? super V, ? extends V> reducer;
5472 if ((reducer = this.reducer) != null) {
5473 for (int i = baseIndex, f, h; batch > 0 &&
5474 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5475 addToPendingCount(1);
5476 (rights = new ReduceValuesTask<K,V>
5477 (this, batch >>>= 1, baseLimit = h, f, tab,
5478 rights, reducer)).fork();
5479 }
5480 V r = null;
5481 for (Node<K,V> p; (p = advance()) != null; ) {
5482 V v = p.val;
5483 r = (r == null) ? v : reducer.apply(r, v);
5484 }
5485 result = r;
5486 CountedCompleter<?> c;
5487 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5488 @SuppressWarnings("unchecked")
5489 ReduceValuesTask<K,V>
5490 t = (ReduceValuesTask<K,V>)c,
5491 s = t.rights;
5492 while (s != null) {
5493 V tr, sr;
5494 if ((sr = s.result) != null)
5495 t.result = (((tr = t.result) == null) ? sr :
5496 reducer.apply(tr, sr));
5497 s = t.rights = s.nextRight;
5498 }
5499 }
5500 }
5501 }
5502 }
5503
5504 @SuppressWarnings("serial")
5505 static final class ReduceEntriesTask<K,V>
5506 extends BulkTask<K,V,Map.Entry<K,V>> {
5507 final BiFunction<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer;
5508 Map.Entry<K,V> result;
5509 ReduceEntriesTask<K,V> rights, nextRight;
5510 ReduceEntriesTask
5511 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5512 ReduceEntriesTask<K,V> nextRight,
5513 BiFunction<Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer) {
5514 super(p, b, i, f, t); this.nextRight = nextRight;
5515 this.reducer = reducer;
5516 }
5517 public final Map.Entry<K,V> getRawResult() { return result; }
5518 public final void compute() {
5519 final BiFunction<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer;
5520 if ((reducer = this.reducer) != null) {
5521 for (int i = baseIndex, f, h; batch > 0 &&
5522 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5523 addToPendingCount(1);
5524 (rights = new ReduceEntriesTask<K,V>
5525 (this, batch >>>= 1, baseLimit = h, f, tab,
5526 rights, reducer)).fork();
5527 }
5528 Map.Entry<K,V> r = null;
5529 for (Node<K,V> p; (p = advance()) != null; )
5530 r = (r == null) ? p : reducer.apply(r, p);
5531 result = r;
5532 CountedCompleter<?> c;
5533 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5534 @SuppressWarnings("unchecked")
5535 ReduceEntriesTask<K,V>
5536 t = (ReduceEntriesTask<K,V>)c,
5537 s = t.rights;
5538 while (s != null) {
5539 Map.Entry<K,V> tr, sr;
5540 if ((sr = s.result) != null)
5541 t.result = (((tr = t.result) == null) ? sr :
5542 reducer.apply(tr, sr));
5543 s = t.rights = s.nextRight;
5544 }
5545 }
5546 }
5547 }
5548 }
5549
5550 @SuppressWarnings("serial")
5551 static final class MapReduceKeysTask<K,V,U>
5552 extends BulkTask<K,V,U> {
5553 final Function<? super K, ? extends U> transformer;
5554 final BiFunction<? super U, ? super U, ? extends U> reducer;
5555 U result;
5556 MapReduceKeysTask<K,V,U> rights, nextRight;
5557 MapReduceKeysTask
5558 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5559 MapReduceKeysTask<K,V,U> nextRight,
5560 Function<? super K, ? extends U> transformer,
5561 BiFunction<? super U, ? super U, ? extends U> reducer) {
5562 super(p, b, i, f, t); this.nextRight = nextRight;
5563 this.transformer = transformer;
5564 this.reducer = reducer;
5565 }
5566 public final U getRawResult() { return result; }
5567 public final void compute() {
5568 final Function<? super K, ? extends U> transformer;
5569 final BiFunction<? super U, ? super U, ? extends U> reducer;
5570 if ((transformer = this.transformer) != null &&
5571 (reducer = this.reducer) != null) {
5572 for (int i = baseIndex, f, h; batch > 0 &&
5573 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5574 addToPendingCount(1);
5575 (rights = new MapReduceKeysTask<K,V,U>
5576 (this, batch >>>= 1, baseLimit = h, f, tab,
5577 rights, transformer, reducer)).fork();
5578 }
5579 U r = null;
5580 for (Node<K,V> p; (p = advance()) != null; ) {
5581 U u;
5582 if ((u = transformer.apply(p.key)) != null)
5583 r = (r == null) ? u : reducer.apply(r, u);
5584 }
5585 result = r;
5586 CountedCompleter<?> c;
5587 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5588 @SuppressWarnings("unchecked")
5589 MapReduceKeysTask<K,V,U>
5590 t = (MapReduceKeysTask<K,V,U>)c,
5591 s = t.rights;
5592 while (s != null) {
5593 U tr, sr;
5594 if ((sr = s.result) != null)
5595 t.result = (((tr = t.result) == null) ? sr :
5596 reducer.apply(tr, sr));
5597 s = t.rights = s.nextRight;
5598 }
5599 }
5600 }
5601 }
5602 }
5603
5604 @SuppressWarnings("serial")
5605 static final class MapReduceValuesTask<K,V,U>
5606 extends BulkTask<K,V,U> {
5607 final Function<? super V, ? extends U> transformer;
5608 final BiFunction<? super U, ? super U, ? extends U> reducer;
5609 U result;
5610 MapReduceValuesTask<K,V,U> rights, nextRight;
5611 MapReduceValuesTask
5612 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5613 MapReduceValuesTask<K,V,U> nextRight,
5614 Function<? super V, ? extends U> transformer,
5615 BiFunction<? super U, ? super U, ? extends U> reducer) {
5616 super(p, b, i, f, t); this.nextRight = nextRight;
5617 this.transformer = transformer;
5618 this.reducer = reducer;
5619 }
5620 public final U getRawResult() { return result; }
5621 public final void compute() {
5622 final Function<? super V, ? extends U> transformer;
5623 final BiFunction<? super U, ? super U, ? extends U> reducer;
5624 if ((transformer = this.transformer) != null &&
5625 (reducer = this.reducer) != null) {
5626 for (int i = baseIndex, f, h; batch > 0 &&
5627 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5628 addToPendingCount(1);
5629 (rights = new MapReduceValuesTask<K,V,U>
5630 (this, batch >>>= 1, baseLimit = h, f, tab,
5631 rights, transformer, reducer)).fork();
5632 }
5633 U r = null;
5634 for (Node<K,V> p; (p = advance()) != null; ) {
5635 U u;
5636 if ((u = transformer.apply(p.val)) != null)
5637 r = (r == null) ? u : reducer.apply(r, u);
5638 }
5639 result = r;
5640 CountedCompleter<?> c;
5641 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5642 @SuppressWarnings("unchecked")
5643 MapReduceValuesTask<K,V,U>
5644 t = (MapReduceValuesTask<K,V,U>)c,
5645 s = t.rights;
5646 while (s != null) {
5647 U tr, sr;
5648 if ((sr = s.result) != null)
5649 t.result = (((tr = t.result) == null) ? sr :
5650 reducer.apply(tr, sr));
5651 s = t.rights = s.nextRight;
5652 }
5653 }
5654 }
5655 }
5656 }
5657
5658 @SuppressWarnings("serial")
5659 static final class MapReduceEntriesTask<K,V,U>
5660 extends BulkTask<K,V,U> {
5661 final Function<Map.Entry<K,V>, ? extends U> transformer;
5662 final BiFunction<? super U, ? super U, ? extends U> reducer;
5663 U result;
5664 MapReduceEntriesTask<K,V,U> rights, nextRight;
5665 MapReduceEntriesTask
5666 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5667 MapReduceEntriesTask<K,V,U> nextRight,
5668 Function<Map.Entry<K,V>, ? extends U> transformer,
5669 BiFunction<? super U, ? super U, ? extends U> reducer) {
5670 super(p, b, i, f, t); this.nextRight = nextRight;
5671 this.transformer = transformer;
5672 this.reducer = reducer;
5673 }
5674 public final U getRawResult() { return result; }
5675 public final void compute() {
5676 final Function<Map.Entry<K,V>, ? extends U> transformer;
5677 final BiFunction<? super U, ? super U, ? extends U> reducer;
5678 if ((transformer = this.transformer) != null &&
5679 (reducer = this.reducer) != null) {
5680 for (int i = baseIndex, f, h; batch > 0 &&
5681 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5682 addToPendingCount(1);
5683 (rights = new MapReduceEntriesTask<K,V,U>
5684 (this, batch >>>= 1, baseLimit = h, f, tab,
5685 rights, transformer, reducer)).fork();
5686 }
5687 U r = null;
5688 for (Node<K,V> p; (p = advance()) != null; ) {
5689 U u;
5690 if ((u = transformer.apply(p)) != null)
5691 r = (r == null) ? u : reducer.apply(r, u);
5692 }
5693 result = r;
5694 CountedCompleter<?> c;
5695 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5696 @SuppressWarnings("unchecked")
5697 MapReduceEntriesTask<K,V,U>
5698 t = (MapReduceEntriesTask<K,V,U>)c,
5699 s = t.rights;
5700 while (s != null) {
5701 U tr, sr;
5702 if ((sr = s.result) != null)
5703 t.result = (((tr = t.result) == null) ? sr :
5704 reducer.apply(tr, sr));
5705 s = t.rights = s.nextRight;
5706 }
5707 }
5708 }
5709 }
5710 }
5711
5712 @SuppressWarnings("serial")
5713 static final class MapReduceMappingsTask<K,V,U>
5714 extends BulkTask<K,V,U> {
5715 final BiFunction<? super K, ? super V, ? extends U> transformer;
5716 final BiFunction<? super U, ? super U, ? extends U> reducer;
5717 U result;
5718 MapReduceMappingsTask<K,V,U> rights, nextRight;
5719 MapReduceMappingsTask
5720 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5721 MapReduceMappingsTask<K,V,U> nextRight,
5722 BiFunction<? super K, ? super V, ? extends U> transformer,
5723 BiFunction<? super U, ? super U, ? extends U> reducer) {
5724 super(p, b, i, f, t); this.nextRight = nextRight;
5725 this.transformer = transformer;
5726 this.reducer = reducer;
5727 }
5728 public final U getRawResult() { return result; }
5729 public final void compute() {
5730 final BiFunction<? super K, ? super V, ? extends U> transformer;
5731 final BiFunction<? super U, ? super U, ? extends U> reducer;
5732 if ((transformer = this.transformer) != null &&
5733 (reducer = this.reducer) != null) {
5734 for (int i = baseIndex, f, h; batch > 0 &&
5735 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5736 addToPendingCount(1);
5737 (rights = new MapReduceMappingsTask<K,V,U>
5738 (this, batch >>>= 1, baseLimit = h, f, tab,
5739 rights, transformer, reducer)).fork();
5740 }
5741 U r = null;
5742 for (Node<K,V> p; (p = advance()) != null; ) {
5743 U u;
5744 if ((u = transformer.apply(p.key, p.val)) != null)
5745 r = (r == null) ? u : reducer.apply(r, u);
5746 }
5747 result = r;
5748 CountedCompleter<?> c;
5749 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5750 @SuppressWarnings("unchecked")
5751 MapReduceMappingsTask<K,V,U>
5752 t = (MapReduceMappingsTask<K,V,U>)c,
5753 s = t.rights;
5754 while (s != null) {
5755 U tr, sr;
5756 if ((sr = s.result) != null)
5757 t.result = (((tr = t.result) == null) ? sr :
5758 reducer.apply(tr, sr));
5759 s = t.rights = s.nextRight;
5760 }
5761 }
5762 }
5763 }
5764 }
5765
5766 @SuppressWarnings("serial")
5767 static final class MapReduceKeysToDoubleTask<K,V>
5768 extends BulkTask<K,V,Double> {
5769 final ToDoubleFunction<? super K> transformer;
5770 final DoubleBinaryOperator reducer;
5771 final double basis;
5772 double result;
5773 MapReduceKeysToDoubleTask<K,V> rights, nextRight;
5774 MapReduceKeysToDoubleTask
5775 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5776 MapReduceKeysToDoubleTask<K,V> nextRight,
5777 ToDoubleFunction<? super K> transformer,
5778 double basis,
5779 DoubleBinaryOperator reducer) {
5780 super(p, b, i, f, t); this.nextRight = nextRight;
5781 this.transformer = transformer;
5782 this.basis = basis; this.reducer = reducer;
5783 }
5784 public final Double getRawResult() { return result; }
5785 public final void compute() {
5786 final ToDoubleFunction<? super K> transformer;
5787 final DoubleBinaryOperator reducer;
5788 if ((transformer = this.transformer) != null &&
5789 (reducer = this.reducer) != null) {
5790 double r = this.basis;
5791 for (int i = baseIndex, f, h; batch > 0 &&
5792 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5793 addToPendingCount(1);
5794 (rights = new MapReduceKeysToDoubleTask<K,V>
5795 (this, batch >>>= 1, baseLimit = h, f, tab,
5796 rights, transformer, r, reducer)).fork();
5797 }
5798 for (Node<K,V> p; (p = advance()) != null; )
5799 r = reducer.applyAsDouble(r, transformer.applyAsDouble(p.key));
5800 result = r;
5801 CountedCompleter<?> c;
5802 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5803 @SuppressWarnings("unchecked")
5804 MapReduceKeysToDoubleTask<K,V>
5805 t = (MapReduceKeysToDoubleTask<K,V>)c,
5806 s = t.rights;
5807 while (s != null) {
5808 t.result = reducer.applyAsDouble(t.result, s.result);
5809 s = t.rights = s.nextRight;
5810 }
5811 }
5812 }
5813 }
5814 }
5815
5816 @SuppressWarnings("serial")
5817 static final class MapReduceValuesToDoubleTask<K,V>
5818 extends BulkTask<K,V,Double> {
5819 final ToDoubleFunction<? super V> transformer;
5820 final DoubleBinaryOperator reducer;
5821 final double basis;
5822 double result;
5823 MapReduceValuesToDoubleTask<K,V> rights, nextRight;
5824 MapReduceValuesToDoubleTask
5825 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5826 MapReduceValuesToDoubleTask<K,V> nextRight,
5827 ToDoubleFunction<? super V> transformer,
5828 double basis,
5829 DoubleBinaryOperator reducer) {
5830 super(p, b, i, f, t); this.nextRight = nextRight;
5831 this.transformer = transformer;
5832 this.basis = basis; this.reducer = reducer;
5833 }
5834 public final Double getRawResult() { return result; }
5835 public final void compute() {
5836 final ToDoubleFunction<? super V> transformer;
5837 final DoubleBinaryOperator reducer;
5838 if ((transformer = this.transformer) != null &&
5839 (reducer = this.reducer) != null) {
5840 double r = this.basis;
5841 for (int i = baseIndex, f, h; batch > 0 &&
5842 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5843 addToPendingCount(1);
5844 (rights = new MapReduceValuesToDoubleTask<K,V>
5845 (this, batch >>>= 1, baseLimit = h, f, tab,
5846 rights, transformer, r, reducer)).fork();
5847 }
5848 for (Node<K,V> p; (p = advance()) != null; )
5849 r = reducer.applyAsDouble(r, transformer.applyAsDouble(p.val));
5850 result = r;
5851 CountedCompleter<?> c;
5852 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5853 @SuppressWarnings("unchecked")
5854 MapReduceValuesToDoubleTask<K,V>
5855 t = (MapReduceValuesToDoubleTask<K,V>)c,
5856 s = t.rights;
5857 while (s != null) {
5858 t.result = reducer.applyAsDouble(t.result, s.result);
5859 s = t.rights = s.nextRight;
5860 }
5861 }
5862 }
5863 }
5864 }
5865
5866 @SuppressWarnings("serial")
5867 static final class MapReduceEntriesToDoubleTask<K,V>
5868 extends BulkTask<K,V,Double> {
5869 final ToDoubleFunction<Map.Entry<K,V>> transformer;
5870 final DoubleBinaryOperator reducer;
5871 final double basis;
5872 double result;
5873 MapReduceEntriesToDoubleTask<K,V> rights, nextRight;
5874 MapReduceEntriesToDoubleTask
5875 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5876 MapReduceEntriesToDoubleTask<K,V> nextRight,
5877 ToDoubleFunction<Map.Entry<K,V>> transformer,
5878 double basis,
5879 DoubleBinaryOperator reducer) {
5880 super(p, b, i, f, t); this.nextRight = nextRight;
5881 this.transformer = transformer;
5882 this.basis = basis; this.reducer = reducer;
5883 }
5884 public final Double getRawResult() { return result; }
5885 public final void compute() {
5886 final ToDoubleFunction<Map.Entry<K,V>> transformer;
5887 final DoubleBinaryOperator reducer;
5888 if ((transformer = this.transformer) != null &&
5889 (reducer = this.reducer) != null) {
5890 double r = this.basis;
5891 for (int i = baseIndex, f, h; batch > 0 &&
5892 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5893 addToPendingCount(1);
5894 (rights = new MapReduceEntriesToDoubleTask<K,V>
5895 (this, batch >>>= 1, baseLimit = h, f, tab,
5896 rights, transformer, r, reducer)).fork();
5897 }
5898 for (Node<K,V> p; (p = advance()) != null; )
5899 r = reducer.applyAsDouble(r, transformer.applyAsDouble(p));
5900 result = r;
5901 CountedCompleter<?> c;
5902 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5903 @SuppressWarnings("unchecked")
5904 MapReduceEntriesToDoubleTask<K,V>
5905 t = (MapReduceEntriesToDoubleTask<K,V>)c,
5906 s = t.rights;
5907 while (s != null) {
5908 t.result = reducer.applyAsDouble(t.result, s.result);
5909 s = t.rights = s.nextRight;
5910 }
5911 }
5912 }
5913 }
5914 }
5915
5916 @SuppressWarnings("serial")
5917 static final class MapReduceMappingsToDoubleTask<K,V>
5918 extends BulkTask<K,V,Double> {
5919 final ToDoubleBiFunction<? super K, ? super V> transformer;
5920 final DoubleBinaryOperator reducer;
5921 final double basis;
5922 double result;
5923 MapReduceMappingsToDoubleTask<K,V> rights, nextRight;
5924 MapReduceMappingsToDoubleTask
5925 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5926 MapReduceMappingsToDoubleTask<K,V> nextRight,
5927 ToDoubleBiFunction<? super K, ? super V> transformer,
5928 double basis,
5929 DoubleBinaryOperator reducer) {
5930 super(p, b, i, f, t); this.nextRight = nextRight;
5931 this.transformer = transformer;
5932 this.basis = basis; this.reducer = reducer;
5933 }
5934 public final Double getRawResult() { return result; }
5935 public final void compute() {
5936 final ToDoubleBiFunction<? super K, ? super V> transformer;
5937 final DoubleBinaryOperator reducer;
5938 if ((transformer = this.transformer) != null &&
5939 (reducer = this.reducer) != null) {
5940 double r = this.basis;
5941 for (int i = baseIndex, f, h; batch > 0 &&
5942 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5943 addToPendingCount(1);
5944 (rights = new MapReduceMappingsToDoubleTask<K,V>
5945 (this, batch >>>= 1, baseLimit = h, f, tab,
5946 rights, transformer, r, reducer)).fork();
5947 }
5948 for (Node<K,V> p; (p = advance()) != null; )
5949 r = reducer.applyAsDouble(r, transformer.applyAsDouble(p.key, p.val));
5950 result = r;
5951 CountedCompleter<?> c;
5952 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5953 @SuppressWarnings("unchecked")
5954 MapReduceMappingsToDoubleTask<K,V>
5955 t = (MapReduceMappingsToDoubleTask<K,V>)c,
5956 s = t.rights;
5957 while (s != null) {
5958 t.result = reducer.applyAsDouble(t.result, s.result);
5959 s = t.rights = s.nextRight;
5960 }
5961 }
5962 }
5963 }
5964 }
5965
5966 @SuppressWarnings("serial")
5967 static final class MapReduceKeysToLongTask<K,V>
5968 extends BulkTask<K,V,Long> {
5969 final ToLongFunction<? super K> transformer;
5970 final LongBinaryOperator reducer;
5971 final long basis;
5972 long result;
5973 MapReduceKeysToLongTask<K,V> rights, nextRight;
5974 MapReduceKeysToLongTask
5975 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5976 MapReduceKeysToLongTask<K,V> nextRight,
5977 ToLongFunction<? super K> transformer,
5978 long basis,
5979 LongBinaryOperator reducer) {
5980 super(p, b, i, f, t); this.nextRight = nextRight;
5981 this.transformer = transformer;
5982 this.basis = basis; this.reducer = reducer;
5983 }
5984 public final Long getRawResult() { return result; }
5985 public final void compute() {
5986 final ToLongFunction<? super K> transformer;
5987 final LongBinaryOperator reducer;
5988 if ((transformer = this.transformer) != null &&
5989 (reducer = this.reducer) != null) {
5990 long r = this.basis;
5991 for (int i = baseIndex, f, h; batch > 0 &&
5992 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5993 addToPendingCount(1);
5994 (rights = new MapReduceKeysToLongTask<K,V>
5995 (this, batch >>>= 1, baseLimit = h, f, tab,
5996 rights, transformer, r, reducer)).fork();
5997 }
5998 for (Node<K,V> p; (p = advance()) != null; )
5999 r = reducer.applyAsLong(r, transformer.applyAsLong(p.key));
6000 result = r;
6001 CountedCompleter<?> c;
6002 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6003 @SuppressWarnings("unchecked")
6004 MapReduceKeysToLongTask<K,V>
6005 t = (MapReduceKeysToLongTask<K,V>)c,
6006 s = t.rights;
6007 while (s != null) {
6008 t.result = reducer.applyAsLong(t.result, s.result);
6009 s = t.rights = s.nextRight;
6010 }
6011 }
6012 }
6013 }
6014 }
6015
6016 @SuppressWarnings("serial")
6017 static final class MapReduceValuesToLongTask<K,V>
6018 extends BulkTask<K,V,Long> {
6019 final ToLongFunction<? super V> transformer;
6020 final LongBinaryOperator reducer;
6021 final long basis;
6022 long result;
6023 MapReduceValuesToLongTask<K,V> rights, nextRight;
6024 MapReduceValuesToLongTask
6025 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6026 MapReduceValuesToLongTask<K,V> nextRight,
6027 ToLongFunction<? super V> transformer,
6028 long basis,
6029 LongBinaryOperator reducer) {
6030 super(p, b, i, f, t); this.nextRight = nextRight;
6031 this.transformer = transformer;
6032 this.basis = basis; this.reducer = reducer;
6033 }
6034 public final Long getRawResult() { return result; }
6035 public final void compute() {
6036 final ToLongFunction<? super V> transformer;
6037 final LongBinaryOperator reducer;
6038 if ((transformer = this.transformer) != null &&
6039 (reducer = this.reducer) != null) {
6040 long r = this.basis;
6041 for (int i = baseIndex, f, h; batch > 0 &&
6042 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6043 addToPendingCount(1);
6044 (rights = new MapReduceValuesToLongTask<K,V>
6045 (this, batch >>>= 1, baseLimit = h, f, tab,
6046 rights, transformer, r, reducer)).fork();
6047 }
6048 for (Node<K,V> p; (p = advance()) != null; )
6049 r = reducer.applyAsLong(r, transformer.applyAsLong(p.val));
6050 result = r;
6051 CountedCompleter<?> c;
6052 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6053 @SuppressWarnings("unchecked")
6054 MapReduceValuesToLongTask<K,V>
6055 t = (MapReduceValuesToLongTask<K,V>)c,
6056 s = t.rights;
6057 while (s != null) {
6058 t.result = reducer.applyAsLong(t.result, s.result);
6059 s = t.rights = s.nextRight;
6060 }
6061 }
6062 }
6063 }
6064 }
6065
6066 @SuppressWarnings("serial")
6067 static final class MapReduceEntriesToLongTask<K,V>
6068 extends BulkTask<K,V,Long> {
6069 final ToLongFunction<Map.Entry<K,V>> transformer;
6070 final LongBinaryOperator reducer;
6071 final long basis;
6072 long result;
6073 MapReduceEntriesToLongTask<K,V> rights, nextRight;
6074 MapReduceEntriesToLongTask
6075 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6076 MapReduceEntriesToLongTask<K,V> nextRight,
6077 ToLongFunction<Map.Entry<K,V>> transformer,
6078 long basis,
6079 LongBinaryOperator reducer) {
6080 super(p, b, i, f, t); this.nextRight = nextRight;
6081 this.transformer = transformer;
6082 this.basis = basis; this.reducer = reducer;
6083 }
6084 public final Long getRawResult() { return result; }
6085 public final void compute() {
6086 final ToLongFunction<Map.Entry<K,V>> transformer;
6087 final LongBinaryOperator reducer;
6088 if ((transformer = this.transformer) != null &&
6089 (reducer = this.reducer) != null) {
6090 long r = this.basis;
6091 for (int i = baseIndex, f, h; batch > 0 &&
6092 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6093 addToPendingCount(1);
6094 (rights = new MapReduceEntriesToLongTask<K,V>
6095 (this, batch >>>= 1, baseLimit = h, f, tab,
6096 rights, transformer, r, reducer)).fork();
6097 }
6098 for (Node<K,V> p; (p = advance()) != null; )
6099 r = reducer.applyAsLong(r, transformer.applyAsLong(p));
6100 result = r;
6101 CountedCompleter<?> c;
6102 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6103 @SuppressWarnings("unchecked")
6104 MapReduceEntriesToLongTask<K,V>
6105 t = (MapReduceEntriesToLongTask<K,V>)c,
6106 s = t.rights;
6107 while (s != null) {
6108 t.result = reducer.applyAsLong(t.result, s.result);
6109 s = t.rights = s.nextRight;
6110 }
6111 }
6112 }
6113 }
6114 }
6115
6116 @SuppressWarnings("serial")
6117 static final class MapReduceMappingsToLongTask<K,V>
6118 extends BulkTask<K,V,Long> {
6119 final ToLongBiFunction<? super K, ? super V> transformer;
6120 final LongBinaryOperator reducer;
6121 final long basis;
6122 long result;
6123 MapReduceMappingsToLongTask<K,V> rights, nextRight;
6124 MapReduceMappingsToLongTask
6125 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6126 MapReduceMappingsToLongTask<K,V> nextRight,
6127 ToLongBiFunction<? super K, ? super V> transformer,
6128 long basis,
6129 LongBinaryOperator reducer) {
6130 super(p, b, i, f, t); this.nextRight = nextRight;
6131 this.transformer = transformer;
6132 this.basis = basis; this.reducer = reducer;
6133 }
6134 public final Long getRawResult() { return result; }
6135 public final void compute() {
6136 final ToLongBiFunction<? super K, ? super V> transformer;
6137 final LongBinaryOperator reducer;
6138 if ((transformer = this.transformer) != null &&
6139 (reducer = this.reducer) != null) {
6140 long r = this.basis;
6141 for (int i = baseIndex, f, h; batch > 0 &&
6142 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6143 addToPendingCount(1);
6144 (rights = new MapReduceMappingsToLongTask<K,V>
6145 (this, batch >>>= 1, baseLimit = h, f, tab,
6146 rights, transformer, r, reducer)).fork();
6147 }
6148 for (Node<K,V> p; (p = advance()) != null; )
6149 r = reducer.applyAsLong(r, transformer.applyAsLong(p.key, p.val));
6150 result = r;
6151 CountedCompleter<?> c;
6152 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6153 @SuppressWarnings("unchecked")
6154 MapReduceMappingsToLongTask<K,V>
6155 t = (MapReduceMappingsToLongTask<K,V>)c,
6156 s = t.rights;
6157 while (s != null) {
6158 t.result = reducer.applyAsLong(t.result, s.result);
6159 s = t.rights = s.nextRight;
6160 }
6161 }
6162 }
6163 }
6164 }
6165
6166 @SuppressWarnings("serial")
6167 static final class MapReduceKeysToIntTask<K,V>
6168 extends BulkTask<K,V,Integer> {
6169 final ToIntFunction<? super K> transformer;
6170 final IntBinaryOperator reducer;
6171 final int basis;
6172 int result;
6173 MapReduceKeysToIntTask<K,V> rights, nextRight;
6174 MapReduceKeysToIntTask
6175 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6176 MapReduceKeysToIntTask<K,V> nextRight,
6177 ToIntFunction<? super K> transformer,
6178 int basis,
6179 IntBinaryOperator reducer) {
6180 super(p, b, i, f, t); this.nextRight = nextRight;
6181 this.transformer = transformer;
6182 this.basis = basis; this.reducer = reducer;
6183 }
6184 public final Integer getRawResult() { return result; }
6185 public final void compute() {
6186 final ToIntFunction<? super K> transformer;
6187 final IntBinaryOperator reducer;
6188 if ((transformer = this.transformer) != null &&
6189 (reducer = this.reducer) != null) {
6190 int r = this.basis;
6191 for (int i = baseIndex, f, h; batch > 0 &&
6192 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6193 addToPendingCount(1);
6194 (rights = new MapReduceKeysToIntTask<K,V>
6195 (this, batch >>>= 1, baseLimit = h, f, tab,
6196 rights, transformer, r, reducer)).fork();
6197 }
6198 for (Node<K,V> p; (p = advance()) != null; )
6199 r = reducer.applyAsInt(r, transformer.applyAsInt(p.key));
6200 result = r;
6201 CountedCompleter<?> c;
6202 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6203 @SuppressWarnings("unchecked")
6204 MapReduceKeysToIntTask<K,V>
6205 t = (MapReduceKeysToIntTask<K,V>)c,
6206 s = t.rights;
6207 while (s != null) {
6208 t.result = reducer.applyAsInt(t.result, s.result);
6209 s = t.rights = s.nextRight;
6210 }
6211 }
6212 }
6213 }
6214 }
6215
6216 @SuppressWarnings("serial")
6217 static final class MapReduceValuesToIntTask<K,V>
6218 extends BulkTask<K,V,Integer> {
6219 final ToIntFunction<? super V> transformer;
6220 final IntBinaryOperator reducer;
6221 final int basis;
6222 int result;
6223 MapReduceValuesToIntTask<K,V> rights, nextRight;
6224 MapReduceValuesToIntTask
6225 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6226 MapReduceValuesToIntTask<K,V> nextRight,
6227 ToIntFunction<? super V> transformer,
6228 int basis,
6229 IntBinaryOperator reducer) {
6230 super(p, b, i, f, t); this.nextRight = nextRight;
6231 this.transformer = transformer;
6232 this.basis = basis; this.reducer = reducer;
6233 }
6234 public final Integer getRawResult() { return result; }
6235 public final void compute() {
6236 final ToIntFunction<? super V> transformer;
6237 final IntBinaryOperator reducer;
6238 if ((transformer = this.transformer) != null &&
6239 (reducer = this.reducer) != null) {
6240 int r = this.basis;
6241 for (int i = baseIndex, f, h; batch > 0 &&
6242 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6243 addToPendingCount(1);
6244 (rights = new MapReduceValuesToIntTask<K,V>
6245 (this, batch >>>= 1, baseLimit = h, f, tab,
6246 rights, transformer, r, reducer)).fork();
6247 }
6248 for (Node<K,V> p; (p = advance()) != null; )
6249 r = reducer.applyAsInt(r, transformer.applyAsInt(p.val));
6250 result = r;
6251 CountedCompleter<?> c;
6252 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6253 @SuppressWarnings("unchecked")
6254 MapReduceValuesToIntTask<K,V>
6255 t = (MapReduceValuesToIntTask<K,V>)c,
6256 s = t.rights;
6257 while (s != null) {
6258 t.result = reducer.applyAsInt(t.result, s.result);
6259 s = t.rights = s.nextRight;
6260 }
6261 }
6262 }
6263 }
6264 }
6265
6266 @SuppressWarnings("serial")
6267 static final class MapReduceEntriesToIntTask<K,V>
6268 extends BulkTask<K,V,Integer> {
6269 final ToIntFunction<Map.Entry<K,V>> transformer;
6270 final IntBinaryOperator reducer;
6271 final int basis;
6272 int result;
6273 MapReduceEntriesToIntTask<K,V> rights, nextRight;
6274 MapReduceEntriesToIntTask
6275 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6276 MapReduceEntriesToIntTask<K,V> nextRight,
6277 ToIntFunction<Map.Entry<K,V>> transformer,
6278 int basis,
6279 IntBinaryOperator reducer) {
6280 super(p, b, i, f, t); this.nextRight = nextRight;
6281 this.transformer = transformer;
6282 this.basis = basis; this.reducer = reducer;
6283 }
6284 public final Integer getRawResult() { return result; }
6285 public final void compute() {
6286 final ToIntFunction<Map.Entry<K,V>> transformer;
6287 final IntBinaryOperator reducer;
6288 if ((transformer = this.transformer) != null &&
6289 (reducer = this.reducer) != null) {
6290 int r = this.basis;
6291 for (int i = baseIndex, f, h; batch > 0 &&
6292 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6293 addToPendingCount(1);
6294 (rights = new MapReduceEntriesToIntTask<K,V>
6295 (this, batch >>>= 1, baseLimit = h, f, tab,
6296 rights, transformer, r, reducer)).fork();
6297 }
6298 for (Node<K,V> p; (p = advance()) != null; )
6299 r = reducer.applyAsInt(r, transformer.applyAsInt(p));
6300 result = r;
6301 CountedCompleter<?> c;
6302 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6303 @SuppressWarnings("unchecked")
6304 MapReduceEntriesToIntTask<K,V>
6305 t = (MapReduceEntriesToIntTask<K,V>)c,
6306 s = t.rights;
6307 while (s != null) {
6308 t.result = reducer.applyAsInt(t.result, s.result);
6309 s = t.rights = s.nextRight;
6310 }
6311 }
6312 }
6313 }
6314 }
6315
6316 @SuppressWarnings("serial")
6317 static final class MapReduceMappingsToIntTask<K,V>
6318 extends BulkTask<K,V,Integer> {
6319 final ToIntBiFunction<? super K, ? super V> transformer;
6320 final IntBinaryOperator reducer;
6321 final int basis;
6322 int result;
6323 MapReduceMappingsToIntTask<K,V> rights, nextRight;
6324 MapReduceMappingsToIntTask
6325 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
6326 MapReduceMappingsToIntTask<K,V> nextRight,
6327 ToIntBiFunction<? super K, ? super V> transformer,
6328 int basis,
6329 IntBinaryOperator reducer) {
6330 super(p, b, i, f, t); this.nextRight = nextRight;
6331 this.transformer = transformer;
6332 this.basis = basis; this.reducer = reducer;
6333 }
6334 public final Integer getRawResult() { return result; }
6335 public final void compute() {
6336 final ToIntBiFunction<? super K, ? super V> transformer;
6337 final IntBinaryOperator reducer;
6338 if ((transformer = this.transformer) != null &&
6339 (reducer = this.reducer) != null) {
6340 int r = this.basis;
6341 for (int i = baseIndex, f, h; batch > 0 &&
6342 (h = ((f = baseLimit) + i) >>> 1) > i;) {
6343 addToPendingCount(1);
6344 (rights = new MapReduceMappingsToIntTask<K,V>
6345 (this, batch >>>= 1, baseLimit = h, f, tab,
6346 rights, transformer, r, reducer)).fork();
6347 }
6348 for (Node<K,V> p; (p = advance()) != null; )
6349 r = reducer.applyAsInt(r, transformer.applyAsInt(p.key, p.val));
6350 result = r;
6351 CountedCompleter<?> c;
6352 for (c = firstComplete(); c != null; c = c.nextComplete()) {
6353 @SuppressWarnings("unchecked")
6354 MapReduceMappingsToIntTask<K,V>
6355 t = (MapReduceMappingsToIntTask<K,V>)c,
6356 s = t.rights;
6357 while (s != null) {
6358 t.result = reducer.applyAsInt(t.result, s.result);
6359 s = t.rights = s.nextRight;
6360 }
6361 }
6362 }
6363 }
6364 }
6365
6366 // Unsafe mechanics
6367 private static final Unsafe U = Unsafe.getUnsafe();
6368 private static final long SIZECTL;
6369 private static final long TRANSFERINDEX;
6370 private static final long BASECOUNT;
6371 private static final long CELLSBUSY;
6372 private static final long CELLVALUE;
6373 private static final int ABASE;
6374 private static final int ASHIFT;
6375
6376 static {
6377 SIZECTL = U.objectFieldOffset
6378 (ConcurrentHashMap.class, "sizeCtl");
6379 TRANSFERINDEX = U.objectFieldOffset
6380 (ConcurrentHashMap.class, "transferIndex");
6381 BASECOUNT = U.objectFieldOffset
6382 (ConcurrentHashMap.class, "baseCount");
6383 CELLSBUSY = U.objectFieldOffset
6384 (ConcurrentHashMap.class, "cellsBusy");
6385
6386 CELLVALUE = U.objectFieldOffset
6387 (CounterCell.class, "value");
6388
6389 ABASE = U.arrayBaseOffset(Node[].class);
6390 int scale = U.arrayIndexScale(Node[].class);
6391 if ((scale & (scale - 1)) != 0)
6392 throw new Error("array index scale not a power of two");
6393 ASHIFT = 31 - Integer.numberOfLeadingZeros(scale);
6394
6395 // Reduce the risk of rare disastrous classloading in first call to
6396 // LockSupport.park: https://bugs.openjdk.java.net/browse/JDK-8074773
6397 Class<?> ensureLoaded = LockSupport.class;
6398 }
6399 }