ViewVC Help
View File | Revision Log | Show Annotations | Download File | Root Listing
root/jsr166/jsr166/src/main/java/util/concurrent/ConcurrentHashMap.java
Revision: 1.295
Committed: Sun Jul 17 04:23:31 2016 UTC (7 years, 10 months ago) by jsr166
Branch: MAIN
Changes since 1.294: +24 -5 lines
Log Message:
optimize view set removeAll using heuristics

File Contents

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