ViewVC Help
View File | Revision Log | Show Annotations | Download File | Root Listing
root/jsr166/jsr166/src/jsr166e/ConcurrentHashMapV8.java
Revision: 1.110
Committed: Thu Jul 4 18:34:49 2013 UTC (10 years, 10 months ago) by dl
Branch: MAIN
Changes since 1.109: +31 -16 lines
Log Message:
Avoid unbounded recursion

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 jsr166e;
8
9 import jsr166e.ForkJoinPool;
10
11 import java.io.ObjectStreamField;
12 import java.io.Serializable;
13 import java.lang.reflect.ParameterizedType;
14 import java.lang.reflect.Type;
15 import java.util.Arrays;
16 import java.util.Collection;
17 import java.util.Comparator;
18 import java.util.ConcurrentModificationException;
19 import java.util.Enumeration;
20 import java.util.HashMap;
21 import java.util.Hashtable;
22 import java.util.Iterator;
23 import java.util.Map;
24 import java.util.NoSuchElementException;
25 import java.util.Set;
26 import java.util.concurrent.ConcurrentMap;
27 import java.util.concurrent.atomic.AtomicReference;
28 import java.util.concurrent.atomic.AtomicInteger;
29 import java.util.concurrent.locks.LockSupport;
30 import java.util.concurrent.locks.ReentrantLock;
31
32 /**
33 * A hash table supporting full concurrency of retrievals and
34 * high expected concurrency for updates. This class obeys the
35 * same functional specification as {@link java.util.Hashtable}, and
36 * includes versions of methods corresponding to each method of
37 * {@code Hashtable}. However, even though all operations are
38 * thread-safe, retrieval operations do <em>not</em> entail locking,
39 * and there is <em>not</em> any support for locking the entire table
40 * in a way that prevents all access. This class is fully
41 * interoperable with {@code Hashtable} in programs that rely on its
42 * thread safety but not on its synchronization details.
43 *
44 * <p>Retrieval operations (including {@code get}) generally do not
45 * block, so may overlap with update operations (including {@code put}
46 * and {@code remove}). Retrievals reflect the results of the most
47 * recently <em>completed</em> update operations holding upon their
48 * onset. (More formally, an update operation for a given key bears a
49 * <em>happens-before</em> relation with any (non-null) retrieval for
50 * that key reporting the updated value.) For aggregate operations
51 * such as {@code putAll} and {@code clear}, concurrent retrievals may
52 * reflect insertion or removal of only some entries. Similarly,
53 * Iterators and Enumerations return elements reflecting the state of
54 * the hash table at some point at or since the creation of the
55 * iterator/enumeration. They do <em>not</em> throw {@link
56 * ConcurrentModificationException}. However, iterators are designed
57 * to be used by only one thread at a time. Bear in mind that the
58 * results of aggregate status methods including {@code size}, {@code
59 * isEmpty}, and {@code containsValue} are typically useful only when
60 * a map is not undergoing concurrent updates in other threads.
61 * Otherwise the results of these methods reflect transient states
62 * that may be adequate for monitoring or estimation purposes, but not
63 * for program control.
64 *
65 * <p>The table is dynamically expanded when there are too many
66 * collisions (i.e., keys that have distinct hash codes but fall into
67 * the same slot modulo the table size), with the expected average
68 * effect of maintaining roughly two bins per mapping (corresponding
69 * to a 0.75 load factor threshold for resizing). There may be much
70 * variance around this average as mappings are added and removed, but
71 * overall, this maintains a commonly accepted time/space tradeoff for
72 * hash tables. However, resizing this or any other kind of hash
73 * table may be a relatively slow operation. When possible, it is a
74 * good idea to provide a size estimate as an optional {@code
75 * initialCapacity} constructor argument. An additional optional
76 * {@code loadFactor} constructor argument provides a further means of
77 * customizing initial table capacity by specifying the table density
78 * to be used in calculating the amount of space to allocate for the
79 * given number of elements. Also, for compatibility with previous
80 * versions of this class, constructors may optionally specify an
81 * expected {@code concurrencyLevel} as an additional hint for
82 * internal sizing. Note that using many keys with exactly the same
83 * {@code hashCode()} is a sure way to slow down performance of any
84 * hash table. To ameliorate impact, when keys are {@link Comparable},
85 * this class may use comparison order among keys to help break ties.
86 *
87 * <p>A {@link Set} projection of a ConcurrentHashMapV8 may be created
88 * (using {@link #newKeySet()} or {@link #newKeySet(int)}), or viewed
89 * (using {@link #keySet(Object)} when only keys are of interest, and the
90 * mapped values are (perhaps transiently) not used or all take the
91 * same mapping value.
92 *
93 * <p>This class and its views and iterators implement all of the
94 * <em>optional</em> methods of the {@link Map} and {@link Iterator}
95 * interfaces.
96 *
97 * <p>Like {@link Hashtable} but unlike {@link HashMap}, this class
98 * does <em>not</em> allow {@code null} to be used as a key or value.
99 *
100 * <p>ConcurrentHashMapV8s support a set of sequential and parallel bulk
101 * operations that are designed
102 * to be safely, and often sensibly, applied even with maps that are
103 * being concurrently updated by other threads; for example, when
104 * computing a snapshot summary of the values in a shared registry.
105 * There are three kinds of operation, each with four forms, accepting
106 * functions with Keys, Values, Entries, and (Key, Value) arguments
107 * and/or return values. Because the elements of a ConcurrentHashMapV8
108 * are not ordered in any particular way, and may be processed in
109 * different orders in different parallel executions, the correctness
110 * of supplied functions should not depend on any ordering, or on any
111 * other objects or values that may transiently change while
112 * computation is in progress; and except for forEach actions, should
113 * ideally be side-effect-free. Bulk operations on {@link java.util.Map.Entry}
114 * objects do not support method {@code setValue}.
115 *
116 * <ul>
117 * <li> forEach: Perform a given action on each element.
118 * A variant form applies a given transformation on each element
119 * before performing the action.</li>
120 *
121 * <li> search: Return the first available non-null result of
122 * applying a given function on each element; skipping further
123 * search when a result is found.</li>
124 *
125 * <li> reduce: Accumulate each element. The supplied reduction
126 * function cannot rely on ordering (more formally, it should be
127 * both associative and commutative). There are five variants:
128 *
129 * <ul>
130 *
131 * <li> Plain reductions. (There is not a form of this method for
132 * (key, value) function arguments since there is no corresponding
133 * return type.)</li>
134 *
135 * <li> Mapped reductions that accumulate the results of a given
136 * function applied to each element.</li>
137 *
138 * <li> Reductions to scalar doubles, longs, and ints, using a
139 * given basis value.</li>
140 *
141 * </ul>
142 * </li>
143 * </ul>
144 *
145 * <p>These bulk operations accept a {@code parallelismThreshold}
146 * argument. Methods proceed sequentially if the current map size is
147 * estimated to be less than the given threshold. Using a value of
148 * {@code Long.MAX_VALUE} suppresses all parallelism. Using a value
149 * of {@code 1} results in maximal parallelism by partitioning into
150 * enough subtasks to fully utilize the {@link
151 * ForkJoinPool#commonPool()} that is used for all parallel
152 * computations. Normally, you would initially choose one of these
153 * extreme values, and then measure performance of using in-between
154 * values that trade off overhead versus throughput.
155 *
156 * <p>The concurrency properties of bulk operations follow
157 * from those of ConcurrentHashMapV8: Any non-null result returned
158 * from {@code get(key)} and related access methods bears a
159 * happens-before relation with the associated insertion or
160 * update. The result of any bulk operation reflects the
161 * composition of these per-element relations (but is not
162 * necessarily atomic with respect to the map as a whole unless it
163 * is somehow known to be quiescent). Conversely, because keys
164 * and values in the map are never null, null serves as a reliable
165 * atomic indicator of the current lack of any result. To
166 * maintain this property, null serves as an implicit basis for
167 * all non-scalar reduction operations. For the double, long, and
168 * int versions, the basis should be one that, when combined with
169 * any other value, returns that other value (more formally, it
170 * should be the identity element for the reduction). Most common
171 * reductions have these properties; for example, computing a sum
172 * with basis 0 or a minimum with basis MAX_VALUE.
173 *
174 * <p>Search and transformation functions provided as arguments
175 * should similarly return null to indicate the lack of any result
176 * (in which case it is not used). In the case of mapped
177 * reductions, this also enables transformations to serve as
178 * filters, returning null (or, in the case of primitive
179 * specializations, the identity basis) if the element should not
180 * be combined. You can create compound transformations and
181 * filterings by composing them yourself under this "null means
182 * there is nothing there now" rule before using them in search or
183 * reduce operations.
184 *
185 * <p>Methods accepting and/or returning Entry arguments maintain
186 * key-value associations. They may be useful for example when
187 * finding the key for the greatest value. Note that "plain" Entry
188 * arguments can be supplied using {@code new
189 * AbstractMap.SimpleEntry(k,v)}.
190 *
191 * <p>Bulk operations may complete abruptly, throwing an
192 * exception encountered in the application of a supplied
193 * function. Bear in mind when handling such exceptions that other
194 * concurrently executing functions could also have thrown
195 * exceptions, or would have done so if the first exception had
196 * not occurred.
197 *
198 * <p>Speedups for parallel compared to sequential forms are common
199 * but not guaranteed. Parallel operations involving brief functions
200 * on small maps may execute more slowly than sequential forms if the
201 * underlying work to parallelize the computation is more expensive
202 * than the computation itself. Similarly, parallelization may not
203 * lead to much actual parallelism if all processors are busy
204 * performing unrelated tasks.
205 *
206 * <p>All arguments to all task methods must be non-null.
207 *
208 * <p><em>jsr166e note: During transition, this class
209 * uses nested functional interfaces with different names but the
210 * same forms as those expected for JDK8.</em>
211 *
212 * <p>This class is a member of the
213 * <a href="{@docRoot}/../technotes/guides/collections/index.html">
214 * Java Collections Framework</a>.
215 *
216 * @since 1.5
217 * @author Doug Lea
218 * @param <K> the type of keys maintained by this map
219 * @param <V> the type of mapped values
220 */
221 public class ConcurrentHashMapV8<K,V>
222 implements ConcurrentMap<K,V>, Serializable {
223 private static final long serialVersionUID = 7249069246763182397L;
224
225 /**
226 * An object for traversing and partitioning elements of a source.
227 * This interface provides a subset of the functionality of JDK8
228 * java.util.Spliterator.
229 */
230 public static interface ConcurrentHashMapSpliterator<T> {
231 /**
232 * If possible, returns a new spliterator covering
233 * approximately one half of the elements, which will not be
234 * covered by this spliterator. Returns null if cannot be
235 * split.
236 */
237 ConcurrentHashMapSpliterator<T> trySplit();
238 /**
239 * Returns an estimate of the number of elements covered by
240 * this Spliterator.
241 */
242 long estimateSize();
243
244 /** Applies the action to each untraversed element */
245 void forEachRemaining(Action<? super T> action);
246 /** If an element remains, applies the action and returns true. */
247 boolean tryAdvance(Action<? super T> action);
248 }
249
250 // Sams
251 /** Interface describing a void action of one argument */
252 public interface Action<A> { void apply(A a); }
253 /** Interface describing a void action of two arguments */
254 public interface BiAction<A,B> { void apply(A a, B b); }
255 /** Interface describing a function of one argument */
256 public interface Fun<A,T> { T apply(A a); }
257 /** Interface describing a function of two arguments */
258 public interface BiFun<A,B,T> { T apply(A a, B b); }
259 /** Interface describing a function mapping its argument to a double */
260 public interface ObjectToDouble<A> { double apply(A a); }
261 /** Interface describing a function mapping its argument to a long */
262 public interface ObjectToLong<A> { long apply(A a); }
263 /** Interface describing a function mapping its argument to an int */
264 public interface ObjectToInt<A> {int apply(A a); }
265 /** Interface describing a function mapping two arguments to a double */
266 public interface ObjectByObjectToDouble<A,B> { double apply(A a, B b); }
267 /** Interface describing a function mapping two arguments to a long */
268 public interface ObjectByObjectToLong<A,B> { long apply(A a, B b); }
269 /** Interface describing a function mapping two arguments to an int */
270 public interface ObjectByObjectToInt<A,B> {int apply(A a, B b); }
271 /** Interface describing a function mapping two doubles to a double */
272 public interface DoubleByDoubleToDouble { double apply(double a, double b); }
273 /** Interface describing a function mapping two longs to a long */
274 public interface LongByLongToLong { long apply(long a, long b); }
275 /** Interface describing a function mapping two ints to an int */
276 public interface IntByIntToInt { int apply(int a, int b); }
277
278 /*
279 * Overview:
280 *
281 * The primary design goal of this hash table is to maintain
282 * concurrent readability (typically method get(), but also
283 * iterators and related methods) while minimizing update
284 * contention. Secondary goals are to keep space consumption about
285 * the same or better than java.util.HashMap, and to support high
286 * initial insertion rates on an empty table by many threads.
287 *
288 * This map usually acts as a binned (bucketed) hash table. Each
289 * key-value mapping is held in a Node. Most nodes are instances
290 * of the basic Node class with hash, key, value, and next
291 * fields. However, various subclasses exist: TreeNodes are
292 * arranged in balanced trees, not lists. TreeBins hold the roots
293 * of sets of TreeNodes. ForwardingNodes are placed at the heads
294 * of bins during resizing. ReservationNodes are used as
295 * placeholders while establishing values in computeIfAbsent and
296 * related methods. The types TreeBin, ForwardingNode, and
297 * ReservationNode do not hold normal user keys, values, or
298 * hashes, and are readily distinguishable during search etc
299 * because they have negative hash fields and null key and value
300 * fields. (These special nodes are either uncommon or transient,
301 * so the impact of carrying around some unused fields is
302 * insignificant.)
303 *
304 * The table is lazily initialized to a power-of-two size upon the
305 * first insertion. Each bin in the table normally contains a
306 * list of Nodes (most often, the list has only zero or one Node).
307 * Table accesses require volatile/atomic reads, writes, and
308 * CASes. Because there is no other way to arrange this without
309 * adding further indirections, we use intrinsics
310 * (sun.misc.Unsafe) operations.
311 *
312 * We use the top (sign) bit of Node hash fields for control
313 * purposes -- it is available anyway because of addressing
314 * constraints. Nodes with negative hash fields are specially
315 * handled or ignored in map methods.
316 *
317 * Insertion (via put or its variants) of the first node in an
318 * empty bin is performed by just CASing it to the bin. This is
319 * by far the most common case for put operations under most
320 * key/hash distributions. Other update operations (insert,
321 * delete, and replace) require locks. We do not want to waste
322 * the space required to associate a distinct lock object with
323 * each bin, so instead use the first node of a bin list itself as
324 * a lock. Locking support for these locks relies on builtin
325 * "synchronized" monitors.
326 *
327 * Using the first node of a list as a lock does not by itself
328 * suffice though: When a node is locked, any update must first
329 * validate that it is still the first node after locking it, and
330 * retry if not. Because new nodes are always appended to lists,
331 * once a node is first in a bin, it remains first until deleted
332 * or the bin becomes invalidated (upon resizing).
333 *
334 * The main disadvantage of per-bin locks is that other update
335 * operations on other nodes in a bin list protected by the same
336 * lock can stall, for example when user equals() or mapping
337 * functions take a long time. However, statistically, under
338 * random hash codes, this is not a common problem. Ideally, the
339 * frequency of nodes in bins follows a Poisson distribution
340 * (http://en.wikipedia.org/wiki/Poisson_distribution) with a
341 * parameter of about 0.5 on average, given the resizing threshold
342 * of 0.75, although with a large variance because of resizing
343 * granularity. Ignoring variance, the expected occurrences of
344 * list size k are (exp(-0.5) * pow(0.5, k) / factorial(k)). The
345 * first values are:
346 *
347 * 0: 0.60653066
348 * 1: 0.30326533
349 * 2: 0.07581633
350 * 3: 0.01263606
351 * 4: 0.00157952
352 * 5: 0.00015795
353 * 6: 0.00001316
354 * 7: 0.00000094
355 * 8: 0.00000006
356 * more: less than 1 in ten million
357 *
358 * Lock contention probability for two threads accessing distinct
359 * elements is roughly 1 / (8 * #elements) under random hashes.
360 *
361 * Actual hash code distributions encountered in practice
362 * sometimes deviate significantly from uniform randomness. This
363 * includes the case when N > (1<<30), so some keys MUST collide.
364 * Similarly for dumb or hostile usages in which multiple keys are
365 * designed to have identical hash codes or ones that differs only
366 * in masked-out high bits. So we use a secondary strategy that
367 * applies when the number of nodes in a bin exceeds a
368 * threshold. These TreeBins use a balanced tree to hold nodes (a
369 * specialized form of red-black trees), bounding search time to
370 * O(log N). Each search step in a TreeBin is at least twice as
371 * slow as in a regular list, but given that N cannot exceed
372 * (1<<64) (before running out of addresses) this bounds search
373 * steps, lock hold times, etc, to reasonable constants (roughly
374 * 100 nodes inspected per operation worst case) so long as keys
375 * are Comparable (which is very common -- String, Long, etc).
376 * TreeBin nodes (TreeNodes) also maintain the same "next"
377 * traversal pointers as regular nodes, so can be traversed in
378 * iterators in the same way.
379 *
380 * The table is resized when occupancy exceeds a percentage
381 * threshold (nominally, 0.75, but see below). Any thread
382 * noticing an overfull bin may assist in resizing after the
383 * initiating thread allocates and sets up the replacement
384 * array. However, rather than stalling, these other threads may
385 * proceed with insertions etc. The use of TreeBins shields us
386 * from the worst case effects of overfilling while resizes are in
387 * progress. Resizing proceeds by transferring bins, one by one,
388 * from the table to the next table. To enable concurrency, the
389 * next table must be (incrementally) prefilled with place-holders
390 * serving as reverse forwarders to the old table. Because we are
391 * using power-of-two expansion, the elements from each bin must
392 * either stay at same index, or move with a power of two
393 * offset. We eliminate unnecessary node creation by catching
394 * cases where old nodes can be reused because their next fields
395 * won't change. On average, only about one-sixth of them need
396 * cloning when a table doubles. The nodes they replace will be
397 * garbage collectable as soon as they are no longer referenced by
398 * any reader thread that may be in the midst of concurrently
399 * traversing table. Upon transfer, the old table bin contains
400 * only a special forwarding node (with hash field "MOVED") that
401 * contains the next table as its key. On encountering a
402 * forwarding node, access and update operations restart, using
403 * the new table.
404 *
405 * Each bin transfer requires its bin lock, which can stall
406 * waiting for locks while resizing. However, because other
407 * threads can join in and help resize rather than contend for
408 * locks, average aggregate waits become shorter as resizing
409 * progresses. The transfer operation must also ensure that all
410 * accessible bins in both the old and new table are usable by any
411 * traversal. This is arranged by proceeding from the last bin
412 * (table.length - 1) up towards the first. Upon seeing a
413 * forwarding node, traversals (see class Traverser) arrange to
414 * move to the new table without revisiting nodes. However, to
415 * ensure that no intervening nodes are skipped, bin splitting can
416 * only begin after the associated reverse-forwarders are in
417 * place.
418 *
419 * The traversal scheme also applies to partial traversals of
420 * ranges of bins (via an alternate Traverser constructor)
421 * to support partitioned aggregate operations. Also, read-only
422 * operations give up if ever forwarded to a null table, which
423 * provides support for shutdown-style clearing, which is also not
424 * currently implemented.
425 *
426 * Lazy table initialization minimizes footprint until first use,
427 * and also avoids resizings when the first operation is from a
428 * putAll, constructor with map argument, or deserialization.
429 * These cases attempt to override the initial capacity settings,
430 * but harmlessly fail to take effect in cases of races.
431 *
432 * The element count is maintained using a specialization of
433 * LongAdder. We need to incorporate a specialization rather than
434 * just use a LongAdder in order to access implicit
435 * contention-sensing that leads to creation of multiple
436 * CounterCells. The counter mechanics avoid contention on
437 * updates but can encounter cache thrashing if read too
438 * frequently during concurrent access. To avoid reading so often,
439 * resizing under contention is attempted only upon adding to a
440 * bin already holding two or more nodes. Under uniform hash
441 * distributions, the probability of this occurring at threshold
442 * is around 13%, meaning that only about 1 in 8 puts check
443 * threshold (and after resizing, many fewer do so).
444 *
445 * TreeBins use a special form of comparison for search and
446 * related operations (which is the main reason we cannot use
447 * existing collections such as TreeMaps). TreeBins contain
448 * Comparable elements, but may contain others, as well as
449 * elements that are Comparable but not necessarily Comparable
450 * for the same T, so we cannot invoke compareTo among them. To
451 * handle this, the tree is ordered primarily by hash value, then
452 * by Comparable.compareTo order if applicable. On lookup at a
453 * node, if elements are not comparable or compare as 0 then both
454 * left and right children may need to be searched in the case of
455 * tied hash values. (This corresponds to the full list search
456 * that would be necessary if all elements were non-Comparable and
457 * had tied hashes.) The red-black balancing code is updated from
458 * pre-jdk-collections
459 * (http://gee.cs.oswego.edu/dl/classes/collections/RBCell.java)
460 * based in turn on Cormen, Leiserson, and Rivest "Introduction to
461 * Algorithms" (CLR).
462 *
463 * TreeBins also require an additional locking mechanism. While
464 * list traversal is always possible by readers even during
465 * updates, tree traversal is not, mainly because of tree-rotations
466 * that may change the root node and/or its linkages. TreeBins
467 * include a simple read-write lock mechanism parasitic on the
468 * main bin-synchronization strategy: Structural adjustments
469 * associated with an insertion or removal are already bin-locked
470 * (and so cannot conflict with other writers) but must wait for
471 * ongoing readers to finish. Since there can be only one such
472 * waiter, we use a simple scheme using a single "waiter" field to
473 * block writers. However, readers need never block. If the root
474 * lock is held, they proceed along the slow traversal path (via
475 * next-pointers) until the lock becomes available or the list is
476 * exhausted, whichever comes first. These cases are not fast, but
477 * maximize aggregate expected throughput.
478 *
479 * Maintaining API and serialization compatibility with previous
480 * versions of this class introduces several oddities. Mainly: We
481 * leave untouched but unused constructor arguments refering to
482 * concurrencyLevel. We accept a loadFactor constructor argument,
483 * but apply it only to initial table capacity (which is the only
484 * time that we can guarantee to honor it.) We also declare an
485 * unused "Segment" class that is instantiated in minimal form
486 * only when serializing.
487 *
488 * This file is organized to make things a little easier to follow
489 * while reading than they might otherwise: First the main static
490 * declarations and utilities, then fields, then main public
491 * methods (with a few factorings of multiple public methods into
492 * internal ones), then sizing methods, trees, traversers, and
493 * bulk operations.
494 */
495
496 /* ---------------- Constants -------------- */
497
498 /**
499 * The largest possible table capacity. This value must be
500 * exactly 1<<30 to stay within Java array allocation and indexing
501 * bounds for power of two table sizes, and is further required
502 * because the top two bits of 32bit hash fields are used for
503 * control purposes.
504 */
505 private static final int MAXIMUM_CAPACITY = 1 << 30;
506
507 /**
508 * The default initial table capacity. Must be a power of 2
509 * (i.e., at least 1) and at most MAXIMUM_CAPACITY.
510 */
511 private static final int DEFAULT_CAPACITY = 16;
512
513 /**
514 * The largest possible (non-power of two) array size.
515 * Needed by toArray and related methods.
516 */
517 static final int MAX_ARRAY_SIZE = Integer.MAX_VALUE - 8;
518
519 /**
520 * The default concurrency level for this table. Unused but
521 * defined for compatibility with previous versions of this class.
522 */
523 private static final int DEFAULT_CONCURRENCY_LEVEL = 16;
524
525 /**
526 * The load factor for this table. Overrides of this value in
527 * constructors affect only the initial table capacity. The
528 * actual floating point value isn't normally used -- it is
529 * simpler to use expressions such as {@code n - (n >>> 2)} for
530 * the associated resizing threshold.
531 */
532 private static final float LOAD_FACTOR = 0.75f;
533
534 /**
535 * The bin count threshold for using a tree rather than list for a
536 * bin. Bins are converted to trees when adding an element to a
537 * bin with at least this many nodes. The value must be greater
538 * than 2, and should be at least 8 to mesh with assumptions in
539 * tree removal about conversion back to plain bins upon
540 * shrinkage.
541 */
542 static final int TREEIFY_THRESHOLD = 8;
543
544 /**
545 * The bin count threshold for untreeifying a (split) bin during a
546 * resize operation. Should be less than TREEIFY_THRESHOLD, and at
547 * most 6 to mesh with shrinkage detection under removal.
548 */
549 static final int UNTREEIFY_THRESHOLD = 6;
550
551 /**
552 * The smallest table capacity for which bins may be treeified.
553 * (Otherwise the table is resized if too many nodes in a bin.)
554 * The value should be at least 4 * TREEIFY_THRESHOLD to avoid
555 * conflicts between resizing and treeification thresholds.
556 */
557 static final int MIN_TREEIFY_CAPACITY = 64;
558
559 /**
560 * Minimum number of rebinnings per transfer step. Ranges are
561 * subdivided to allow multiple resizer threads. This value
562 * serves as a lower bound to avoid resizers encountering
563 * excessive memory contention. The value should be at least
564 * DEFAULT_CAPACITY.
565 */
566 private static final int MIN_TRANSFER_STRIDE = 16;
567
568 /*
569 * Encodings for Node hash fields. See above for explanation.
570 */
571 static final int MOVED = -1; // hash for forwarding nodes
572 static final int TREEBIN = -2; // hash for roots of trees
573 static final int RESERVED = -3; // hash for transient reservations
574 static final int HASH_BITS = 0x7fffffff; // usable bits of normal node hash
575
576 /** Number of CPUS, to place bounds on some sizings */
577 static final int NCPU = Runtime.getRuntime().availableProcessors();
578
579 /** For serialization compatibility. */
580 private static final ObjectStreamField[] serialPersistentFields = {
581 new ObjectStreamField("segments", Segment[].class),
582 new ObjectStreamField("segmentMask", Integer.TYPE),
583 new ObjectStreamField("segmentShift", Integer.TYPE)
584 };
585
586 /* ---------------- Nodes -------------- */
587
588 /**
589 * Key-value entry. This class is never exported out as a
590 * user-mutable Map.Entry (i.e., one supporting setValue; see
591 * MapEntry below), but can be used for read-only traversals used
592 * in bulk tasks. Subclasses of Node with a negative hash field
593 * are special, and contain null keys and values (but are never
594 * exported). Otherwise, keys and vals are never null.
595 */
596 static class Node<K,V> implements Map.Entry<K,V> {
597 final int hash;
598 final K key;
599 volatile V val;
600 volatile Node<K,V> next;
601
602 Node(int hash, K key, V val, Node<K,V> next) {
603 this.hash = hash;
604 this.key = key;
605 this.val = val;
606 this.next = next;
607 }
608
609 public final K getKey() { return key; }
610 public final V getValue() { return val; }
611 public final int hashCode() { return key.hashCode() ^ val.hashCode(); }
612 public final String toString(){ return key + "=" + val; }
613 public final V setValue(V value) {
614 throw new UnsupportedOperationException();
615 }
616
617 public final boolean equals(Object o) {
618 Object k, v, u; Map.Entry<?,?> e;
619 return ((o instanceof Map.Entry) &&
620 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
621 (v = e.getValue()) != null &&
622 (k == key || k.equals(key)) &&
623 (v == (u = val) || v.equals(u)));
624 }
625
626 /**
627 * Virtualized support for map.get(); overridden in subclasses.
628 */
629 Node<K,V> find(int h, Object k) {
630 Node<K,V> e = this;
631 if (k != null) {
632 do {
633 K ek;
634 if (e.hash == h &&
635 ((ek = e.key) == k || (ek != null && k.equals(ek))))
636 return e;
637 } while ((e = e.next) != null);
638 }
639 return null;
640 }
641 }
642
643 /* ---------------- Static utilities -------------- */
644
645 /**
646 * Spreads (XORs) higher bits of hash to lower and also forces top
647 * bit to 0. Because the table uses power-of-two masking, sets of
648 * hashes that vary only in bits above the current mask will
649 * always collide. (Among known examples are sets of Float keys
650 * holding consecutive whole numbers in small tables.) So we
651 * apply a transform that spreads the impact of higher bits
652 * downward. There is a tradeoff between speed, utility, and
653 * quality of bit-spreading. Because many common sets of hashes
654 * are already reasonably distributed (so don't benefit from
655 * spreading), and because we use trees to handle large sets of
656 * collisions in bins, we just XOR some shifted bits in the
657 * cheapest possible way to reduce systematic lossage, as well as
658 * to incorporate impact of the highest bits that would otherwise
659 * never be used in index calculations because of table bounds.
660 */
661 static final int spread(int h) {
662 return (h ^ (h >>> 16)) & HASH_BITS;
663 }
664
665 /**
666 * Returns a power of two table size for the given desired capacity.
667 * See Hackers Delight, sec 3.2
668 */
669 private static final int tableSizeFor(int c) {
670 int n = c - 1;
671 n |= n >>> 1;
672 n |= n >>> 2;
673 n |= n >>> 4;
674 n |= n >>> 8;
675 n |= n >>> 16;
676 return (n < 0) ? 1 : (n >= MAXIMUM_CAPACITY) ? MAXIMUM_CAPACITY : n + 1;
677 }
678
679 /**
680 * Returns x's Class if it is of the form "class C implements
681 * Comparable<C>", else null.
682 */
683 static Class<?> comparableClassFor(Object x) {
684 if (x instanceof Comparable) {
685 Class<?> c; Type[] ts, as; Type t; ParameterizedType p;
686 if ((c = x.getClass()) == String.class) // bypass checks
687 return c;
688 if ((ts = c.getGenericInterfaces()) != null) {
689 for (int i = 0; i < ts.length; ++i) {
690 if (((t = ts[i]) instanceof ParameterizedType) &&
691 ((p = (ParameterizedType)t).getRawType() ==
692 Comparable.class) &&
693 (as = p.getActualTypeArguments()) != null &&
694 as.length == 1 && as[0] == c) // type arg is c
695 return c;
696 }
697 }
698 }
699 return null;
700 }
701
702 /**
703 * Returns k.compareTo(x) if x matches kc (k's screened comparable
704 * class), else 0.
705 */
706 @SuppressWarnings({"rawtypes","unchecked"}) // for cast to Comparable
707 static int compareComparables(Class<?> kc, Object k, Object x) {
708 return (x == null || x.getClass() != kc ? 0 :
709 ((Comparable)k).compareTo(x));
710 }
711
712 /* ---------------- Table element access -------------- */
713
714 /*
715 * Volatile access methods are used for table elements as well as
716 * elements of in-progress next table while resizing. All uses of
717 * the tab arguments must be null checked by callers. All callers
718 * also paranoically precheck that tab's length is not zero (or an
719 * equivalent check), thus ensuring that any index argument taking
720 * the form of a hash value anded with (length - 1) is a valid
721 * index. Note that, to be correct wrt arbitrary concurrency
722 * errors by users, these checks must operate on local variables,
723 * which accounts for some odd-looking inline assignments below.
724 * Note that calls to setTabAt always occur within locked regions,
725 * and so in principle require only release ordering, not need
726 * full volatile semantics, but are currently coded as volatile
727 * writes to be conservative.
728 */
729
730 @SuppressWarnings("unchecked")
731 static final <K,V> Node<K,V> tabAt(Node<K,V>[] tab, int i) {
732 return (Node<K,V>)U.getObjectVolatile(tab, ((long)i << ASHIFT) + ABASE);
733 }
734
735 static final <K,V> boolean casTabAt(Node<K,V>[] tab, int i,
736 Node<K,V> c, Node<K,V> v) {
737 return U.compareAndSwapObject(tab, ((long)i << ASHIFT) + ABASE, c, v);
738 }
739
740 static final <K,V> void setTabAt(Node<K,V>[] tab, int i, Node<K,V> v) {
741 U.putObjectVolatile(tab, ((long)i << ASHIFT) + ABASE, v);
742 }
743
744 /* ---------------- Fields -------------- */
745
746 /**
747 * The array of bins. Lazily initialized upon first insertion.
748 * Size is always a power of two. Accessed directly by iterators.
749 */
750 transient volatile Node<K,V>[] table;
751
752 /**
753 * The next table to use; non-null only while resizing.
754 */
755 private transient volatile Node<K,V>[] nextTable;
756
757 /**
758 * Base counter value, used mainly when there is no contention,
759 * but also as a fallback during table initialization
760 * races. Updated via CAS.
761 */
762 private transient volatile long baseCount;
763
764 /**
765 * Table initialization and resizing control. When negative, the
766 * table is being initialized or resized: -1 for initialization,
767 * else -(1 + the number of active resizing threads). Otherwise,
768 * when table is null, holds the initial table size to use upon
769 * creation, or 0 for default. After initialization, holds the
770 * next element count value upon which to resize the table.
771 */
772 private transient volatile int sizeCtl;
773
774 /**
775 * The next table index (plus one) to split while resizing.
776 */
777 private transient volatile int transferIndex;
778
779 /**
780 * The least available table index to split while resizing.
781 */
782 private transient volatile int transferOrigin;
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 ConcurrentHashMapV8() {
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 ConcurrentHashMapV8(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 ConcurrentHashMapV8(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 ConcurrentHashMapV8(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 ConcurrentHashMapV8(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,
1002 new Node<K,V>(hash, key, value, null)))
1003 break; // no lock when adding to empty bin
1004 }
1005 else if ((fh = f.hash) == MOVED)
1006 tab = helpTransfer(tab, f);
1007 else {
1008 V oldVal = null;
1009 synchronized (f) {
1010 if (tabAt(tab, i) == f) {
1011 if (fh >= 0) {
1012 binCount = 1;
1013 for (Node<K,V> e = f;; ++binCount) {
1014 K ek;
1015 if (e.hash == hash &&
1016 ((ek = e.key) == key ||
1017 (ek != null && key.equals(ek)))) {
1018 oldVal = e.val;
1019 if (!onlyIfAbsent)
1020 e.val = value;
1021 break;
1022 }
1023 Node<K,V> pred = e;
1024 if ((e = e.next) == null) {
1025 pred.next = new Node<K,V>(hash, key,
1026 value, null);
1027 break;
1028 }
1029 }
1030 }
1031 else if (f instanceof TreeBin) {
1032 Node<K,V> p;
1033 binCount = 2;
1034 if ((p = ((TreeBin<K,V>)f).putTreeVal(hash, key,
1035 value)) != null) {
1036 oldVal = p.val;
1037 if (!onlyIfAbsent)
1038 p.val = value;
1039 }
1040 }
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 }
1144 }
1145 if (validated) {
1146 if (oldVal != null) {
1147 if (value == null)
1148 addCount(-1L, -1);
1149 return oldVal;
1150 }
1151 break;
1152 }
1153 }
1154 }
1155 return null;
1156 }
1157
1158 /**
1159 * Removes all of the mappings from this map.
1160 */
1161 public void clear() {
1162 long delta = 0L; // negative number of deletions
1163 int i = 0;
1164 Node<K,V>[] tab = table;
1165 while (tab != null && i < tab.length) {
1166 int fh;
1167 Node<K,V> f = tabAt(tab, i);
1168 if (f == null)
1169 ++i;
1170 else if ((fh = f.hash) == MOVED) {
1171 tab = helpTransfer(tab, f);
1172 i = 0; // restart
1173 }
1174 else {
1175 synchronized (f) {
1176 if (tabAt(tab, i) == f) {
1177 Node<K,V> p = (fh >= 0 ? f :
1178 (f instanceof TreeBin) ?
1179 ((TreeBin<K,V>)f).first : null);
1180 while (p != null) {
1181 --delta;
1182 p = p.next;
1183 }
1184 setTabAt(tab, i++, null);
1185 }
1186 }
1187 }
1188 }
1189 if (delta != 0L)
1190 addCount(delta, -1);
1191 }
1192
1193 /**
1194 * Returns a {@link Set} view of the keys contained in this map.
1195 * The set is backed by the map, so changes to the map are
1196 * reflected in the set, and vice-versa. The set supports element
1197 * removal, which removes the corresponding mapping from this map,
1198 * via the {@code Iterator.remove}, {@code Set.remove},
1199 * {@code removeAll}, {@code retainAll}, and {@code clear}
1200 * operations. It does not support the {@code add} or
1201 * {@code addAll} operations.
1202 *
1203 * <p>The view's {@code iterator} is a "weakly consistent" iterator
1204 * that will never throw {@link ConcurrentModificationException},
1205 * and guarantees to traverse elements as they existed upon
1206 * construction of the iterator, and may (but is not guaranteed to)
1207 * reflect any modifications subsequent to construction.
1208 *
1209 * @return the set view
1210 */
1211 public KeySetView<K,V> keySet() {
1212 KeySetView<K,V> ks;
1213 return (ks = keySet) != null ? ks : (keySet = new KeySetView<K,V>(this, null));
1214 }
1215
1216 /**
1217 * Returns a {@link Collection} view of the values contained in this map.
1218 * The collection is backed by the map, so changes to the map are
1219 * reflected in the collection, and vice-versa. The collection
1220 * supports element removal, which removes the corresponding
1221 * mapping from this map, via the {@code Iterator.remove},
1222 * {@code Collection.remove}, {@code removeAll},
1223 * {@code retainAll}, and {@code clear} operations. It does not
1224 * support the {@code add} or {@code addAll} operations.
1225 *
1226 * <p>The view's {@code iterator} is a "weakly consistent" iterator
1227 * that will never throw {@link ConcurrentModificationException},
1228 * and guarantees to traverse elements as they existed upon
1229 * construction of the iterator, and may (but is not guaranteed to)
1230 * reflect any modifications subsequent to construction.
1231 *
1232 * @return the collection view
1233 */
1234 public Collection<V> values() {
1235 ValuesView<K,V> vs;
1236 return (vs = values) != null ? vs : (values = new ValuesView<K,V>(this));
1237 }
1238
1239 /**
1240 * Returns a {@link Set} view of the mappings contained in this map.
1241 * The set is backed by the map, so changes to the map are
1242 * reflected in the set, and vice-versa. The set supports element
1243 * removal, which removes the corresponding mapping from the map,
1244 * via the {@code Iterator.remove}, {@code Set.remove},
1245 * {@code removeAll}, {@code retainAll}, and {@code clear}
1246 * operations.
1247 *
1248 * <p>The view's {@code iterator} is a "weakly consistent" iterator
1249 * that will never throw {@link ConcurrentModificationException},
1250 * and guarantees to traverse elements as they existed upon
1251 * construction of the iterator, and may (but is not guaranteed to)
1252 * reflect any modifications subsequent to construction.
1253 *
1254 * @return the set view
1255 */
1256 public Set<Map.Entry<K,V>> entrySet() {
1257 EntrySetView<K,V> es;
1258 return (es = entrySet) != null ? es : (entrySet = new EntrySetView<K,V>(this));
1259 }
1260
1261 /**
1262 * Returns the hash code value for this {@link Map}, i.e.,
1263 * the sum of, for each key-value pair in the map,
1264 * {@code key.hashCode() ^ value.hashCode()}.
1265 *
1266 * @return the hash code value for this map
1267 */
1268 public int hashCode() {
1269 int h = 0;
1270 Node<K,V>[] t;
1271 if ((t = table) != null) {
1272 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1273 for (Node<K,V> p; (p = it.advance()) != null; )
1274 h += p.key.hashCode() ^ p.val.hashCode();
1275 }
1276 return h;
1277 }
1278
1279 /**
1280 * Returns a string representation of this map. The string
1281 * representation consists of a list of key-value mappings (in no
1282 * particular order) enclosed in braces ("{@code {}}"). Adjacent
1283 * mappings are separated by the characters {@code ", "} (comma
1284 * and space). Each key-value mapping is rendered as the key
1285 * followed by an equals sign ("{@code =}") followed by the
1286 * associated value.
1287 *
1288 * @return a string representation of this map
1289 */
1290 public String toString() {
1291 Node<K,V>[] t;
1292 int f = (t = table) == null ? 0 : t.length;
1293 Traverser<K,V> it = new Traverser<K,V>(t, f, 0, f);
1294 StringBuilder sb = new StringBuilder();
1295 sb.append('{');
1296 Node<K,V> p;
1297 if ((p = it.advance()) != null) {
1298 for (;;) {
1299 K k = p.key;
1300 V v = p.val;
1301 sb.append(k == this ? "(this Map)" : k);
1302 sb.append('=');
1303 sb.append(v == this ? "(this Map)" : v);
1304 if ((p = it.advance()) == null)
1305 break;
1306 sb.append(',').append(' ');
1307 }
1308 }
1309 return sb.append('}').toString();
1310 }
1311
1312 /**
1313 * Compares the specified object with this map for equality.
1314 * Returns {@code true} if the given object is a map with the same
1315 * mappings as this map. This operation may return misleading
1316 * results if either map is concurrently modified during execution
1317 * of this method.
1318 *
1319 * @param o object to be compared for equality with this map
1320 * @return {@code true} if the specified object is equal to this map
1321 */
1322 public boolean equals(Object o) {
1323 if (o != this) {
1324 if (!(o instanceof Map))
1325 return false;
1326 Map<?,?> m = (Map<?,?>) o;
1327 Node<K,V>[] t;
1328 int f = (t = table) == null ? 0 : t.length;
1329 Traverser<K,V> it = new Traverser<K,V>(t, f, 0, f);
1330 for (Node<K,V> p; (p = it.advance()) != null; ) {
1331 V val = p.val;
1332 Object v = m.get(p.key);
1333 if (v == null || (v != val && !v.equals(val)))
1334 return false;
1335 }
1336 for (Map.Entry<?,?> e : m.entrySet()) {
1337 Object mk, mv, v;
1338 if ((mk = e.getKey()) == null ||
1339 (mv = e.getValue()) == null ||
1340 (v = get(mk)) == null ||
1341 (mv != v && !mv.equals(v)))
1342 return false;
1343 }
1344 }
1345 return true;
1346 }
1347
1348 /**
1349 * Stripped-down version of helper class used in previous version,
1350 * declared for the sake of serialization compatibility
1351 */
1352 static class Segment<K,V> extends ReentrantLock implements Serializable {
1353 private static final long serialVersionUID = 2249069246763182397L;
1354 final float loadFactor;
1355 Segment(float lf) { this.loadFactor = lf; }
1356 }
1357
1358 /**
1359 * Saves the state of the {@code ConcurrentHashMapV8} instance to a
1360 * stream (i.e., serializes it).
1361 * @param s the stream
1362 * @serialData
1363 * the key (Object) and value (Object)
1364 * for each key-value mapping, followed by a null pair.
1365 * The key-value mappings are emitted in no particular order.
1366 */
1367 private void writeObject(java.io.ObjectOutputStream s)
1368 throws java.io.IOException {
1369 // For serialization compatibility
1370 // Emulate segment calculation from previous version of this class
1371 int sshift = 0;
1372 int ssize = 1;
1373 while (ssize < DEFAULT_CONCURRENCY_LEVEL) {
1374 ++sshift;
1375 ssize <<= 1;
1376 }
1377 int segmentShift = 32 - sshift;
1378 int segmentMask = ssize - 1;
1379 @SuppressWarnings("unchecked") Segment<K,V>[] segments = (Segment<K,V>[])
1380 new Segment<?,?>[DEFAULT_CONCURRENCY_LEVEL];
1381 for (int i = 0; i < segments.length; ++i)
1382 segments[i] = new Segment<K,V>(LOAD_FACTOR);
1383 s.putFields().put("segments", segments);
1384 s.putFields().put("segmentShift", segmentShift);
1385 s.putFields().put("segmentMask", segmentMask);
1386 s.writeFields();
1387
1388 Node<K,V>[] t;
1389 if ((t = table) != null) {
1390 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1391 for (Node<K,V> p; (p = it.advance()) != null; ) {
1392 s.writeObject(p.key);
1393 s.writeObject(p.val);
1394 }
1395 }
1396 s.writeObject(null);
1397 s.writeObject(null);
1398 segments = null; // throw away
1399 }
1400
1401 /**
1402 * Reconstitutes the instance from a stream (that is, deserializes it).
1403 * @param s the stream
1404 */
1405 private void readObject(java.io.ObjectInputStream s)
1406 throws java.io.IOException, ClassNotFoundException {
1407 /*
1408 * To improve performance in typical cases, we create nodes
1409 * while reading, then place in table once size is known.
1410 * However, we must also validate uniqueness and deal with
1411 * overpopulated bins while doing so, which requires
1412 * specialized versions of putVal mechanics.
1413 */
1414 sizeCtl = -1; // force exclusion for table construction
1415 s.defaultReadObject();
1416 long size = 0L;
1417 Node<K,V> p = null;
1418 for (;;) {
1419 @SuppressWarnings("unchecked") K k = (K) s.readObject();
1420 @SuppressWarnings("unchecked") V v = (V) s.readObject();
1421 if (k != null && v != null) {
1422 p = new Node<K,V>(spread(k.hashCode()), k, v, p);
1423 ++size;
1424 }
1425 else
1426 break;
1427 }
1428 if (size == 0L)
1429 sizeCtl = 0;
1430 else {
1431 int n;
1432 if (size >= (long)(MAXIMUM_CAPACITY >>> 1))
1433 n = MAXIMUM_CAPACITY;
1434 else {
1435 int sz = (int)size;
1436 n = tableSizeFor(sz + (sz >>> 1) + 1);
1437 }
1438 @SuppressWarnings({"rawtypes","unchecked"})
1439 Node<K,V>[] tab = (Node<K,V>[])new Node[n];
1440 int mask = n - 1;
1441 long added = 0L;
1442 while (p != null) {
1443 boolean insertAtFront;
1444 Node<K,V> next = p.next, first;
1445 int h = p.hash, j = h & mask;
1446 if ((first = tabAt(tab, j)) == null)
1447 insertAtFront = true;
1448 else {
1449 K k = p.key;
1450 if (first.hash < 0) {
1451 TreeBin<K,V> t = (TreeBin<K,V>)first;
1452 if (t.putTreeVal(h, k, p.val) == null)
1453 ++added;
1454 insertAtFront = false;
1455 }
1456 else {
1457 int binCount = 0;
1458 insertAtFront = true;
1459 Node<K,V> q; K qk;
1460 for (q = first; q != null; q = q.next) {
1461 if (q.hash == h &&
1462 ((qk = q.key) == k ||
1463 (qk != null && k.equals(qk)))) {
1464 insertAtFront = false;
1465 break;
1466 }
1467 ++binCount;
1468 }
1469 if (insertAtFront && binCount >= TREEIFY_THRESHOLD) {
1470 insertAtFront = false;
1471 ++added;
1472 p.next = first;
1473 TreeNode<K,V> hd = null, tl = null;
1474 for (q = p; q != null; q = q.next) {
1475 TreeNode<K,V> t = new TreeNode<K,V>
1476 (q.hash, q.key, q.val, null, null);
1477 if ((t.prev = tl) == null)
1478 hd = t;
1479 else
1480 tl.next = t;
1481 tl = t;
1482 }
1483 setTabAt(tab, j, new TreeBin<K,V>(hd));
1484 }
1485 }
1486 }
1487 if (insertAtFront) {
1488 ++added;
1489 p.next = first;
1490 setTabAt(tab, j, p);
1491 }
1492 p = next;
1493 }
1494 table = tab;
1495 sizeCtl = n - (n >>> 2);
1496 baseCount = added;
1497 }
1498 }
1499
1500 // ConcurrentMap methods
1501
1502 /**
1503 * {@inheritDoc}
1504 *
1505 * @return the previous value associated with the specified key,
1506 * or {@code null} if there was no mapping for the key
1507 * @throws NullPointerException if the specified key or value is null
1508 */
1509 public V putIfAbsent(K key, V value) {
1510 return putVal(key, value, true);
1511 }
1512
1513 /**
1514 * {@inheritDoc}
1515 *
1516 * @throws NullPointerException if the specified key is null
1517 */
1518 public boolean remove(Object key, Object value) {
1519 if (key == null)
1520 throw new NullPointerException();
1521 return value != null && replaceNode(key, null, value) != null;
1522 }
1523
1524 /**
1525 * {@inheritDoc}
1526 *
1527 * @throws NullPointerException if any of the arguments are null
1528 */
1529 public boolean replace(K key, V oldValue, V newValue) {
1530 if (key == null || oldValue == null || newValue == null)
1531 throw new NullPointerException();
1532 return replaceNode(key, newValue, oldValue) != null;
1533 }
1534
1535 /**
1536 * {@inheritDoc}
1537 *
1538 * @return the previous value associated with the specified key,
1539 * or {@code null} if there was no mapping for the key
1540 * @throws NullPointerException if the specified key or value is null
1541 */
1542 public V replace(K key, V value) {
1543 if (key == null || value == null)
1544 throw new NullPointerException();
1545 return replaceNode(key, value, null);
1546 }
1547
1548 // Overrides of JDK8+ Map extension method defaults
1549
1550 /**
1551 * Returns the value to which the specified key is mapped, or the
1552 * given default value if this map contains no mapping for the
1553 * key.
1554 *
1555 * @param key the key whose associated value is to be returned
1556 * @param defaultValue the value to return if this map contains
1557 * no mapping for the given key
1558 * @return the mapping for the key, if present; else the default value
1559 * @throws NullPointerException if the specified key is null
1560 */
1561 public V getOrDefault(Object key, V defaultValue) {
1562 V v;
1563 return (v = get(key)) == null ? defaultValue : v;
1564 }
1565
1566 public void forEach(BiAction<? super K, ? super V> action) {
1567 if (action == null) throw new NullPointerException();
1568 Node<K,V>[] t;
1569 if ((t = table) != null) {
1570 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1571 for (Node<K,V> p; (p = it.advance()) != null; ) {
1572 action.apply(p.key, p.val);
1573 }
1574 }
1575 }
1576
1577 public void replaceAll(BiFun<? super K, ? super V, ? extends V> function) {
1578 if (function == null) throw new NullPointerException();
1579 Node<K,V>[] t;
1580 if ((t = table) != null) {
1581 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
1582 for (Node<K,V> p; (p = it.advance()) != null; ) {
1583 V oldValue = p.val;
1584 for (K key = p.key;;) {
1585 V newValue = function.apply(key, oldValue);
1586 if (newValue == null)
1587 throw new NullPointerException();
1588 if (replaceNode(key, newValue, oldValue) != null ||
1589 (oldValue = get(key)) == null)
1590 break;
1591 }
1592 }
1593 }
1594 }
1595
1596 /**
1597 * If the specified key is not already associated with a value,
1598 * attempts to compute its value using the given mapping function
1599 * and enters it into this map unless {@code null}. The entire
1600 * method invocation is performed atomically, so the function is
1601 * applied at most once per key. Some attempted update operations
1602 * on this map by other threads may be blocked while computation
1603 * is in progress, so the computation should be short and simple,
1604 * and must not attempt to update any other mappings of this map.
1605 *
1606 * @param key key with which the specified value is to be associated
1607 * @param mappingFunction the function to compute a value
1608 * @return the current (existing or computed) value associated with
1609 * the specified key, or null if the computed value is null
1610 * @throws NullPointerException if the specified key or mappingFunction
1611 * is null
1612 * @throws IllegalStateException if the computation detectably
1613 * attempts a recursive update to this map that would
1614 * otherwise never complete
1615 * @throws RuntimeException or Error if the mappingFunction does so,
1616 * in which case the mapping is left unestablished
1617 */
1618 public V computeIfAbsent(K key, Fun<? super K, ? extends V> mappingFunction) {
1619 if (key == null || mappingFunction == null)
1620 throw new NullPointerException();
1621 int h = spread(key.hashCode());
1622 V val = null;
1623 int binCount = 0;
1624 for (Node<K,V>[] tab = table;;) {
1625 Node<K,V> f; int n, i, fh;
1626 if (tab == null || (n = tab.length) == 0)
1627 tab = initTable();
1628 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
1629 Node<K,V> r = new ReservationNode<K,V>();
1630 synchronized (r) {
1631 if (casTabAt(tab, i, null, r)) {
1632 binCount = 1;
1633 Node<K,V> node = null;
1634 try {
1635 if ((val = mappingFunction.apply(key)) != null)
1636 node = new Node<K,V>(h, key, val, null);
1637 } finally {
1638 setTabAt(tab, i, node);
1639 }
1640 }
1641 }
1642 if (binCount != 0)
1643 break;
1644 }
1645 else if ((fh = f.hash) == MOVED)
1646 tab = helpTransfer(tab, f);
1647 else {
1648 boolean added = false;
1649 synchronized (f) {
1650 if (tabAt(tab, i) == f) {
1651 if (fh >= 0) {
1652 binCount = 1;
1653 for (Node<K,V> e = f;; ++binCount) {
1654 K ek; V ev;
1655 if (e.hash == h &&
1656 ((ek = e.key) == key ||
1657 (ek != null && key.equals(ek)))) {
1658 val = e.val;
1659 break;
1660 }
1661 Node<K,V> pred = e;
1662 if ((e = e.next) == null) {
1663 if ((val = mappingFunction.apply(key)) != null) {
1664 added = true;
1665 pred.next = new Node<K,V>(h, key, val, null);
1666 }
1667 break;
1668 }
1669 }
1670 }
1671 else if (f instanceof TreeBin) {
1672 binCount = 2;
1673 TreeBin<K,V> t = (TreeBin<K,V>)f;
1674 TreeNode<K,V> r, p;
1675 if ((r = t.root) != null &&
1676 (p = r.findTreeNode(h, key, null)) != null)
1677 val = p.val;
1678 else if ((val = mappingFunction.apply(key)) != null) {
1679 added = true;
1680 t.putTreeVal(h, key, val);
1681 }
1682 }
1683 }
1684 }
1685 if (binCount != 0) {
1686 if (binCount >= TREEIFY_THRESHOLD)
1687 treeifyBin(tab, i);
1688 if (!added)
1689 return val;
1690 break;
1691 }
1692 }
1693 }
1694 if (val != null)
1695 addCount(1L, binCount);
1696 return val;
1697 }
1698
1699 /**
1700 * If the value for the specified key is present, attempts to
1701 * compute a new mapping given the key and its current mapped
1702 * value. The entire method invocation is performed atomically.
1703 * Some attempted update operations on this map by other threads
1704 * may be blocked while computation is in progress, so the
1705 * computation should be short and simple, and must not attempt to
1706 * update any other mappings of this map.
1707 *
1708 * @param key key with which a value may be associated
1709 * @param remappingFunction the function to compute a value
1710 * @return the new value associated with the specified key, or null if none
1711 * @throws NullPointerException if the specified key or remappingFunction
1712 * is null
1713 * @throws IllegalStateException if the computation detectably
1714 * attempts a recursive update to this map that would
1715 * otherwise never complete
1716 * @throws RuntimeException or Error if the remappingFunction does so,
1717 * in which case the mapping is unchanged
1718 */
1719 public V computeIfPresent(K key, BiFun<? super K, ? super V, ? extends V> remappingFunction) {
1720 if (key == null || remappingFunction == null)
1721 throw new NullPointerException();
1722 int h = spread(key.hashCode());
1723 V val = null;
1724 int delta = 0;
1725 int binCount = 0;
1726 for (Node<K,V>[] tab = table;;) {
1727 Node<K,V> f; int n, i, fh;
1728 if (tab == null || (n = tab.length) == 0)
1729 tab = initTable();
1730 else if ((f = tabAt(tab, i = (n - 1) & h)) == null)
1731 break;
1732 else if ((fh = f.hash) == MOVED)
1733 tab = helpTransfer(tab, f);
1734 else {
1735 synchronized (f) {
1736 if (tabAt(tab, i) == f) {
1737 if (fh >= 0) {
1738 binCount = 1;
1739 for (Node<K,V> e = f, pred = null;; ++binCount) {
1740 K ek;
1741 if (e.hash == h &&
1742 ((ek = e.key) == key ||
1743 (ek != null && key.equals(ek)))) {
1744 val = remappingFunction.apply(key, e.val);
1745 if (val != null)
1746 e.val = val;
1747 else {
1748 delta = -1;
1749 Node<K,V> en = e.next;
1750 if (pred != null)
1751 pred.next = en;
1752 else
1753 setTabAt(tab, i, en);
1754 }
1755 break;
1756 }
1757 pred = e;
1758 if ((e = e.next) == null)
1759 break;
1760 }
1761 }
1762 else if (f instanceof TreeBin) {
1763 binCount = 2;
1764 TreeBin<K,V> t = (TreeBin<K,V>)f;
1765 TreeNode<K,V> r, p;
1766 if ((r = t.root) != null &&
1767 (p = r.findTreeNode(h, key, null)) != null) {
1768 val = remappingFunction.apply(key, p.val);
1769 if (val != null)
1770 p.val = val;
1771 else {
1772 delta = -1;
1773 if (t.removeTreeNode(p))
1774 setTabAt(tab, i, untreeify(t.first));
1775 }
1776 }
1777 }
1778 }
1779 }
1780 if (binCount != 0)
1781 break;
1782 }
1783 }
1784 if (delta != 0)
1785 addCount((long)delta, binCount);
1786 return val;
1787 }
1788
1789 /**
1790 * Attempts to compute a mapping for the specified key and its
1791 * current mapped value (or {@code null} if there is no current
1792 * mapping). The entire method invocation is performed atomically.
1793 * Some attempted update operations on this map by other threads
1794 * may be blocked while computation is in progress, so the
1795 * computation should be short and simple, and must not attempt to
1796 * update any other mappings of this Map.
1797 *
1798 * @param key key with which the specified value is to be associated
1799 * @param remappingFunction the function to compute a value
1800 * @return the new value associated with the specified key, or null if none
1801 * @throws NullPointerException if the specified key or remappingFunction
1802 * is null
1803 * @throws IllegalStateException if the computation detectably
1804 * attempts a recursive update to this map that would
1805 * otherwise never complete
1806 * @throws RuntimeException or Error if the remappingFunction does so,
1807 * in which case the mapping is unchanged
1808 */
1809 public V compute(K key,
1810 BiFun<? super K, ? super V, ? extends V> remappingFunction) {
1811 if (key == null || remappingFunction == null)
1812 throw new NullPointerException();
1813 int h = spread(key.hashCode());
1814 V val = null;
1815 int delta = 0;
1816 int binCount = 0;
1817 for (Node<K,V>[] tab = table;;) {
1818 Node<K,V> f; int n, i, fh;
1819 if (tab == null || (n = tab.length) == 0)
1820 tab = initTable();
1821 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
1822 Node<K,V> r = new ReservationNode<K,V>();
1823 synchronized (r) {
1824 if (casTabAt(tab, i, null, r)) {
1825 binCount = 1;
1826 Node<K,V> node = null;
1827 try {
1828 if ((val = remappingFunction.apply(key, null)) != null) {
1829 delta = 1;
1830 node = new Node<K,V>(h, key, val, null);
1831 }
1832 } finally {
1833 setTabAt(tab, i, node);
1834 }
1835 }
1836 }
1837 if (binCount != 0)
1838 break;
1839 }
1840 else if ((fh = f.hash) == MOVED)
1841 tab = helpTransfer(tab, f);
1842 else {
1843 synchronized (f) {
1844 if (tabAt(tab, i) == f) {
1845 if (fh >= 0) {
1846 binCount = 1;
1847 for (Node<K,V> e = f, pred = null;; ++binCount) {
1848 K ek;
1849 if (e.hash == h &&
1850 ((ek = e.key) == key ||
1851 (ek != null && key.equals(ek)))) {
1852 val = remappingFunction.apply(key, e.val);
1853 if (val != null)
1854 e.val = val;
1855 else {
1856 delta = -1;
1857 Node<K,V> en = e.next;
1858 if (pred != null)
1859 pred.next = en;
1860 else
1861 setTabAt(tab, i, en);
1862 }
1863 break;
1864 }
1865 pred = e;
1866 if ((e = e.next) == null) {
1867 val = remappingFunction.apply(key, null);
1868 if (val != null) {
1869 delta = 1;
1870 pred.next =
1871 new Node<K,V>(h, key, val, null);
1872 }
1873 break;
1874 }
1875 }
1876 }
1877 else if (f instanceof TreeBin) {
1878 binCount = 1;
1879 TreeBin<K,V> t = (TreeBin<K,V>)f;
1880 TreeNode<K,V> r, p;
1881 if ((r = t.root) != null)
1882 p = r.findTreeNode(h, key, null);
1883 else
1884 p = null;
1885 V pv = (p == null) ? null : p.val;
1886 val = remappingFunction.apply(key, pv);
1887 if (val != null) {
1888 if (p != null)
1889 p.val = val;
1890 else {
1891 delta = 1;
1892 t.putTreeVal(h, key, val);
1893 }
1894 }
1895 else if (p != null) {
1896 delta = -1;
1897 if (t.removeTreeNode(p))
1898 setTabAt(tab, i, untreeify(t.first));
1899 }
1900 }
1901 }
1902 }
1903 if (binCount != 0) {
1904 if (binCount >= TREEIFY_THRESHOLD)
1905 treeifyBin(tab, i);
1906 break;
1907 }
1908 }
1909 }
1910 if (delta != 0)
1911 addCount((long)delta, binCount);
1912 return val;
1913 }
1914
1915 /**
1916 * If the specified key is not already associated with a
1917 * (non-null) value, associates it with the given value.
1918 * Otherwise, replaces the value with the results of the given
1919 * remapping function, or removes if {@code null}. The entire
1920 * method invocation is performed atomically. Some attempted
1921 * update operations on this map by other threads may be blocked
1922 * while computation is in progress, so the computation should be
1923 * short and simple, and must not attempt to update any other
1924 * mappings of this Map.
1925 *
1926 * @param key key with which the specified value is to be associated
1927 * @param value the value to use if absent
1928 * @param remappingFunction the function to recompute a value if present
1929 * @return the new value associated with the specified key, or null if none
1930 * @throws NullPointerException if the specified key or the
1931 * remappingFunction is null
1932 * @throws RuntimeException or Error if the remappingFunction does so,
1933 * in which case the mapping is unchanged
1934 */
1935 public V merge(K key, V value, BiFun<? super V, ? super V, ? extends V> remappingFunction) {
1936 if (key == null || value == null || remappingFunction == null)
1937 throw new NullPointerException();
1938 int h = spread(key.hashCode());
1939 V val = null;
1940 int delta = 0;
1941 int binCount = 0;
1942 for (Node<K,V>[] tab = table;;) {
1943 Node<K,V> f; int n, i, fh;
1944 if (tab == null || (n = tab.length) == 0)
1945 tab = initTable();
1946 else if ((f = tabAt(tab, i = (n - 1) & h)) == null) {
1947 if (casTabAt(tab, i, null, new Node<K,V>(h, key, value, null))) {
1948 delta = 1;
1949 val = value;
1950 break;
1951 }
1952 }
1953 else if ((fh = f.hash) == MOVED)
1954 tab = helpTransfer(tab, f);
1955 else {
1956 synchronized (f) {
1957 if (tabAt(tab, i) == f) {
1958 if (fh >= 0) {
1959 binCount = 1;
1960 for (Node<K,V> e = f, pred = null;; ++binCount) {
1961 K ek;
1962 if (e.hash == h &&
1963 ((ek = e.key) == key ||
1964 (ek != null && key.equals(ek)))) {
1965 val = remappingFunction.apply(e.val, value);
1966 if (val != null)
1967 e.val = val;
1968 else {
1969 delta = -1;
1970 Node<K,V> en = e.next;
1971 if (pred != null)
1972 pred.next = en;
1973 else
1974 setTabAt(tab, i, en);
1975 }
1976 break;
1977 }
1978 pred = e;
1979 if ((e = e.next) == null) {
1980 delta = 1;
1981 val = value;
1982 pred.next =
1983 new Node<K,V>(h, key, val, null);
1984 break;
1985 }
1986 }
1987 }
1988 else if (f instanceof TreeBin) {
1989 binCount = 2;
1990 TreeBin<K,V> t = (TreeBin<K,V>)f;
1991 TreeNode<K,V> r = t.root;
1992 TreeNode<K,V> p = (r == null) ? null :
1993 r.findTreeNode(h, key, null);
1994 val = (p == null) ? value :
1995 remappingFunction.apply(p.val, value);
1996 if (val != null) {
1997 if (p != null)
1998 p.val = val;
1999 else {
2000 delta = 1;
2001 t.putTreeVal(h, key, val);
2002 }
2003 }
2004 else if (p != null) {
2005 delta = -1;
2006 if (t.removeTreeNode(p))
2007 setTabAt(tab, i, untreeify(t.first));
2008 }
2009 }
2010 }
2011 }
2012 if (binCount != 0) {
2013 if (binCount >= TREEIFY_THRESHOLD)
2014 treeifyBin(tab, i);
2015 break;
2016 }
2017 }
2018 }
2019 if (delta != 0)
2020 addCount((long)delta, binCount);
2021 return val;
2022 }
2023
2024 // Hashtable legacy methods
2025
2026 /**
2027 * Legacy method testing if some key maps into the specified value
2028 * in this table. This method is identical in functionality to
2029 * {@link #containsValue(Object)}, and exists solely to ensure
2030 * full compatibility with class {@link java.util.Hashtable},
2031 * which supported this method prior to introduction of the
2032 * Java Collections framework.
2033 *
2034 * @param value a value to search for
2035 * @return {@code true} if and only if some key maps to the
2036 * {@code value} argument in this table as
2037 * determined by the {@code equals} method;
2038 * {@code false} otherwise
2039 * @throws NullPointerException if the specified value is null
2040 */
2041 @Deprecated public boolean contains(Object value) {
2042 return containsValue(value);
2043 }
2044
2045 /**
2046 * Returns an enumeration of the keys in this table.
2047 *
2048 * @return an enumeration of the keys in this table
2049 * @see #keySet()
2050 */
2051 public Enumeration<K> keys() {
2052 Node<K,V>[] t;
2053 int f = (t = table) == null ? 0 : t.length;
2054 return new KeyIterator<K,V>(t, f, 0, f, this);
2055 }
2056
2057 /**
2058 * Returns an enumeration of the values in this table.
2059 *
2060 * @return an enumeration of the values in this table
2061 * @see #values()
2062 */
2063 public Enumeration<V> elements() {
2064 Node<K,V>[] t;
2065 int f = (t = table) == null ? 0 : t.length;
2066 return new ValueIterator<K,V>(t, f, 0, f, this);
2067 }
2068
2069 // ConcurrentHashMapV8-only methods
2070
2071 /**
2072 * Returns the number of mappings. This method should be used
2073 * instead of {@link #size} because a ConcurrentHashMapV8 may
2074 * contain more mappings than can be represented as an int. The
2075 * value returned is an estimate; the actual count may differ if
2076 * there are concurrent insertions or removals.
2077 *
2078 * @return the number of mappings
2079 * @since 1.8
2080 */
2081 public long mappingCount() {
2082 long n = sumCount();
2083 return (n < 0L) ? 0L : n; // ignore transient negative values
2084 }
2085
2086 /**
2087 * Creates a new {@link Set} backed by a ConcurrentHashMapV8
2088 * from the given type to {@code Boolean.TRUE}.
2089 *
2090 * @return the new set
2091 * @since 1.8
2092 */
2093 public static <K> KeySetView<K,Boolean> newKeySet() {
2094 return new KeySetView<K,Boolean>
2095 (new ConcurrentHashMapV8<K,Boolean>(), Boolean.TRUE);
2096 }
2097
2098 /**
2099 * Creates a new {@link Set} backed by a ConcurrentHashMapV8
2100 * from the given type to {@code Boolean.TRUE}.
2101 *
2102 * @param initialCapacity The implementation performs internal
2103 * sizing to accommodate this many elements.
2104 * @throws IllegalArgumentException if the initial capacity of
2105 * elements is negative
2106 * @return the new set
2107 * @since 1.8
2108 */
2109 public static <K> KeySetView<K,Boolean> newKeySet(int initialCapacity) {
2110 return new KeySetView<K,Boolean>
2111 (new ConcurrentHashMapV8<K,Boolean>(initialCapacity), Boolean.TRUE);
2112 }
2113
2114 /**
2115 * Returns a {@link Set} view of the keys in this map, using the
2116 * given common mapped value for any additions (i.e., {@link
2117 * Collection#add} and {@link Collection#addAll(Collection)}).
2118 * This is of course only appropriate if it is acceptable to use
2119 * the same value for all additions from this view.
2120 *
2121 * @param mappedValue the mapped value to use for any additions
2122 * @return the set view
2123 * @throws NullPointerException if the mappedValue is null
2124 */
2125 public KeySetView<K,V> keySet(V mappedValue) {
2126 if (mappedValue == null)
2127 throw new NullPointerException();
2128 return new KeySetView<K,V>(this, mappedValue);
2129 }
2130
2131 /* ---------------- Special Nodes -------------- */
2132
2133 /**
2134 * A node inserted at head of bins during transfer operations.
2135 */
2136 static final class ForwardingNode<K,V> extends Node<K,V> {
2137 final Node<K,V>[] nextTable;
2138 ForwardingNode(Node<K,V>[] tab) {
2139 super(MOVED, null, null, null);
2140 this.nextTable = tab;
2141 }
2142
2143 Node<K,V> find(int h, Object k) {
2144 // loop to avoid arbitrarily deep recursion on forwarding nodes
2145 outer: for (Node<K,V>[] tab = nextTable;;) {
2146 Node<K,V> e; int n;
2147 if (k == null || tab == null || (n = tab.length) == 0 ||
2148 (e = tabAt(tab, (n - 1) & h)) == null)
2149 return null;
2150 for (;;) {
2151 int eh; K ek;
2152 if ((eh = e.hash) == h &&
2153 ((ek = e.key) == k || (ek != null && k.equals(ek))))
2154 return e;
2155 if (eh < 0) {
2156 if (e instanceof ForwardingNode) {
2157 tab = ((ForwardingNode<K,V>)e).nextTable;
2158 continue outer;
2159 }
2160 else
2161 return e.find(h, k);
2162 }
2163 if ((e = e.next) == null)
2164 return null;
2165 }
2166 }
2167 }
2168 }
2169
2170 /**
2171 * A place-holder node used in computeIfAbsent and compute
2172 */
2173 static final class ReservationNode<K,V> extends Node<K,V> {
2174 ReservationNode() {
2175 super(RESERVED, null, null, null);
2176 }
2177
2178 Node<K,V> find(int h, Object k) {
2179 return null;
2180 }
2181 }
2182
2183 /* ---------------- Table Initialization and Resizing -------------- */
2184
2185 /**
2186 * Initializes table, using the size recorded in sizeCtl.
2187 */
2188 private final Node<K,V>[] initTable() {
2189 Node<K,V>[] tab; int sc;
2190 while ((tab = table) == null || tab.length == 0) {
2191 if ((sc = sizeCtl) < 0)
2192 Thread.yield(); // lost initialization race; just spin
2193 else if (U.compareAndSwapInt(this, SIZECTL, sc, -1)) {
2194 try {
2195 if ((tab = table) == null || tab.length == 0) {
2196 int n = (sc > 0) ? sc : DEFAULT_CAPACITY;
2197 @SuppressWarnings({"rawtypes","unchecked"})
2198 Node<K,V>[] nt = (Node<K,V>[])new Node[n];
2199 table = tab = nt;
2200 sc = n - (n >>> 2);
2201 }
2202 } finally {
2203 sizeCtl = sc;
2204 }
2205 break;
2206 }
2207 }
2208 return tab;
2209 }
2210
2211 /**
2212 * Adds to count, and if table is too small and not already
2213 * resizing, initiates transfer. If already resizing, helps
2214 * perform transfer if work is available. Rechecks occupancy
2215 * after a transfer to see if another resize is already needed
2216 * because resizings are lagging additions.
2217 *
2218 * @param x the count to add
2219 * @param check if <0, don't check resize, if <= 1 only check if uncontended
2220 */
2221 private final void addCount(long x, int check) {
2222 CounterCell[] as; long b, s;
2223 if ((as = counterCells) != null ||
2224 !U.compareAndSwapLong(this, BASECOUNT, b = baseCount, s = b + x)) {
2225 CounterHashCode hc; CounterCell a; long v; int m;
2226 boolean uncontended = true;
2227 if ((hc = threadCounterHashCode.get()) == null ||
2228 as == null || (m = as.length - 1) < 0 ||
2229 (a = as[m & hc.code]) == null ||
2230 !(uncontended =
2231 U.compareAndSwapLong(a, CELLVALUE, v = a.value, v + x))) {
2232 fullAddCount(x, hc, uncontended);
2233 return;
2234 }
2235 if (check <= 1)
2236 return;
2237 s = sumCount();
2238 }
2239 if (check >= 0) {
2240 Node<K,V>[] tab, nt; int sc;
2241 while (s >= (long)(sc = sizeCtl) && (tab = table) != null &&
2242 tab.length < MAXIMUM_CAPACITY) {
2243 if (sc < 0) {
2244 if (sc == -1 || transferIndex <= transferOrigin ||
2245 (nt = nextTable) == null)
2246 break;
2247 if (U.compareAndSwapInt(this, SIZECTL, sc, sc - 1))
2248 transfer(tab, nt);
2249 }
2250 else if (U.compareAndSwapInt(this, SIZECTL, sc, -2))
2251 transfer(tab, null);
2252 s = sumCount();
2253 }
2254 }
2255 }
2256
2257 /**
2258 * Helps transfer if a resize is in progress.
2259 */
2260 final Node<K,V>[] helpTransfer(Node<K,V>[] tab, Node<K,V> f) {
2261 Node<K,V>[] nextTab; int sc;
2262 if ((f instanceof ForwardingNode) &&
2263 (nextTab = ((ForwardingNode<K,V>)f).nextTable) != null) {
2264 if (nextTab == nextTable && tab == table &&
2265 transferIndex > transferOrigin && (sc = sizeCtl) < -1 &&
2266 U.compareAndSwapInt(this, SIZECTL, sc, sc - 1))
2267 transfer(tab, nextTab);
2268 return nextTab;
2269 }
2270 return table;
2271 }
2272
2273 /**
2274 * Tries to presize table to accommodate the given number of elements.
2275 *
2276 * @param size number of elements (doesn't need to be perfectly accurate)
2277 */
2278 private final void tryPresize(int size) {
2279 int c = (size >= (MAXIMUM_CAPACITY >>> 1)) ? MAXIMUM_CAPACITY :
2280 tableSizeFor(size + (size >>> 1) + 1);
2281 int sc;
2282 while ((sc = sizeCtl) >= 0) {
2283 Node<K,V>[] tab = table; int n;
2284 if (tab == null || (n = tab.length) == 0) {
2285 n = (sc > c) ? sc : c;
2286 if (U.compareAndSwapInt(this, SIZECTL, sc, -1)) {
2287 try {
2288 if (table == tab) {
2289 @SuppressWarnings({"rawtypes","unchecked"})
2290 Node<K,V>[] nt = (Node<K,V>[])new Node[n];
2291 table = nt;
2292 sc = n - (n >>> 2);
2293 }
2294 } finally {
2295 sizeCtl = sc;
2296 }
2297 }
2298 }
2299 else if (c <= sc || n >= MAXIMUM_CAPACITY)
2300 break;
2301 else if (tab == table &&
2302 U.compareAndSwapInt(this, SIZECTL, sc, -2))
2303 transfer(tab, null);
2304 }
2305 }
2306
2307 /**
2308 * Moves and/or copies the nodes in each bin to new table. See
2309 * above for explanation.
2310 */
2311 private final void transfer(Node<K,V>[] tab, Node<K,V>[] nextTab) {
2312 int n = tab.length, stride;
2313 if ((stride = (NCPU > 1) ? (n >>> 3) / NCPU : n) < MIN_TRANSFER_STRIDE)
2314 stride = MIN_TRANSFER_STRIDE; // subdivide range
2315 if (nextTab == null) { // initiating
2316 try {
2317 @SuppressWarnings({"rawtypes","unchecked"})
2318 Node<K,V>[] nt = (Node<K,V>[])new Node[n << 1];
2319 nextTab = nt;
2320 } catch (Throwable ex) { // try to cope with OOME
2321 sizeCtl = Integer.MAX_VALUE;
2322 return;
2323 }
2324 nextTable = nextTab;
2325 transferOrigin = n;
2326 transferIndex = n;
2327 ForwardingNode<K,V> rev = new ForwardingNode<K,V>(tab);
2328 for (int k = n; k > 0;) { // progressively reveal ready slots
2329 int nextk = (k > stride) ? k - stride : 0;
2330 for (int m = nextk; m < k; ++m)
2331 nextTab[m] = rev;
2332 for (int m = n + nextk; m < n + k; ++m)
2333 nextTab[m] = rev;
2334 U.putOrderedInt(this, TRANSFERORIGIN, k = nextk);
2335 }
2336 }
2337 int nextn = nextTab.length;
2338 ForwardingNode<K,V> fwd = new ForwardingNode<K,V>(nextTab);
2339 boolean advance = true;
2340 boolean finishing = false; // to ensure sweep before committing nextTab
2341 for (int i = 0, bound = 0;;) {
2342 int nextIndex, nextBound, fh; Node<K,V> f;
2343 while (advance) {
2344 if (--i >= bound || finishing)
2345 advance = false;
2346 else if ((nextIndex = transferIndex) <= transferOrigin) {
2347 i = -1;
2348 advance = false;
2349 }
2350 else if (U.compareAndSwapInt
2351 (this, TRANSFERINDEX, nextIndex,
2352 nextBound = (nextIndex > stride ?
2353 nextIndex - stride : 0))) {
2354 bound = nextBound;
2355 i = nextIndex - 1;
2356 advance = false;
2357 }
2358 }
2359 if (i < 0 || i >= n || i + n >= nextn) {
2360 if (finishing) {
2361 nextTable = null;
2362 table = nextTab;
2363 sizeCtl = (n << 1) - (n >>> 1);
2364 return;
2365 }
2366 for (int sc;;) {
2367 if (U.compareAndSwapInt(this, SIZECTL, sc = sizeCtl, ++sc)) {
2368 if (sc != -1)
2369 return;
2370 finishing = advance = true;
2371 i = n; // recheck before commit
2372 break;
2373 }
2374 }
2375 }
2376 else if ((f = tabAt(tab, i)) == null) {
2377 if (casTabAt(tab, i, null, fwd)) {
2378 setTabAt(nextTab, i, null);
2379 setTabAt(nextTab, i + n, null);
2380 advance = true;
2381 }
2382 }
2383 else if ((fh = f.hash) == MOVED)
2384 advance = true; // already processed
2385 else {
2386 synchronized (f) {
2387 if (tabAt(tab, i) == f) {
2388 Node<K,V> ln, hn;
2389 if (fh >= 0) {
2390 int runBit = fh & n;
2391 Node<K,V> lastRun = f;
2392 for (Node<K,V> p = f.next; p != null; p = p.next) {
2393 int b = p.hash & n;
2394 if (b != runBit) {
2395 runBit = b;
2396 lastRun = p;
2397 }
2398 }
2399 if (runBit == 0) {
2400 ln = lastRun;
2401 hn = null;
2402 }
2403 else {
2404 hn = lastRun;
2405 ln = null;
2406 }
2407 for (Node<K,V> p = f; p != lastRun; p = p.next) {
2408 int ph = p.hash; K pk = p.key; V pv = p.val;
2409 if ((ph & n) == 0)
2410 ln = new Node<K,V>(ph, pk, pv, ln);
2411 else
2412 hn = new Node<K,V>(ph, pk, pv, hn);
2413 }
2414 setTabAt(nextTab, i, ln);
2415 setTabAt(nextTab, i + n, hn);
2416 setTabAt(tab, i, fwd);
2417 advance = true;
2418 }
2419 else if (f instanceof TreeBin) {
2420 TreeBin<K,V> t = (TreeBin<K,V>)f;
2421 TreeNode<K,V> lo = null, loTail = null;
2422 TreeNode<K,V> hi = null, hiTail = null;
2423 int lc = 0, hc = 0;
2424 for (Node<K,V> e = t.first; e != null; e = e.next) {
2425 int h = e.hash;
2426 TreeNode<K,V> p = new TreeNode<K,V>
2427 (h, e.key, e.val, null, null);
2428 if ((h & n) == 0) {
2429 if ((p.prev = loTail) == null)
2430 lo = p;
2431 else
2432 loTail.next = p;
2433 loTail = p;
2434 ++lc;
2435 }
2436 else {
2437 if ((p.prev = hiTail) == null)
2438 hi = p;
2439 else
2440 hiTail.next = p;
2441 hiTail = p;
2442 ++hc;
2443 }
2444 }
2445 ln = (lc <= UNTREEIFY_THRESHOLD) ? untreeify(lo) :
2446 (hc != 0) ? new TreeBin<K,V>(lo) : t;
2447 hn = (hc <= UNTREEIFY_THRESHOLD) ? untreeify(hi) :
2448 (lc != 0) ? new TreeBin<K,V>(hi) : t;
2449 setTabAt(nextTab, i, ln);
2450 setTabAt(nextTab, i + n, hn);
2451 setTabAt(tab, i, fwd);
2452 advance = true;
2453 }
2454 }
2455 }
2456 }
2457 }
2458 }
2459
2460 /* ---------------- Conversion from/to TreeBins -------------- */
2461
2462 /**
2463 * Replaces all linked nodes in bin at given index unless table is
2464 * too small, in which case resizes instead.
2465 */
2466 private final void treeifyBin(Node<K,V>[] tab, int index) {
2467 Node<K,V> b; int n, sc;
2468 if (tab != null) {
2469 if ((n = tab.length) < MIN_TREEIFY_CAPACITY) {
2470 if (tab == table && (sc = sizeCtl) >= 0 &&
2471 U.compareAndSwapInt(this, SIZECTL, sc, -2))
2472 transfer(tab, null);
2473 }
2474 else if ((b = tabAt(tab, index)) != null && b.hash >= 0) {
2475 synchronized (b) {
2476 if (tabAt(tab, index) == b) {
2477 TreeNode<K,V> hd = null, tl = null;
2478 for (Node<K,V> e = b; e != null; e = e.next) {
2479 TreeNode<K,V> p =
2480 new TreeNode<K,V>(e.hash, e.key, e.val,
2481 null, null);
2482 if ((p.prev = tl) == null)
2483 hd = p;
2484 else
2485 tl.next = p;
2486 tl = p;
2487 }
2488 setTabAt(tab, index, new TreeBin<K,V>(hd));
2489 }
2490 }
2491 }
2492 }
2493 }
2494
2495 /**
2496 * Returns a list on non-TreeNodes replacing those in given list.
2497 */
2498 static <K,V> Node<K,V> untreeify(Node<K,V> b) {
2499 Node<K,V> hd = null, tl = null;
2500 for (Node<K,V> q = b; q != null; q = q.next) {
2501 Node<K,V> p = new Node<K,V>(q.hash, q.key, q.val, null);
2502 if (tl == null)
2503 hd = p;
2504 else
2505 tl.next = p;
2506 tl = p;
2507 }
2508 return hd;
2509 }
2510
2511 /* ---------------- TreeNodes -------------- */
2512
2513 /**
2514 * Nodes for use in TreeBins
2515 */
2516 static final class TreeNode<K,V> extends Node<K,V> {
2517 TreeNode<K,V> parent; // red-black tree links
2518 TreeNode<K,V> left;
2519 TreeNode<K,V> right;
2520 TreeNode<K,V> prev; // needed to unlink next upon deletion
2521 boolean red;
2522
2523 TreeNode(int hash, K key, V val, Node<K,V> next,
2524 TreeNode<K,V> parent) {
2525 super(hash, key, val, next);
2526 this.parent = parent;
2527 }
2528
2529 Node<K,V> find(int h, Object k) {
2530 return findTreeNode(h, k, null);
2531 }
2532
2533 /**
2534 * Returns the TreeNode (or null if not found) for the given key
2535 * starting at given root.
2536 */
2537 final TreeNode<K,V> findTreeNode(int h, Object k, Class<?> kc) {
2538 if (k != null) {
2539 TreeNode<K,V> p = this;
2540 do {
2541 int ph, dir; K pk; TreeNode<K,V> q;
2542 TreeNode<K,V> pl = p.left, pr = p.right;
2543 if ((ph = p.hash) > h)
2544 p = pl;
2545 else if (ph < h)
2546 p = pr;
2547 else if ((pk = p.key) == k || (pk != null && k.equals(pk)))
2548 return p;
2549 else if (pl == null && pr == null)
2550 break;
2551 else if ((kc != null ||
2552 (kc = comparableClassFor(k)) != null) &&
2553 (dir = compareComparables(kc, k, pk)) != 0)
2554 p = (dir < 0) ? pl : pr;
2555 else if (pl == null)
2556 p = pr;
2557 else if (pr == null ||
2558 (q = pr.findTreeNode(h, k, kc)) == null)
2559 p = pl;
2560 else
2561 return q;
2562 } while (p != null);
2563 }
2564 return null;
2565 }
2566 }
2567
2568 /* ---------------- TreeBins -------------- */
2569
2570 /**
2571 * TreeNodes used at the heads of bins. TreeBins do not hold user
2572 * keys or values, but instead point to list of TreeNodes and
2573 * their root. They also maintain a parasitic read-write lock
2574 * forcing writers (who hold bin lock) to wait for readers (who do
2575 * not) to complete before tree restructuring operations.
2576 */
2577 static final class TreeBin<K,V> extends Node<K,V> {
2578 TreeNode<K,V> root;
2579 volatile TreeNode<K,V> first;
2580 volatile Thread waiter;
2581 volatile int lockState;
2582 // values for lockState
2583 static final int WRITER = 1; // set while holding write lock
2584 static final int WAITER = 2; // set when waiting for write lock
2585 static final int READER = 4; // increment value for setting read lock
2586
2587 /**
2588 * Creates bin with initial set of nodes headed by b.
2589 */
2590 TreeBin(TreeNode<K,V> b) {
2591 super(TREEBIN, null, null, null);
2592 this.first = b;
2593 TreeNode<K,V> r = null;
2594 for (TreeNode<K,V> x = b, next; x != null; x = next) {
2595 next = (TreeNode<K,V>)x.next;
2596 x.left = x.right = null;
2597 if (r == null) {
2598 x.parent = null;
2599 x.red = false;
2600 r = x;
2601 }
2602 else {
2603 Object key = x.key;
2604 int hash = x.hash;
2605 Class<?> kc = null;
2606 for (TreeNode<K,V> p = r;;) {
2607 int dir, ph;
2608 if ((ph = p.hash) > hash)
2609 dir = -1;
2610 else if (ph < hash)
2611 dir = 1;
2612 else if ((kc != null ||
2613 (kc = comparableClassFor(key)) != null))
2614 dir = compareComparables(kc, key, p.key);
2615 else
2616 dir = 0;
2617 TreeNode<K,V> xp = p;
2618 if ((p = (dir <= 0) ? p.left : p.right) == null) {
2619 x.parent = xp;
2620 if (dir <= 0)
2621 xp.left = x;
2622 else
2623 xp.right = x;
2624 r = balanceInsertion(r, x);
2625 break;
2626 }
2627 }
2628 }
2629 }
2630 this.root = r;
2631 }
2632
2633 /**
2634 * Acquires write lock for tree restructuring.
2635 */
2636 private final void lockRoot() {
2637 if (!U.compareAndSwapInt(this, LOCKSTATE, 0, WRITER))
2638 contendedLock(); // offload to separate method
2639 }
2640
2641 /**
2642 * Releases write lock for tree restructuring.
2643 */
2644 private final void unlockRoot() {
2645 lockState = 0;
2646 }
2647
2648 /**
2649 * Possibly blocks awaiting root lock.
2650 */
2651 private final void contendedLock() {
2652 boolean waiting = false;
2653 for (int s;;) {
2654 if (((s = lockState) & WRITER) == 0) {
2655 if (U.compareAndSwapInt(this, LOCKSTATE, s, WRITER)) {
2656 if (waiting)
2657 waiter = null;
2658 return;
2659 }
2660 }
2661 else if ((s | WAITER) == 0) {
2662 if (U.compareAndSwapInt(this, LOCKSTATE, s, s | WAITER)) {
2663 waiting = true;
2664 waiter = Thread.currentThread();
2665 }
2666 }
2667 else if (waiting)
2668 LockSupport.park(this);
2669 }
2670 }
2671
2672 /**
2673 * Returns matching node or null if none. Tries to search
2674 * using tree comparisons from root, but continues linear
2675 * search when lock not available.
2676 */
2677 final Node<K,V> find(int h, Object k) {
2678 if (k != null) {
2679 for (Node<K,V> e = first; e != null; e = e.next) {
2680 int s; K ek;
2681 if (((s = lockState) & (WAITER|WRITER)) != 0) {
2682 if (e.hash == h &&
2683 ((ek = e.key) == k || (ek != null && k.equals(ek))))
2684 return e;
2685 }
2686 else if (U.compareAndSwapInt(this, LOCKSTATE, s,
2687 s + READER)) {
2688 TreeNode<K,V> r, p;
2689 try {
2690 p = ((r = root) == null ? null :
2691 r.findTreeNode(h, k, null));
2692 } finally {
2693 Thread w;
2694 int ls;
2695 do {} while (!U.compareAndSwapInt
2696 (this, LOCKSTATE,
2697 ls = lockState, ls - READER));
2698 if (ls == (READER|WAITER) && (w = waiter) != null)
2699 LockSupport.unpark(w);
2700 }
2701 return p;
2702 }
2703 }
2704 }
2705 return null;
2706 }
2707
2708 /**
2709 * Finds or adds a node.
2710 * @return null if added
2711 */
2712 final TreeNode<K,V> putTreeVal(int h, K k, V v) {
2713 Class<?> kc = null;
2714 for (TreeNode<K,V> p = root;;) {
2715 int dir, ph; K pk; TreeNode<K,V> q, pr;
2716 if (p == null) {
2717 first = root = new TreeNode<K,V>(h, k, v, null, null);
2718 break;
2719 }
2720 else if ((ph = p.hash) > h)
2721 dir = -1;
2722 else if (ph < h)
2723 dir = 1;
2724 else if ((pk = p.key) == k || (pk != null && k.equals(pk)))
2725 return p;
2726 else if ((kc == null &&
2727 (kc = comparableClassFor(k)) == null) ||
2728 (dir = compareComparables(kc, k, pk)) == 0) {
2729 if (p.left == null)
2730 dir = 1;
2731 else if ((pr = p.right) == null ||
2732 (q = pr.findTreeNode(h, k, kc)) == null)
2733 dir = -1;
2734 else
2735 return q;
2736 }
2737 TreeNode<K,V> xp = p;
2738 if ((p = (dir < 0) ? p.left : p.right) == null) {
2739 TreeNode<K,V> x, f = first;
2740 first = x = new TreeNode<K,V>(h, k, v, f, xp);
2741 if (f != null)
2742 f.prev = x;
2743 if (dir < 0)
2744 xp.left = x;
2745 else
2746 xp.right = x;
2747 if (!xp.red)
2748 x.red = true;
2749 else {
2750 lockRoot();
2751 try {
2752 root = balanceInsertion(root, x);
2753 } finally {
2754 unlockRoot();
2755 }
2756 }
2757 break;
2758 }
2759 }
2760 assert checkInvariants(root);
2761 return null;
2762 }
2763
2764 /**
2765 * Removes the given node, that must be present before this
2766 * call. This is messier than typical red-black deletion code
2767 * because we cannot swap the contents of an interior node
2768 * with a leaf successor that is pinned by "next" pointers
2769 * that are accessible independently of lock. So instead we
2770 * swap the tree linkages.
2771 *
2772 * @return true if now too small, so should be untreeified
2773 */
2774 final boolean removeTreeNode(TreeNode<K,V> p) {
2775 TreeNode<K,V> next = (TreeNode<K,V>)p.next;
2776 TreeNode<K,V> pred = p.prev; // unlink traversal pointers
2777 TreeNode<K,V> r, rl;
2778 if (pred == null)
2779 first = next;
2780 else
2781 pred.next = next;
2782 if (next != null)
2783 next.prev = pred;
2784 if (first == null) {
2785 root = null;
2786 return true;
2787 }
2788 if ((r = root) == null || r.right == null || // too small
2789 (rl = r.left) == null || rl.left == null)
2790 return true;
2791 lockRoot();
2792 try {
2793 TreeNode<K,V> replacement;
2794 TreeNode<K,V> pl = p.left;
2795 TreeNode<K,V> pr = p.right;
2796 if (pl != null && pr != null) {
2797 TreeNode<K,V> s = pr, sl;
2798 while ((sl = s.left) != null) // find successor
2799 s = sl;
2800 boolean c = s.red; s.red = p.red; p.red = c; // swap colors
2801 TreeNode<K,V> sr = s.right;
2802 TreeNode<K,V> pp = p.parent;
2803 if (s == pr) { // p was s's direct parent
2804 p.parent = s;
2805 s.right = p;
2806 }
2807 else {
2808 TreeNode<K,V> sp = s.parent;
2809 if ((p.parent = sp) != null) {
2810 if (s == sp.left)
2811 sp.left = p;
2812 else
2813 sp.right = p;
2814 }
2815 if ((s.right = pr) != null)
2816 pr.parent = s;
2817 }
2818 p.left = null;
2819 if ((p.right = sr) != null)
2820 sr.parent = p;
2821 if ((s.left = pl) != null)
2822 pl.parent = s;
2823 if ((s.parent = pp) == null)
2824 r = s;
2825 else if (p == pp.left)
2826 pp.left = s;
2827 else
2828 pp.right = s;
2829 if (sr != null)
2830 replacement = sr;
2831 else
2832 replacement = p;
2833 }
2834 else if (pl != null)
2835 replacement = pl;
2836 else if (pr != null)
2837 replacement = pr;
2838 else
2839 replacement = p;
2840 if (replacement != p) {
2841 TreeNode<K,V> pp = replacement.parent = p.parent;
2842 if (pp == null)
2843 r = replacement;
2844 else if (p == pp.left)
2845 pp.left = replacement;
2846 else
2847 pp.right = replacement;
2848 p.left = p.right = p.parent = null;
2849 }
2850
2851 root = (p.red) ? r : balanceDeletion(r, replacement);
2852
2853 if (p == replacement) { // detach pointers
2854 TreeNode<K,V> pp;
2855 if ((pp = p.parent) != null) {
2856 if (p == pp.left)
2857 pp.left = null;
2858 else if (p == pp.right)
2859 pp.right = null;
2860 p.parent = null;
2861 }
2862 }
2863 } finally {
2864 unlockRoot();
2865 }
2866 assert checkInvariants(root);
2867 return false;
2868 }
2869
2870 /* ------------------------------------------------------------ */
2871 // Red-black tree methods, all adapted from CLR
2872
2873 static <K,V> TreeNode<K,V> rotateLeft(TreeNode<K,V> root,
2874 TreeNode<K,V> p) {
2875 TreeNode<K,V> r, pp, rl;
2876 if (p != null && (r = p.right) != null) {
2877 if ((rl = p.right = r.left) != null)
2878 rl.parent = p;
2879 if ((pp = r.parent = p.parent) == null)
2880 (root = r).red = false;
2881 else if (pp.left == p)
2882 pp.left = r;
2883 else
2884 pp.right = r;
2885 r.left = p;
2886 p.parent = r;
2887 }
2888 return root;
2889 }
2890
2891 static <K,V> TreeNode<K,V> rotateRight(TreeNode<K,V> root,
2892 TreeNode<K,V> p) {
2893 TreeNode<K,V> l, pp, lr;
2894 if (p != null && (l = p.left) != null) {
2895 if ((lr = p.left = l.right) != null)
2896 lr.parent = p;
2897 if ((pp = l.parent = p.parent) == null)
2898 (root = l).red = false;
2899 else if (pp.right == p)
2900 pp.right = l;
2901 else
2902 pp.left = l;
2903 l.right = p;
2904 p.parent = l;
2905 }
2906 return root;
2907 }
2908
2909 static <K,V> TreeNode<K,V> balanceInsertion(TreeNode<K,V> root,
2910 TreeNode<K,V> x) {
2911 x.red = true;
2912 for (TreeNode<K,V> xp, xpp, xppl, xppr;;) {
2913 if ((xp = x.parent) == null) {
2914 x.red = false;
2915 return x;
2916 }
2917 else if (!xp.red || (xpp = xp.parent) == null)
2918 return root;
2919 if (xp == (xppl = xpp.left)) {
2920 if ((xppr = xpp.right) != null && xppr.red) {
2921 xppr.red = false;
2922 xp.red = false;
2923 xpp.red = true;
2924 x = xpp;
2925 }
2926 else {
2927 if (x == xp.right) {
2928 root = rotateLeft(root, x = xp);
2929 xpp = (xp = x.parent) == null ? null : xp.parent;
2930 }
2931 if (xp != null) {
2932 xp.red = false;
2933 if (xpp != null) {
2934 xpp.red = true;
2935 root = rotateRight(root, xpp);
2936 }
2937 }
2938 }
2939 }
2940 else {
2941 if (xppl != null && xppl.red) {
2942 xppl.red = false;
2943 xp.red = false;
2944 xpp.red = true;
2945 x = xpp;
2946 }
2947 else {
2948 if (x == xp.left) {
2949 root = rotateRight(root, x = xp);
2950 xpp = (xp = x.parent) == null ? null : xp.parent;
2951 }
2952 if (xp != null) {
2953 xp.red = false;
2954 if (xpp != null) {
2955 xpp.red = true;
2956 root = rotateLeft(root, xpp);
2957 }
2958 }
2959 }
2960 }
2961 }
2962 }
2963
2964 static <K,V> TreeNode<K,V> balanceDeletion(TreeNode<K,V> root,
2965 TreeNode<K,V> x) {
2966 for (TreeNode<K,V> xp, xpl, xpr;;) {
2967 if (x == null || x == root)
2968 return root;
2969 else if ((xp = x.parent) == null) {
2970 x.red = false;
2971 return x;
2972 }
2973 else if (x.red) {
2974 x.red = false;
2975 return root;
2976 }
2977 else if ((xpl = xp.left) == x) {
2978 if ((xpr = xp.right) != null && xpr.red) {
2979 xpr.red = false;
2980 xp.red = true;
2981 root = rotateLeft(root, xp);
2982 xpr = (xp = x.parent) == null ? null : xp.right;
2983 }
2984 if (xpr == null)
2985 x = xp;
2986 else {
2987 TreeNode<K,V> sl = xpr.left, sr = xpr.right;
2988 if ((sr == null || !sr.red) &&
2989 (sl == null || !sl.red)) {
2990 xpr.red = true;
2991 x = xp;
2992 }
2993 else {
2994 if (sr == null || !sr.red) {
2995 if (sl != null)
2996 sl.red = false;
2997 xpr.red = true;
2998 root = rotateRight(root, xpr);
2999 xpr = (xp = x.parent) == null ?
3000 null : xp.right;
3001 }
3002 if (xpr != null) {
3003 xpr.red = (xp == null) ? false : xp.red;
3004 if ((sr = xpr.right) != null)
3005 sr.red = false;
3006 }
3007 if (xp != null) {
3008 xp.red = false;
3009 root = rotateLeft(root, xp);
3010 }
3011 x = root;
3012 }
3013 }
3014 }
3015 else { // symmetric
3016 if (xpl != null && xpl.red) {
3017 xpl.red = false;
3018 xp.red = true;
3019 root = rotateRight(root, xp);
3020 xpl = (xp = x.parent) == null ? null : xp.left;
3021 }
3022 if (xpl == null)
3023 x = xp;
3024 else {
3025 TreeNode<K,V> sl = xpl.left, sr = xpl.right;
3026 if ((sl == null || !sl.red) &&
3027 (sr == null || !sr.red)) {
3028 xpl.red = true;
3029 x = xp;
3030 }
3031 else {
3032 if (sl == null || !sl.red) {
3033 if (sr != null)
3034 sr.red = false;
3035 xpl.red = true;
3036 root = rotateLeft(root, xpl);
3037 xpl = (xp = x.parent) == null ?
3038 null : xp.left;
3039 }
3040 if (xpl != null) {
3041 xpl.red = (xp == null) ? false : xp.red;
3042 if ((sl = xpl.left) != null)
3043 sl.red = false;
3044 }
3045 if (xp != null) {
3046 xp.red = false;
3047 root = rotateRight(root, xp);
3048 }
3049 x = root;
3050 }
3051 }
3052 }
3053 }
3054 }
3055
3056 /**
3057 * Recursive invariant check
3058 */
3059 static <K,V> boolean checkInvariants(TreeNode<K,V> t) {
3060 TreeNode<K,V> tp = t.parent, tl = t.left, tr = t.right,
3061 tb = t.prev, tn = (TreeNode<K,V>)t.next;
3062 if (tb != null && tb.next != t)
3063 return false;
3064 if (tn != null && tn.prev != t)
3065 return false;
3066 if (tp != null && t != tp.left && t != tp.right)
3067 return false;
3068 if (tl != null && (tl.parent != t || tl.hash > t.hash))
3069 return false;
3070 if (tr != null && (tr.parent != t || tr.hash < t.hash))
3071 return false;
3072 if (t.red && tl != null && tl.red && tr != null && tr.red)
3073 return false;
3074 if (tl != null && !checkInvariants(tl))
3075 return false;
3076 if (tr != null && !checkInvariants(tr))
3077 return false;
3078 return true;
3079 }
3080
3081 private static final sun.misc.Unsafe U;
3082 private static final long LOCKSTATE;
3083 static {
3084 try {
3085 U = getUnsafe();
3086 Class<?> k = TreeBin.class;
3087 LOCKSTATE = U.objectFieldOffset
3088 (k.getDeclaredField("lockState"));
3089 } catch (Exception e) {
3090 throw new Error(e);
3091 }
3092 }
3093 }
3094
3095 /* ----------------Table Traversal -------------- */
3096
3097 /**
3098 * Encapsulates traversal for methods such as containsValue; also
3099 * serves as a base class for other iterators and spliterators.
3100 *
3101 * Method advance visits once each still-valid node that was
3102 * reachable upon iterator construction. It might miss some that
3103 * were added to a bin after the bin was visited, which is OK wrt
3104 * consistency guarantees. Maintaining this property in the face
3105 * of possible ongoing resizes requires a fair amount of
3106 * bookkeeping state that is difficult to optimize away amidst
3107 * volatile accesses. Even so, traversal maintains reasonable
3108 * throughput.
3109 *
3110 * Normally, iteration proceeds bin-by-bin traversing lists.
3111 * However, if the table has been resized, then all future steps
3112 * must traverse both the bin at the current index as well as at
3113 * (index + baseSize); and so on for further resizings. To
3114 * paranoically cope with potential sharing by users of iterators
3115 * across threads, iteration terminates if a bounds checks fails
3116 * for a table read.
3117 */
3118 static class Traverser<K,V> {
3119 Node<K,V>[] tab; // current table; updated if resized
3120 Node<K,V> next; // the next entry to use
3121 int index; // index of bin to use next
3122 int baseIndex; // current index of initial table
3123 int baseLimit; // index bound for initial table
3124 final int baseSize; // initial table size
3125
3126 Traverser(Node<K,V>[] tab, int size, int index, int limit) {
3127 this.tab = tab;
3128 this.baseSize = size;
3129 this.baseIndex = this.index = index;
3130 this.baseLimit = limit;
3131 this.next = null;
3132 }
3133
3134 /**
3135 * Advances if possible, returning next valid node, or null if none.
3136 */
3137 final Node<K,V> advance() {
3138 Node<K,V> e;
3139 if ((e = next) != null)
3140 e = e.next;
3141 for (;;) {
3142 Node<K,V>[] t; int i, n; K ek; // must use locals in checks
3143 if (e != null)
3144 return next = e;
3145 if (baseIndex >= baseLimit || (t = tab) == null ||
3146 (n = t.length) <= (i = index) || i < 0)
3147 return next = null;
3148 if ((e = tabAt(t, index)) != null && e.hash < 0) {
3149 if (e instanceof ForwardingNode) {
3150 tab = ((ForwardingNode<K,V>)e).nextTable;
3151 e = null;
3152 continue;
3153 }
3154 else if (e instanceof TreeBin)
3155 e = ((TreeBin<K,V>)e).first;
3156 else
3157 e = null;
3158 }
3159 if ((index += baseSize) >= n)
3160 index = ++baseIndex; // visit upper slots if present
3161 }
3162 }
3163 }
3164
3165 /**
3166 * Base of key, value, and entry Iterators. Adds fields to
3167 * Traverser to support iterator.remove.
3168 */
3169 static class BaseIterator<K,V> extends Traverser<K,V> {
3170 final ConcurrentHashMapV8<K,V> map;
3171 Node<K,V> lastReturned;
3172 BaseIterator(Node<K,V>[] tab, int size, int index, int limit,
3173 ConcurrentHashMapV8<K,V> map) {
3174 super(tab, size, index, limit);
3175 this.map = map;
3176 advance();
3177 }
3178
3179 public final boolean hasNext() { return next != null; }
3180 public final boolean hasMoreElements() { return next != null; }
3181
3182 public final void remove() {
3183 Node<K,V> p;
3184 if ((p = lastReturned) == null)
3185 throw new IllegalStateException();
3186 lastReturned = null;
3187 map.replaceNode(p.key, null, null);
3188 }
3189 }
3190
3191 static final class KeyIterator<K,V> extends BaseIterator<K,V>
3192 implements Iterator<K>, Enumeration<K> {
3193 KeyIterator(Node<K,V>[] tab, int index, int size, int limit,
3194 ConcurrentHashMapV8<K,V> map) {
3195 super(tab, index, size, limit, map);
3196 }
3197
3198 public final K next() {
3199 Node<K,V> p;
3200 if ((p = next) == null)
3201 throw new NoSuchElementException();
3202 K k = p.key;
3203 lastReturned = p;
3204 advance();
3205 return k;
3206 }
3207
3208 public final K nextElement() { return next(); }
3209 }
3210
3211 static final class ValueIterator<K,V> extends BaseIterator<K,V>
3212 implements Iterator<V>, Enumeration<V> {
3213 ValueIterator(Node<K,V>[] tab, int index, int size, int limit,
3214 ConcurrentHashMapV8<K,V> map) {
3215 super(tab, index, size, limit, map);
3216 }
3217
3218 public final V next() {
3219 Node<K,V> p;
3220 if ((p = next) == null)
3221 throw new NoSuchElementException();
3222 V v = p.val;
3223 lastReturned = p;
3224 advance();
3225 return v;
3226 }
3227
3228 public final V nextElement() { return next(); }
3229 }
3230
3231 static final class EntryIterator<K,V> extends BaseIterator<K,V>
3232 implements Iterator<Map.Entry<K,V>> {
3233 EntryIterator(Node<K,V>[] tab, int index, int size, int limit,
3234 ConcurrentHashMapV8<K,V> map) {
3235 super(tab, index, size, limit, map);
3236 }
3237
3238 public final Map.Entry<K,V> next() {
3239 Node<K,V> p;
3240 if ((p = next) == null)
3241 throw new NoSuchElementException();
3242 K k = p.key;
3243 V v = p.val;
3244 lastReturned = p;
3245 advance();
3246 return new MapEntry<K,V>(k, v, map);
3247 }
3248 }
3249
3250 /**
3251 * Exported Entry for EntryIterator
3252 */
3253 static final class MapEntry<K,V> implements Map.Entry<K,V> {
3254 final K key; // non-null
3255 V val; // non-null
3256 final ConcurrentHashMapV8<K,V> map;
3257 MapEntry(K key, V val, ConcurrentHashMapV8<K,V> map) {
3258 this.key = key;
3259 this.val = val;
3260 this.map = map;
3261 }
3262 public K getKey() { return key; }
3263 public V getValue() { return val; }
3264 public int hashCode() { return key.hashCode() ^ val.hashCode(); }
3265 public String toString() { return key + "=" + val; }
3266
3267 public boolean equals(Object o) {
3268 Object k, v; Map.Entry<?,?> e;
3269 return ((o instanceof Map.Entry) &&
3270 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
3271 (v = e.getValue()) != null &&
3272 (k == key || k.equals(key)) &&
3273 (v == val || v.equals(val)));
3274 }
3275
3276 /**
3277 * Sets our entry's value and writes through to the map. The
3278 * value to return is somewhat arbitrary here. Since we do not
3279 * necessarily track asynchronous changes, the most recent
3280 * "previous" value could be different from what we return (or
3281 * could even have been removed, in which case the put will
3282 * re-establish). We do not and cannot guarantee more.
3283 */
3284 public V setValue(V value) {
3285 if (value == null) throw new NullPointerException();
3286 V v = val;
3287 val = value;
3288 map.put(key, value);
3289 return v;
3290 }
3291 }
3292
3293 static final class KeySpliterator<K,V> extends Traverser<K,V>
3294 implements ConcurrentHashMapSpliterator<K> {
3295 long est; // size estimate
3296 KeySpliterator(Node<K,V>[] tab, int size, int index, int limit,
3297 long est) {
3298 super(tab, size, index, limit);
3299 this.est = est;
3300 }
3301
3302 public ConcurrentHashMapSpliterator<K> trySplit() {
3303 int i, f, h;
3304 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3305 new KeySpliterator<K,V>(tab, baseSize, baseLimit = h,
3306 f, est >>>= 1);
3307 }
3308
3309 public void forEachRemaining(Action<? super K> action) {
3310 if (action == null) throw new NullPointerException();
3311 for (Node<K,V> p; (p = advance()) != null;)
3312 action.apply(p.key);
3313 }
3314
3315 public boolean tryAdvance(Action<? super K> action) {
3316 if (action == null) throw new NullPointerException();
3317 Node<K,V> p;
3318 if ((p = advance()) == null)
3319 return false;
3320 action.apply(p.key);
3321 return true;
3322 }
3323
3324 public long estimateSize() { return est; }
3325
3326 }
3327
3328 static final class ValueSpliterator<K,V> extends Traverser<K,V>
3329 implements ConcurrentHashMapSpliterator<V> {
3330 long est; // size estimate
3331 ValueSpliterator(Node<K,V>[] tab, int size, int index, int limit,
3332 long est) {
3333 super(tab, size, index, limit);
3334 this.est = est;
3335 }
3336
3337 public ConcurrentHashMapSpliterator<V> trySplit() {
3338 int i, f, h;
3339 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3340 new ValueSpliterator<K,V>(tab, baseSize, baseLimit = h,
3341 f, est >>>= 1);
3342 }
3343
3344 public void forEachRemaining(Action<? super V> action) {
3345 if (action == null) throw new NullPointerException();
3346 for (Node<K,V> p; (p = advance()) != null;)
3347 action.apply(p.val);
3348 }
3349
3350 public boolean tryAdvance(Action<? super V> action) {
3351 if (action == null) throw new NullPointerException();
3352 Node<K,V> p;
3353 if ((p = advance()) == null)
3354 return false;
3355 action.apply(p.val);
3356 return true;
3357 }
3358
3359 public long estimateSize() { return est; }
3360
3361 }
3362
3363 static final class EntrySpliterator<K,V> extends Traverser<K,V>
3364 implements ConcurrentHashMapSpliterator<Map.Entry<K,V>> {
3365 final ConcurrentHashMapV8<K,V> map; // To export MapEntry
3366 long est; // size estimate
3367 EntrySpliterator(Node<K,V>[] tab, int size, int index, int limit,
3368 long est, ConcurrentHashMapV8<K,V> map) {
3369 super(tab, size, index, limit);
3370 this.map = map;
3371 this.est = est;
3372 }
3373
3374 public ConcurrentHashMapSpliterator<Map.Entry<K,V>> trySplit() {
3375 int i, f, h;
3376 return (h = ((i = baseIndex) + (f = baseLimit)) >>> 1) <= i ? null :
3377 new EntrySpliterator<K,V>(tab, baseSize, baseLimit = h,
3378 f, est >>>= 1, map);
3379 }
3380
3381 public void forEachRemaining(Action<? super Map.Entry<K,V>> action) {
3382 if (action == null) throw new NullPointerException();
3383 for (Node<K,V> p; (p = advance()) != null; )
3384 action.apply(new MapEntry<K,V>(p.key, p.val, map));
3385 }
3386
3387 public boolean tryAdvance(Action<? super Map.Entry<K,V>> action) {
3388 if (action == null) throw new NullPointerException();
3389 Node<K,V> p;
3390 if ((p = advance()) == null)
3391 return false;
3392 action.apply(new MapEntry<K,V>(p.key, p.val, map));
3393 return true;
3394 }
3395
3396 public long estimateSize() { return est; }
3397
3398 }
3399
3400 // Parallel bulk operations
3401
3402 /**
3403 * Computes initial batch value for bulk tasks. The returned value
3404 * is approximately exp2 of the number of times (minus one) to
3405 * split task by two before executing leaf action. This value is
3406 * faster to compute and more convenient to use as a guide to
3407 * splitting than is the depth, since it is used while dividing by
3408 * two anyway.
3409 */
3410 final int batchFor(long b) {
3411 long n;
3412 if (b == Long.MAX_VALUE || (n = sumCount()) <= 1L || n < b)
3413 return 0;
3414 int sp = ForkJoinPool.getCommonPoolParallelism() << 2; // slack of 4
3415 return (b <= 0L || (n /= b) >= sp) ? sp : (int)n;
3416 }
3417
3418 /**
3419 * Performs the given action for each (key, value).
3420 *
3421 * @param parallelismThreshold the (estimated) number of elements
3422 * needed for this operation to be executed in parallel
3423 * @param action the action
3424 * @since 1.8
3425 */
3426 public void forEach(long parallelismThreshold,
3427 BiAction<? super K,? super V> action) {
3428 if (action == null) throw new NullPointerException();
3429 new ForEachMappingTask<K,V>
3430 (null, batchFor(parallelismThreshold), 0, 0, table,
3431 action).invoke();
3432 }
3433
3434 /**
3435 * Performs the given action for each non-null transformation
3436 * of each (key, value).
3437 *
3438 * @param parallelismThreshold the (estimated) number of elements
3439 * needed for this operation to be executed in parallel
3440 * @param transformer a function returning the transformation
3441 * for an element, or null if there is no transformation (in
3442 * which case the action is not applied)
3443 * @param action the action
3444 * @since 1.8
3445 */
3446 public <U> void forEach(long parallelismThreshold,
3447 BiFun<? super K, ? super V, ? extends U> transformer,
3448 Action<? super U> action) {
3449 if (transformer == null || action == null)
3450 throw new NullPointerException();
3451 new ForEachTransformedMappingTask<K,V,U>
3452 (null, batchFor(parallelismThreshold), 0, 0, table,
3453 transformer, action).invoke();
3454 }
3455
3456 /**
3457 * Returns a non-null result from applying the given search
3458 * function on each (key, value), or null if none. Upon
3459 * success, further element processing is suppressed and the
3460 * results of any other parallel invocations of the search
3461 * function are ignored.
3462 *
3463 * @param parallelismThreshold the (estimated) number of elements
3464 * needed for this operation to be executed in parallel
3465 * @param searchFunction a function returning a non-null
3466 * result on success, else null
3467 * @return a non-null result from applying the given search
3468 * function on each (key, value), or null if none
3469 * @since 1.8
3470 */
3471 public <U> U search(long parallelismThreshold,
3472 BiFun<? super K, ? super V, ? extends U> searchFunction) {
3473 if (searchFunction == null) throw new NullPointerException();
3474 return new SearchMappingsTask<K,V,U>
3475 (null, batchFor(parallelismThreshold), 0, 0, table,
3476 searchFunction, new AtomicReference<U>()).invoke();
3477 }
3478
3479 /**
3480 * Returns the result of accumulating the given transformation
3481 * of all (key, value) pairs using the given reducer to
3482 * combine values, or null if none.
3483 *
3484 * @param parallelismThreshold the (estimated) number of elements
3485 * needed for this operation to be executed in parallel
3486 * @param transformer a function returning the transformation
3487 * for an element, or null if there is no transformation (in
3488 * which case it is not combined)
3489 * @param reducer a commutative associative combining function
3490 * @return the result of accumulating the given transformation
3491 * of all (key, value) pairs
3492 * @since 1.8
3493 */
3494 public <U> U reduce(long parallelismThreshold,
3495 BiFun<? super K, ? super V, ? extends U> transformer,
3496 BiFun<? super U, ? super U, ? extends U> reducer) {
3497 if (transformer == null || reducer == null)
3498 throw new NullPointerException();
3499 return new MapReduceMappingsTask<K,V,U>
3500 (null, batchFor(parallelismThreshold), 0, 0, table,
3501 null, transformer, reducer).invoke();
3502 }
3503
3504 /**
3505 * Returns the result of accumulating the given transformation
3506 * of all (key, value) pairs using the given reducer to
3507 * combine values, and the given basis as an identity value.
3508 *
3509 * @param parallelismThreshold the (estimated) number of elements
3510 * needed for this operation to be executed in parallel
3511 * @param transformer a function returning the transformation
3512 * for an element
3513 * @param basis the identity (initial default value) for the reduction
3514 * @param reducer a commutative associative combining function
3515 * @return the result of accumulating the given transformation
3516 * of all (key, value) pairs
3517 * @since 1.8
3518 */
3519 public double reduceToDouble(long parallelismThreshold,
3520 ObjectByObjectToDouble<? super K, ? super V> transformer,
3521 double basis,
3522 DoubleByDoubleToDouble reducer) {
3523 if (transformer == null || reducer == null)
3524 throw new NullPointerException();
3525 return new MapReduceMappingsToDoubleTask<K,V>
3526 (null, batchFor(parallelismThreshold), 0, 0, table,
3527 null, transformer, basis, reducer).invoke();
3528 }
3529
3530 /**
3531 * Returns the result of accumulating the given transformation
3532 * of all (key, value) pairs using the given reducer to
3533 * combine values, and the given basis as an identity value.
3534 *
3535 * @param parallelismThreshold the (estimated) number of elements
3536 * needed for this operation to be executed in parallel
3537 * @param transformer a function returning the transformation
3538 * for an element
3539 * @param basis the identity (initial default value) for the reduction
3540 * @param reducer a commutative associative combining function
3541 * @return the result of accumulating the given transformation
3542 * of all (key, value) pairs
3543 * @since 1.8
3544 */
3545 public long reduceToLong(long parallelismThreshold,
3546 ObjectByObjectToLong<? super K, ? super V> transformer,
3547 long basis,
3548 LongByLongToLong reducer) {
3549 if (transformer == null || reducer == null)
3550 throw new NullPointerException();
3551 return new MapReduceMappingsToLongTask<K,V>
3552 (null, batchFor(parallelismThreshold), 0, 0, table,
3553 null, transformer, basis, reducer).invoke();
3554 }
3555
3556 /**
3557 * Returns the result of accumulating the given transformation
3558 * of all (key, value) pairs using the given reducer to
3559 * combine values, and the given basis as an identity value.
3560 *
3561 * @param parallelismThreshold the (estimated) number of elements
3562 * needed for this operation to be executed in parallel
3563 * @param transformer a function returning the transformation
3564 * for an element
3565 * @param basis the identity (initial default value) for the reduction
3566 * @param reducer a commutative associative combining function
3567 * @return the result of accumulating the given transformation
3568 * of all (key, value) pairs
3569 * @since 1.8
3570 */
3571 public int reduceToInt(long parallelismThreshold,
3572 ObjectByObjectToInt<? super K, ? super V> transformer,
3573 int basis,
3574 IntByIntToInt reducer) {
3575 if (transformer == null || reducer == null)
3576 throw new NullPointerException();
3577 return new MapReduceMappingsToIntTask<K,V>
3578 (null, batchFor(parallelismThreshold), 0, 0, table,
3579 null, transformer, basis, reducer).invoke();
3580 }
3581
3582 /**
3583 * Performs the given action for each key.
3584 *
3585 * @param parallelismThreshold the (estimated) number of elements
3586 * needed for this operation to be executed in parallel
3587 * @param action the action
3588 * @since 1.8
3589 */
3590 public void forEachKey(long parallelismThreshold,
3591 Action<? super K> action) {
3592 if (action == null) throw new NullPointerException();
3593 new ForEachKeyTask<K,V>
3594 (null, batchFor(parallelismThreshold), 0, 0, table,
3595 action).invoke();
3596 }
3597
3598 /**
3599 * Performs the given action for each non-null transformation
3600 * of each key.
3601 *
3602 * @param parallelismThreshold the (estimated) number of elements
3603 * needed for this operation to be executed in parallel
3604 * @param transformer a function returning the transformation
3605 * for an element, or null if there is no transformation (in
3606 * which case the action is not applied)
3607 * @param action the action
3608 * @since 1.8
3609 */
3610 public <U> void forEachKey(long parallelismThreshold,
3611 Fun<? super K, ? extends U> transformer,
3612 Action<? super U> action) {
3613 if (transformer == null || action == null)
3614 throw new NullPointerException();
3615 new ForEachTransformedKeyTask<K,V,U>
3616 (null, batchFor(parallelismThreshold), 0, 0, table,
3617 transformer, action).invoke();
3618 }
3619
3620 /**
3621 * Returns a non-null result from applying the given search
3622 * function on each key, or null if none. Upon success,
3623 * further element processing is suppressed and the results of
3624 * any other parallel invocations of the search function are
3625 * ignored.
3626 *
3627 * @param parallelismThreshold the (estimated) number of elements
3628 * needed for this operation to be executed in parallel
3629 * @param searchFunction a function returning a non-null
3630 * result on success, else null
3631 * @return a non-null result from applying the given search
3632 * function on each key, or null if none
3633 * @since 1.8
3634 */
3635 public <U> U searchKeys(long parallelismThreshold,
3636 Fun<? super K, ? extends U> searchFunction) {
3637 if (searchFunction == null) throw new NullPointerException();
3638 return new SearchKeysTask<K,V,U>
3639 (null, batchFor(parallelismThreshold), 0, 0, table,
3640 searchFunction, new AtomicReference<U>()).invoke();
3641 }
3642
3643 /**
3644 * Returns the result of accumulating all keys using the given
3645 * reducer to combine values, or null if none.
3646 *
3647 * @param parallelismThreshold the (estimated) number of elements
3648 * needed for this operation to be executed in parallel
3649 * @param reducer a commutative associative combining function
3650 * @return the result of accumulating all keys using the given
3651 * reducer to combine values, or null if none
3652 * @since 1.8
3653 */
3654 public K reduceKeys(long parallelismThreshold,
3655 BiFun<? super K, ? super K, ? extends K> reducer) {
3656 if (reducer == null) throw new NullPointerException();
3657 return new ReduceKeysTask<K,V>
3658 (null, batchFor(parallelismThreshold), 0, 0, table,
3659 null, reducer).invoke();
3660 }
3661
3662 /**
3663 * Returns the result of accumulating the given transformation
3664 * of all keys using the given reducer to combine values, or
3665 * null if none.
3666 *
3667 * @param parallelismThreshold the (estimated) number of elements
3668 * needed for this operation to be executed in parallel
3669 * @param transformer a function returning the transformation
3670 * for an element, or null if there is no transformation (in
3671 * which case it is not combined)
3672 * @param reducer a commutative associative combining function
3673 * @return the result of accumulating the given transformation
3674 * of all keys
3675 * @since 1.8
3676 */
3677 public <U> U reduceKeys(long parallelismThreshold,
3678 Fun<? super K, ? extends U> transformer,
3679 BiFun<? super U, ? super U, ? extends U> reducer) {
3680 if (transformer == null || reducer == null)
3681 throw new NullPointerException();
3682 return new MapReduceKeysTask<K,V,U>
3683 (null, batchFor(parallelismThreshold), 0, 0, table,
3684 null, transformer, reducer).invoke();
3685 }
3686
3687 /**
3688 * Returns the result of accumulating the given transformation
3689 * of all keys using the given reducer to combine values, and
3690 * the given basis as an identity 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
3696 * @param basis the identity (initial default value) for the reduction
3697 * @param reducer a commutative associative combining function
3698 * @return the result of accumulating the given transformation
3699 * of all keys
3700 * @since 1.8
3701 */
3702 public double reduceKeysToDouble(long parallelismThreshold,
3703 ObjectToDouble<? super K> transformer,
3704 double basis,
3705 DoubleByDoubleToDouble reducer) {
3706 if (transformer == null || reducer == null)
3707 throw new NullPointerException();
3708 return new MapReduceKeysToDoubleTask<K,V>
3709 (null, batchFor(parallelismThreshold), 0, 0, table,
3710 null, transformer, basis, reducer).invoke();
3711 }
3712
3713 /**
3714 * Returns the result of accumulating the given transformation
3715 * of all keys using the given reducer to combine values, and
3716 * the given basis as an identity value.
3717 *
3718 * @param parallelismThreshold the (estimated) number of elements
3719 * needed for this operation to be executed in parallel
3720 * @param transformer a function returning the transformation
3721 * for an element
3722 * @param basis the identity (initial default value) for the reduction
3723 * @param reducer a commutative associative combining function
3724 * @return the result of accumulating the given transformation
3725 * of all keys
3726 * @since 1.8
3727 */
3728 public long reduceKeysToLong(long parallelismThreshold,
3729 ObjectToLong<? super K> transformer,
3730 long basis,
3731 LongByLongToLong reducer) {
3732 if (transformer == null || reducer == null)
3733 throw new NullPointerException();
3734 return new MapReduceKeysToLongTask<K,V>
3735 (null, batchFor(parallelismThreshold), 0, 0, table,
3736 null, transformer, basis, reducer).invoke();
3737 }
3738
3739 /**
3740 * Returns the result of accumulating the given transformation
3741 * of all keys using the given reducer to combine values, and
3742 * the given basis as an identity value.
3743 *
3744 * @param parallelismThreshold the (estimated) number of elements
3745 * needed for this operation to be executed in parallel
3746 * @param transformer a function returning the transformation
3747 * for an element
3748 * @param basis the identity (initial default value) for the reduction
3749 * @param reducer a commutative associative combining function
3750 * @return the result of accumulating the given transformation
3751 * of all keys
3752 * @since 1.8
3753 */
3754 public int reduceKeysToInt(long parallelismThreshold,
3755 ObjectToInt<? super K> transformer,
3756 int basis,
3757 IntByIntToInt reducer) {
3758 if (transformer == null || reducer == null)
3759 throw new NullPointerException();
3760 return new MapReduceKeysToIntTask<K,V>
3761 (null, batchFor(parallelismThreshold), 0, 0, table,
3762 null, transformer, basis, reducer).invoke();
3763 }
3764
3765 /**
3766 * Performs the given action for each value.
3767 *
3768 * @param parallelismThreshold the (estimated) number of elements
3769 * needed for this operation to be executed in parallel
3770 * @param action the action
3771 * @since 1.8
3772 */
3773 public void forEachValue(long parallelismThreshold,
3774 Action<? super V> action) {
3775 if (action == null)
3776 throw new NullPointerException();
3777 new ForEachValueTask<K,V>
3778 (null, batchFor(parallelismThreshold), 0, 0, table,
3779 action).invoke();
3780 }
3781
3782 /**
3783 * Performs the given action for each non-null transformation
3784 * of each value.
3785 *
3786 * @param parallelismThreshold the (estimated) number of elements
3787 * needed for this operation to be executed in parallel
3788 * @param transformer a function returning the transformation
3789 * for an element, or null if there is no transformation (in
3790 * which case the action is not applied)
3791 * @param action the action
3792 * @since 1.8
3793 */
3794 public <U> void forEachValue(long parallelismThreshold,
3795 Fun<? super V, ? extends U> transformer,
3796 Action<? super U> action) {
3797 if (transformer == null || action == null)
3798 throw new NullPointerException();
3799 new ForEachTransformedValueTask<K,V,U>
3800 (null, batchFor(parallelismThreshold), 0, 0, table,
3801 transformer, action).invoke();
3802 }
3803
3804 /**
3805 * Returns a non-null result from applying the given search
3806 * function on each value, or null if none. Upon success,
3807 * further element processing is suppressed and the results of
3808 * any other parallel invocations of the search function are
3809 * ignored.
3810 *
3811 * @param parallelismThreshold the (estimated) number of elements
3812 * needed for this operation to be executed in parallel
3813 * @param searchFunction a function returning a non-null
3814 * result on success, else null
3815 * @return a non-null result from applying the given search
3816 * function on each value, or null if none
3817 * @since 1.8
3818 */
3819 public <U> U searchValues(long parallelismThreshold,
3820 Fun<? super V, ? extends U> searchFunction) {
3821 if (searchFunction == null) throw new NullPointerException();
3822 return new SearchValuesTask<K,V,U>
3823 (null, batchFor(parallelismThreshold), 0, 0, table,
3824 searchFunction, new AtomicReference<U>()).invoke();
3825 }
3826
3827 /**
3828 * Returns the result of accumulating all values using the
3829 * given reducer to combine values, or null if none.
3830 *
3831 * @param parallelismThreshold the (estimated) number of elements
3832 * needed for this operation to be executed in parallel
3833 * @param reducer a commutative associative combining function
3834 * @return the result of accumulating all values
3835 * @since 1.8
3836 */
3837 public V reduceValues(long parallelismThreshold,
3838 BiFun<? super V, ? super V, ? extends V> reducer) {
3839 if (reducer == null) throw new NullPointerException();
3840 return new ReduceValuesTask<K,V>
3841 (null, batchFor(parallelismThreshold), 0, 0, table,
3842 null, reducer).invoke();
3843 }
3844
3845 /**
3846 * Returns the result of accumulating the given transformation
3847 * of all values using the given reducer to combine values, or
3848 * null if none.
3849 *
3850 * @param parallelismThreshold the (estimated) number of elements
3851 * needed for this operation to be executed in parallel
3852 * @param transformer a function returning the transformation
3853 * for an element, or null if there is no transformation (in
3854 * which case it is not combined)
3855 * @param reducer a commutative associative combining function
3856 * @return the result of accumulating the given transformation
3857 * of all values
3858 * @since 1.8
3859 */
3860 public <U> U reduceValues(long parallelismThreshold,
3861 Fun<? super V, ? extends U> transformer,
3862 BiFun<? super U, ? super U, ? extends U> reducer) {
3863 if (transformer == null || reducer == null)
3864 throw new NullPointerException();
3865 return new MapReduceValuesTask<K,V,U>
3866 (null, batchFor(parallelismThreshold), 0, 0, table,
3867 null, transformer, reducer).invoke();
3868 }
3869
3870 /**
3871 * Returns the result of accumulating the given transformation
3872 * of all values using the given reducer to combine values,
3873 * and the given basis as an identity value.
3874 *
3875 * @param parallelismThreshold the (estimated) number of elements
3876 * needed for this operation to be executed in parallel
3877 * @param transformer a function returning the transformation
3878 * for an element
3879 * @param basis the identity (initial default value) for the reduction
3880 * @param reducer a commutative associative combining function
3881 * @return the result of accumulating the given transformation
3882 * of all values
3883 * @since 1.8
3884 */
3885 public double reduceValuesToDouble(long parallelismThreshold,
3886 ObjectToDouble<? super V> transformer,
3887 double basis,
3888 DoubleByDoubleToDouble reducer) {
3889 if (transformer == null || reducer == null)
3890 throw new NullPointerException();
3891 return new MapReduceValuesToDoubleTask<K,V>
3892 (null, batchFor(parallelismThreshold), 0, 0, table,
3893 null, transformer, basis, reducer).invoke();
3894 }
3895
3896 /**
3897 * Returns the result of accumulating the given transformation
3898 * of all values using the given reducer to combine values,
3899 * and the given basis as an identity value.
3900 *
3901 * @param parallelismThreshold the (estimated) number of elements
3902 * needed for this operation to be executed in parallel
3903 * @param transformer a function returning the transformation
3904 * for an element
3905 * @param basis the identity (initial default value) for the reduction
3906 * @param reducer a commutative associative combining function
3907 * @return the result of accumulating the given transformation
3908 * of all values
3909 * @since 1.8
3910 */
3911 public long reduceValuesToLong(long parallelismThreshold,
3912 ObjectToLong<? super V> transformer,
3913 long basis,
3914 LongByLongToLong reducer) {
3915 if (transformer == null || reducer == null)
3916 throw new NullPointerException();
3917 return new MapReduceValuesToLongTask<K,V>
3918 (null, batchFor(parallelismThreshold), 0, 0, table,
3919 null, transformer, basis, reducer).invoke();
3920 }
3921
3922 /**
3923 * Returns the result of accumulating the given transformation
3924 * of all values using the given reducer to combine values,
3925 * and the given basis as an identity value.
3926 *
3927 * @param parallelismThreshold the (estimated) number of elements
3928 * needed for this operation to be executed in parallel
3929 * @param transformer a function returning the transformation
3930 * for an element
3931 * @param basis the identity (initial default value) for the reduction
3932 * @param reducer a commutative associative combining function
3933 * @return the result of accumulating the given transformation
3934 * of all values
3935 * @since 1.8
3936 */
3937 public int reduceValuesToInt(long parallelismThreshold,
3938 ObjectToInt<? super V> transformer,
3939 int basis,
3940 IntByIntToInt reducer) {
3941 if (transformer == null || reducer == null)
3942 throw new NullPointerException();
3943 return new MapReduceValuesToIntTask<K,V>
3944 (null, batchFor(parallelismThreshold), 0, 0, table,
3945 null, transformer, basis, reducer).invoke();
3946 }
3947
3948 /**
3949 * Performs the given action for each entry.
3950 *
3951 * @param parallelismThreshold the (estimated) number of elements
3952 * needed for this operation to be executed in parallel
3953 * @param action the action
3954 * @since 1.8
3955 */
3956 public void forEachEntry(long parallelismThreshold,
3957 Action<? super Map.Entry<K,V>> action) {
3958 if (action == null) throw new NullPointerException();
3959 new ForEachEntryTask<K,V>(null, batchFor(parallelismThreshold), 0, 0, table,
3960 action).invoke();
3961 }
3962
3963 /**
3964 * Performs the given action for each non-null transformation
3965 * of each entry.
3966 *
3967 * @param parallelismThreshold the (estimated) number of elements
3968 * needed for this operation to be executed in parallel
3969 * @param transformer a function returning the transformation
3970 * for an element, or null if there is no transformation (in
3971 * which case the action is not applied)
3972 * @param action the action
3973 * @since 1.8
3974 */
3975 public <U> void forEachEntry(long parallelismThreshold,
3976 Fun<Map.Entry<K,V>, ? extends U> transformer,
3977 Action<? super U> action) {
3978 if (transformer == null || action == null)
3979 throw new NullPointerException();
3980 new ForEachTransformedEntryTask<K,V,U>
3981 (null, batchFor(parallelismThreshold), 0, 0, table,
3982 transformer, action).invoke();
3983 }
3984
3985 /**
3986 * Returns a non-null result from applying the given search
3987 * function on each entry, or null if none. Upon success,
3988 * further element processing is suppressed and the results of
3989 * any other parallel invocations of the search function are
3990 * ignored.
3991 *
3992 * @param parallelismThreshold the (estimated) number of elements
3993 * needed for this operation to be executed in parallel
3994 * @param searchFunction a function returning a non-null
3995 * result on success, else null
3996 * @return a non-null result from applying the given search
3997 * function on each entry, or null if none
3998 * @since 1.8
3999 */
4000 public <U> U searchEntries(long parallelismThreshold,
4001 Fun<Map.Entry<K,V>, ? extends U> searchFunction) {
4002 if (searchFunction == null) throw new NullPointerException();
4003 return new SearchEntriesTask<K,V,U>
4004 (null, batchFor(parallelismThreshold), 0, 0, table,
4005 searchFunction, new AtomicReference<U>()).invoke();
4006 }
4007
4008 /**
4009 * Returns the result of accumulating all entries using the
4010 * given reducer to combine values, or null if none.
4011 *
4012 * @param parallelismThreshold the (estimated) number of elements
4013 * needed for this operation to be executed in parallel
4014 * @param reducer a commutative associative combining function
4015 * @return the result of accumulating all entries
4016 * @since 1.8
4017 */
4018 public Map.Entry<K,V> reduceEntries(long parallelismThreshold,
4019 BiFun<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer) {
4020 if (reducer == null) throw new NullPointerException();
4021 return new ReduceEntriesTask<K,V>
4022 (null, batchFor(parallelismThreshold), 0, 0, table,
4023 null, reducer).invoke();
4024 }
4025
4026 /**
4027 * Returns the result of accumulating the given transformation
4028 * of all entries using the given reducer to combine values,
4029 * or null if none.
4030 *
4031 * @param parallelismThreshold the (estimated) number of elements
4032 * needed for this operation to be executed in parallel
4033 * @param transformer a function returning the transformation
4034 * for an element, or null if there is no transformation (in
4035 * which case it is not combined)
4036 * @param reducer a commutative associative combining function
4037 * @return the result of accumulating the given transformation
4038 * of all entries
4039 * @since 1.8
4040 */
4041 public <U> U reduceEntries(long parallelismThreshold,
4042 Fun<Map.Entry<K,V>, ? extends U> transformer,
4043 BiFun<? super U, ? super U, ? extends U> reducer) {
4044 if (transformer == null || reducer == null)
4045 throw new NullPointerException();
4046 return new MapReduceEntriesTask<K,V,U>
4047 (null, batchFor(parallelismThreshold), 0, 0, table,
4048 null, transformer, reducer).invoke();
4049 }
4050
4051 /**
4052 * Returns the result of accumulating the given transformation
4053 * of all entries using the given reducer to combine values,
4054 * and the given basis as an identity value.
4055 *
4056 * @param parallelismThreshold the (estimated) number of elements
4057 * needed for this operation to be executed in parallel
4058 * @param transformer a function returning the transformation
4059 * for an element
4060 * @param basis the identity (initial default value) for the reduction
4061 * @param reducer a commutative associative combining function
4062 * @return the result of accumulating the given transformation
4063 * of all entries
4064 * @since 1.8
4065 */
4066 public double reduceEntriesToDouble(long parallelismThreshold,
4067 ObjectToDouble<Map.Entry<K,V>> transformer,
4068 double basis,
4069 DoubleByDoubleToDouble reducer) {
4070 if (transformer == null || reducer == null)
4071 throw new NullPointerException();
4072 return new MapReduceEntriesToDoubleTask<K,V>
4073 (null, batchFor(parallelismThreshold), 0, 0, table,
4074 null, transformer, basis, reducer).invoke();
4075 }
4076
4077 /**
4078 * Returns the result of accumulating the given transformation
4079 * of all entries using the given reducer to combine values,
4080 * and the given basis as an identity value.
4081 *
4082 * @param parallelismThreshold the (estimated) number of elements
4083 * needed for this operation to be executed in parallel
4084 * @param transformer a function returning the transformation
4085 * for an element
4086 * @param basis the identity (initial default value) for the reduction
4087 * @param reducer a commutative associative combining function
4088 * @return the result of accumulating the given transformation
4089 * of all entries
4090 * @since 1.8
4091 */
4092 public long reduceEntriesToLong(long parallelismThreshold,
4093 ObjectToLong<Map.Entry<K,V>> transformer,
4094 long basis,
4095 LongByLongToLong reducer) {
4096 if (transformer == null || reducer == null)
4097 throw new NullPointerException();
4098 return new MapReduceEntriesToLongTask<K,V>
4099 (null, batchFor(parallelismThreshold), 0, 0, table,
4100 null, transformer, basis, reducer).invoke();
4101 }
4102
4103 /**
4104 * Returns the result of accumulating the given transformation
4105 * of all entries using the given reducer to combine values,
4106 * and the given basis as an identity value.
4107 *
4108 * @param parallelismThreshold the (estimated) number of elements
4109 * needed for this operation to be executed in parallel
4110 * @param transformer a function returning the transformation
4111 * for an element
4112 * @param basis the identity (initial default value) for the reduction
4113 * @param reducer a commutative associative combining function
4114 * @return the result of accumulating the given transformation
4115 * of all entries
4116 * @since 1.8
4117 */
4118 public int reduceEntriesToInt(long parallelismThreshold,
4119 ObjectToInt<Map.Entry<K,V>> transformer,
4120 int basis,
4121 IntByIntToInt reducer) {
4122 if (transformer == null || reducer == null)
4123 throw new NullPointerException();
4124 return new MapReduceEntriesToIntTask<K,V>
4125 (null, batchFor(parallelismThreshold), 0, 0, table,
4126 null, transformer, basis, reducer).invoke();
4127 }
4128
4129
4130 /* ----------------Views -------------- */
4131
4132 /**
4133 * Base class for views.
4134 */
4135 abstract static class CollectionView<K,V,E>
4136 implements Collection<E>, java.io.Serializable {
4137 private static final long serialVersionUID = 7249069246763182397L;
4138 final ConcurrentHashMapV8<K,V> map;
4139 CollectionView(ConcurrentHashMapV8<K,V> map) { this.map = map; }
4140
4141 /**
4142 * Returns the map backing this view.
4143 *
4144 * @return the map backing this view
4145 */
4146 public ConcurrentHashMapV8<K,V> getMap() { return map; }
4147
4148 /**
4149 * Removes all of the elements from this view, by removing all
4150 * the mappings from the map backing this view.
4151 */
4152 public final void clear() { map.clear(); }
4153 public final int size() { return map.size(); }
4154 public final boolean isEmpty() { return map.isEmpty(); }
4155
4156 // implementations below rely on concrete classes supplying these
4157 // abstract methods
4158 /**
4159 * Returns a "weakly consistent" iterator that will never
4160 * throw {@link ConcurrentModificationException}, and
4161 * guarantees to traverse elements as they existed upon
4162 * construction of the iterator, and may (but is not
4163 * guaranteed to) reflect any modifications subsequent to
4164 * construction.
4165 */
4166 public abstract Iterator<E> iterator();
4167 public abstract boolean contains(Object o);
4168 public abstract boolean remove(Object o);
4169
4170 private static final String oomeMsg = "Required array size too large";
4171
4172 public final Object[] toArray() {
4173 long sz = map.mappingCount();
4174 if (sz > MAX_ARRAY_SIZE)
4175 throw new OutOfMemoryError(oomeMsg);
4176 int n = (int)sz;
4177 Object[] r = new Object[n];
4178 int i = 0;
4179 for (E e : this) {
4180 if (i == n) {
4181 if (n >= MAX_ARRAY_SIZE)
4182 throw new OutOfMemoryError(oomeMsg);
4183 if (n >= MAX_ARRAY_SIZE - (MAX_ARRAY_SIZE >>> 1) - 1)
4184 n = MAX_ARRAY_SIZE;
4185 else
4186 n += (n >>> 1) + 1;
4187 r = Arrays.copyOf(r, n);
4188 }
4189 r[i++] = e;
4190 }
4191 return (i == n) ? r : Arrays.copyOf(r, i);
4192 }
4193
4194 @SuppressWarnings("unchecked")
4195 public final <T> T[] toArray(T[] a) {
4196 long sz = map.mappingCount();
4197 if (sz > MAX_ARRAY_SIZE)
4198 throw new OutOfMemoryError(oomeMsg);
4199 int m = (int)sz;
4200 T[] r = (a.length >= m) ? a :
4201 (T[])java.lang.reflect.Array
4202 .newInstance(a.getClass().getComponentType(), m);
4203 int n = r.length;
4204 int i = 0;
4205 for (E e : this) {
4206 if (i == n) {
4207 if (n >= MAX_ARRAY_SIZE)
4208 throw new OutOfMemoryError(oomeMsg);
4209 if (n >= MAX_ARRAY_SIZE - (MAX_ARRAY_SIZE >>> 1) - 1)
4210 n = MAX_ARRAY_SIZE;
4211 else
4212 n += (n >>> 1) + 1;
4213 r = Arrays.copyOf(r, n);
4214 }
4215 r[i++] = (T)e;
4216 }
4217 if (a == r && i < n) {
4218 r[i] = null; // null-terminate
4219 return r;
4220 }
4221 return (i == n) ? r : Arrays.copyOf(r, i);
4222 }
4223
4224 /**
4225 * Returns a string representation of this collection.
4226 * The string representation consists of the string representations
4227 * of the collection's elements in the order they are returned by
4228 * its iterator, enclosed in square brackets ({@code "[]"}).
4229 * Adjacent elements are separated by the characters {@code ", "}
4230 * (comma and space). Elements are converted to strings as by
4231 * {@link String#valueOf(Object)}.
4232 *
4233 * @return a string representation of this collection
4234 */
4235 public final String toString() {
4236 StringBuilder sb = new StringBuilder();
4237 sb.append('[');
4238 Iterator<E> it = iterator();
4239 if (it.hasNext()) {
4240 for (;;) {
4241 Object e = it.next();
4242 sb.append(e == this ? "(this Collection)" : e);
4243 if (!it.hasNext())
4244 break;
4245 sb.append(',').append(' ');
4246 }
4247 }
4248 return sb.append(']').toString();
4249 }
4250
4251 public final boolean containsAll(Collection<?> c) {
4252 if (c != this) {
4253 for (Object e : c) {
4254 if (e == null || !contains(e))
4255 return false;
4256 }
4257 }
4258 return true;
4259 }
4260
4261 public final boolean removeAll(Collection<?> c) {
4262 boolean modified = false;
4263 for (Iterator<E> it = iterator(); it.hasNext();) {
4264 if (c.contains(it.next())) {
4265 it.remove();
4266 modified = true;
4267 }
4268 }
4269 return modified;
4270 }
4271
4272 public final boolean retainAll(Collection<?> c) {
4273 boolean modified = false;
4274 for (Iterator<E> it = iterator(); it.hasNext();) {
4275 if (!c.contains(it.next())) {
4276 it.remove();
4277 modified = true;
4278 }
4279 }
4280 return modified;
4281 }
4282
4283 }
4284
4285 /**
4286 * A view of a ConcurrentHashMapV8 as a {@link Set} of keys, in
4287 * which additions may optionally be enabled by mapping to a
4288 * common value. This class cannot be directly instantiated.
4289 * See {@link #keySet() keySet()},
4290 * {@link #keySet(Object) keySet(V)},
4291 * {@link #newKeySet() newKeySet()},
4292 * {@link #newKeySet(int) newKeySet(int)}.
4293 *
4294 * @since 1.8
4295 */
4296 public static class KeySetView<K,V> extends CollectionView<K,V,K>
4297 implements Set<K>, java.io.Serializable {
4298 private static final long serialVersionUID = 7249069246763182397L;
4299 private final V value;
4300 KeySetView(ConcurrentHashMapV8<K,V> map, V value) { // non-public
4301 super(map);
4302 this.value = value;
4303 }
4304
4305 /**
4306 * Returns the default mapped value for additions,
4307 * or {@code null} if additions are not supported.
4308 *
4309 * @return the default mapped value for additions, or {@code null}
4310 * if not supported
4311 */
4312 public V getMappedValue() { return value; }
4313
4314 /**
4315 * {@inheritDoc}
4316 * @throws NullPointerException if the specified key is null
4317 */
4318 public boolean contains(Object o) { return map.containsKey(o); }
4319
4320 /**
4321 * Removes the key from this map view, by removing the key (and its
4322 * corresponding value) from the backing map. This method does
4323 * nothing if the key is not in the map.
4324 *
4325 * @param o the key to be removed from the backing map
4326 * @return {@code true} if the backing map contained the specified key
4327 * @throws NullPointerException if the specified key is null
4328 */
4329 public boolean remove(Object o) { return map.remove(o) != null; }
4330
4331 /**
4332 * @return an iterator over the keys of the backing map
4333 */
4334 public Iterator<K> iterator() {
4335 Node<K,V>[] t;
4336 ConcurrentHashMapV8<K,V> m = map;
4337 int f = (t = m.table) == null ? 0 : t.length;
4338 return new KeyIterator<K,V>(t, f, 0, f, m);
4339 }
4340
4341 /**
4342 * Adds the specified key to this set view by mapping the key to
4343 * the default mapped value in the backing map, if defined.
4344 *
4345 * @param e key to be added
4346 * @return {@code true} if this set changed as a result of the call
4347 * @throws NullPointerException if the specified key is null
4348 * @throws UnsupportedOperationException if no default mapped value
4349 * for additions was provided
4350 */
4351 public boolean add(K e) {
4352 V v;
4353 if ((v = value) == null)
4354 throw new UnsupportedOperationException();
4355 return map.putVal(e, v, true) == null;
4356 }
4357
4358 /**
4359 * Adds all of the elements in the specified collection to this set,
4360 * as if by calling {@link #add} on each one.
4361 *
4362 * @param c the elements to be inserted into this set
4363 * @return {@code true} if this set changed as a result of the call
4364 * @throws NullPointerException if the collection or any of its
4365 * elements are {@code null}
4366 * @throws UnsupportedOperationException if no default mapped value
4367 * for additions was provided
4368 */
4369 public boolean addAll(Collection<? extends K> c) {
4370 boolean added = false;
4371 V v;
4372 if ((v = value) == null)
4373 throw new UnsupportedOperationException();
4374 for (K e : c) {
4375 if (map.putVal(e, v, true) == null)
4376 added = true;
4377 }
4378 return added;
4379 }
4380
4381 public int hashCode() {
4382 int h = 0;
4383 for (K e : this)
4384 h += e.hashCode();
4385 return h;
4386 }
4387
4388 public boolean equals(Object o) {
4389 Set<?> c;
4390 return ((o instanceof Set) &&
4391 ((c = (Set<?>)o) == this ||
4392 (containsAll(c) && c.containsAll(this))));
4393 }
4394
4395 public ConcurrentHashMapSpliterator<K> spliterator() {
4396 Node<K,V>[] t;
4397 ConcurrentHashMapV8<K,V> m = map;
4398 long n = m.sumCount();
4399 int f = (t = m.table) == null ? 0 : t.length;
4400 return new KeySpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n);
4401 }
4402
4403 public void forEach(Action<? super K> action) {
4404 if (action == null) throw new NullPointerException();
4405 Node<K,V>[] t;
4406 if ((t = map.table) != null) {
4407 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4408 for (Node<K,V> p; (p = it.advance()) != null; )
4409 action.apply(p.key);
4410 }
4411 }
4412 }
4413
4414 /**
4415 * A view of a ConcurrentHashMapV8 as a {@link Collection} of
4416 * values, in which additions are disabled. This class cannot be
4417 * directly instantiated. See {@link #values()}.
4418 */
4419 static final class ValuesView<K,V> extends CollectionView<K,V,V>
4420 implements Collection<V>, java.io.Serializable {
4421 private static final long serialVersionUID = 2249069246763182397L;
4422 ValuesView(ConcurrentHashMapV8<K,V> map) { super(map); }
4423 public final boolean contains(Object o) {
4424 return map.containsValue(o);
4425 }
4426
4427 public final boolean remove(Object o) {
4428 if (o != null) {
4429 for (Iterator<V> it = iterator(); it.hasNext();) {
4430 if (o.equals(it.next())) {
4431 it.remove();
4432 return true;
4433 }
4434 }
4435 }
4436 return false;
4437 }
4438
4439 public final Iterator<V> iterator() {
4440 ConcurrentHashMapV8<K,V> m = map;
4441 Node<K,V>[] t;
4442 int f = (t = m.table) == null ? 0 : t.length;
4443 return new ValueIterator<K,V>(t, f, 0, f, m);
4444 }
4445
4446 public final boolean add(V e) {
4447 throw new UnsupportedOperationException();
4448 }
4449 public final boolean addAll(Collection<? extends V> c) {
4450 throw new UnsupportedOperationException();
4451 }
4452
4453 public ConcurrentHashMapSpliterator<V> spliterator() {
4454 Node<K,V>[] t;
4455 ConcurrentHashMapV8<K,V> m = map;
4456 long n = m.sumCount();
4457 int f = (t = m.table) == null ? 0 : t.length;
4458 return new ValueSpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n);
4459 }
4460
4461 public void forEach(Action<? super V> action) {
4462 if (action == null) throw new NullPointerException();
4463 Node<K,V>[] t;
4464 if ((t = map.table) != null) {
4465 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4466 for (Node<K,V> p; (p = it.advance()) != null; )
4467 action.apply(p.val);
4468 }
4469 }
4470 }
4471
4472 /**
4473 * A view of a ConcurrentHashMapV8 as a {@link Set} of (key, value)
4474 * entries. This class cannot be directly instantiated. See
4475 * {@link #entrySet()}.
4476 */
4477 static final class EntrySetView<K,V> extends CollectionView<K,V,Map.Entry<K,V>>
4478 implements Set<Map.Entry<K,V>>, java.io.Serializable {
4479 private static final long serialVersionUID = 2249069246763182397L;
4480 EntrySetView(ConcurrentHashMapV8<K,V> map) { super(map); }
4481
4482 public boolean contains(Object o) {
4483 Object k, v, r; Map.Entry<?,?> e;
4484 return ((o instanceof Map.Entry) &&
4485 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
4486 (r = map.get(k)) != null &&
4487 (v = e.getValue()) != null &&
4488 (v == r || v.equals(r)));
4489 }
4490
4491 public boolean remove(Object o) {
4492 Object k, v; Map.Entry<?,?> e;
4493 return ((o instanceof Map.Entry) &&
4494 (k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
4495 (v = e.getValue()) != null &&
4496 map.remove(k, v));
4497 }
4498
4499 /**
4500 * @return an iterator over the entries of the backing map
4501 */
4502 public Iterator<Map.Entry<K,V>> iterator() {
4503 ConcurrentHashMapV8<K,V> m = map;
4504 Node<K,V>[] t;
4505 int f = (t = m.table) == null ? 0 : t.length;
4506 return new EntryIterator<K,V>(t, f, 0, f, m);
4507 }
4508
4509 public boolean add(Entry<K,V> e) {
4510 return map.putVal(e.getKey(), e.getValue(), false) == null;
4511 }
4512
4513 public boolean addAll(Collection<? extends Entry<K,V>> c) {
4514 boolean added = false;
4515 for (Entry<K,V> e : c) {
4516 if (add(e))
4517 added = true;
4518 }
4519 return added;
4520 }
4521
4522 public final int hashCode() {
4523 int h = 0;
4524 Node<K,V>[] t;
4525 if ((t = map.table) != null) {
4526 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4527 for (Node<K,V> p; (p = it.advance()) != null; ) {
4528 h += p.hashCode();
4529 }
4530 }
4531 return h;
4532 }
4533
4534 public final boolean equals(Object o) {
4535 Set<?> c;
4536 return ((o instanceof Set) &&
4537 ((c = (Set<?>)o) == this ||
4538 (containsAll(c) && c.containsAll(this))));
4539 }
4540
4541 public ConcurrentHashMapSpliterator<Map.Entry<K,V>> spliterator() {
4542 Node<K,V>[] t;
4543 ConcurrentHashMapV8<K,V> m = map;
4544 long n = m.sumCount();
4545 int f = (t = m.table) == null ? 0 : t.length;
4546 return new EntrySpliterator<K,V>(t, f, 0, f, n < 0L ? 0L : n, m);
4547 }
4548
4549 public void forEach(Action<? super Map.Entry<K,V>> action) {
4550 if (action == null) throw new NullPointerException();
4551 Node<K,V>[] t;
4552 if ((t = map.table) != null) {
4553 Traverser<K,V> it = new Traverser<K,V>(t, t.length, 0, t.length);
4554 for (Node<K,V> p; (p = it.advance()) != null; )
4555 action.apply(new MapEntry<K,V>(p.key, p.val, map));
4556 }
4557 }
4558
4559 }
4560
4561 // -------------------------------------------------------
4562
4563 /**
4564 * Base class for bulk tasks. Repeats some fields and code from
4565 * class Traverser, because we need to subclass CountedCompleter.
4566 */
4567 abstract static class BulkTask<K,V,R> extends CountedCompleter<R> {
4568 Node<K,V>[] tab; // same as Traverser
4569 Node<K,V> next;
4570 int index;
4571 int baseIndex;
4572 int baseLimit;
4573 final int baseSize;
4574 int batch; // split control
4575
4576 BulkTask(BulkTask<K,V,?> par, int b, int i, int f, Node<K,V>[] t) {
4577 super(par);
4578 this.batch = b;
4579 this.index = this.baseIndex = i;
4580 if ((this.tab = t) == null)
4581 this.baseSize = this.baseLimit = 0;
4582 else if (par == null)
4583 this.baseSize = this.baseLimit = t.length;
4584 else {
4585 this.baseLimit = f;
4586 this.baseSize = par.baseSize;
4587 }
4588 }
4589
4590 /**
4591 * Same as Traverser version
4592 */
4593 final Node<K,V> advance() {
4594 Node<K,V> e;
4595 if ((e = next) != null)
4596 e = e.next;
4597 for (;;) {
4598 Node<K,V>[] t; int i, n; K ek; // must use locals in checks
4599 if (e != null)
4600 return next = e;
4601 if (baseIndex >= baseLimit || (t = tab) == null ||
4602 (n = t.length) <= (i = index) || i < 0)
4603 return next = null;
4604 if ((e = tabAt(t, index)) != null && e.hash < 0) {
4605 if (e instanceof ForwardingNode) {
4606 tab = ((ForwardingNode<K,V>)e).nextTable;
4607 e = null;
4608 continue;
4609 }
4610 else if (e instanceof TreeBin)
4611 e = ((TreeBin<K,V>)e).first;
4612 else
4613 e = null;
4614 }
4615 if ((index += baseSize) >= n)
4616 index = ++baseIndex; // visit upper slots if present
4617 }
4618 }
4619 }
4620
4621 /*
4622 * Task classes. Coded in a regular but ugly format/style to
4623 * simplify checks that each variant differs in the right way from
4624 * others. The null screenings exist because compilers cannot tell
4625 * that we've already null-checked task arguments, so we force
4626 * simplest hoisted bypass to help avoid convoluted traps.
4627 */
4628 @SuppressWarnings("serial")
4629 static final class ForEachKeyTask<K,V>
4630 extends BulkTask<K,V,Void> {
4631 final Action<? super K> action;
4632 ForEachKeyTask
4633 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4634 Action<? super K> action) {
4635 super(p, b, i, f, t);
4636 this.action = action;
4637 }
4638 public final void compute() {
4639 final Action<? super K> action;
4640 if ((action = this.action) != null) {
4641 for (int i = baseIndex, f, h; batch > 0 &&
4642 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4643 addToPendingCount(1);
4644 new ForEachKeyTask<K,V>
4645 (this, batch >>>= 1, baseLimit = h, f, tab,
4646 action).fork();
4647 }
4648 for (Node<K,V> p; (p = advance()) != null;)
4649 action.apply(p.key);
4650 propagateCompletion();
4651 }
4652 }
4653 }
4654
4655 @SuppressWarnings("serial")
4656 static final class ForEachValueTask<K,V>
4657 extends BulkTask<K,V,Void> {
4658 final Action<? super V> action;
4659 ForEachValueTask
4660 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4661 Action<? super V> action) {
4662 super(p, b, i, f, t);
4663 this.action = action;
4664 }
4665 public final void compute() {
4666 final Action<? super V> action;
4667 if ((action = this.action) != null) {
4668 for (int i = baseIndex, f, h; batch > 0 &&
4669 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4670 addToPendingCount(1);
4671 new ForEachValueTask<K,V>
4672 (this, batch >>>= 1, baseLimit = h, f, tab,
4673 action).fork();
4674 }
4675 for (Node<K,V> p; (p = advance()) != null;)
4676 action.apply(p.val);
4677 propagateCompletion();
4678 }
4679 }
4680 }
4681
4682 @SuppressWarnings("serial")
4683 static final class ForEachEntryTask<K,V>
4684 extends BulkTask<K,V,Void> {
4685 final Action<? super Entry<K,V>> action;
4686 ForEachEntryTask
4687 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4688 Action<? super Entry<K,V>> action) {
4689 super(p, b, i, f, t);
4690 this.action = action;
4691 }
4692 public final void compute() {
4693 final Action<? super Entry<K,V>> action;
4694 if ((action = this.action) != null) {
4695 for (int i = baseIndex, f, h; batch > 0 &&
4696 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4697 addToPendingCount(1);
4698 new ForEachEntryTask<K,V>
4699 (this, batch >>>= 1, baseLimit = h, f, tab,
4700 action).fork();
4701 }
4702 for (Node<K,V> p; (p = advance()) != null; )
4703 action.apply(p);
4704 propagateCompletion();
4705 }
4706 }
4707 }
4708
4709 @SuppressWarnings("serial")
4710 static final class ForEachMappingTask<K,V>
4711 extends BulkTask<K,V,Void> {
4712 final BiAction<? super K, ? super V> action;
4713 ForEachMappingTask
4714 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4715 BiAction<? super K,? super V> action) {
4716 super(p, b, i, f, t);
4717 this.action = action;
4718 }
4719 public final void compute() {
4720 final BiAction<? super K, ? super V> action;
4721 if ((action = this.action) != null) {
4722 for (int i = baseIndex, f, h; batch > 0 &&
4723 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4724 addToPendingCount(1);
4725 new ForEachMappingTask<K,V>
4726 (this, batch >>>= 1, baseLimit = h, f, tab,
4727 action).fork();
4728 }
4729 for (Node<K,V> p; (p = advance()) != null; )
4730 action.apply(p.key, p.val);
4731 propagateCompletion();
4732 }
4733 }
4734 }
4735
4736 @SuppressWarnings("serial")
4737 static final class ForEachTransformedKeyTask<K,V,U>
4738 extends BulkTask<K,V,Void> {
4739 final Fun<? super K, ? extends U> transformer;
4740 final Action<? super U> action;
4741 ForEachTransformedKeyTask
4742 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4743 Fun<? super K, ? extends U> transformer, Action<? super U> action) {
4744 super(p, b, i, f, t);
4745 this.transformer = transformer; this.action = action;
4746 }
4747 public final void compute() {
4748 final Fun<? super K, ? extends U> transformer;
4749 final Action<? super U> action;
4750 if ((transformer = this.transformer) != null &&
4751 (action = this.action) != null) {
4752 for (int i = baseIndex, f, h; batch > 0 &&
4753 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4754 addToPendingCount(1);
4755 new ForEachTransformedKeyTask<K,V,U>
4756 (this, batch >>>= 1, baseLimit = h, f, tab,
4757 transformer, action).fork();
4758 }
4759 for (Node<K,V> p; (p = advance()) != null; ) {
4760 U u;
4761 if ((u = transformer.apply(p.key)) != null)
4762 action.apply(u);
4763 }
4764 propagateCompletion();
4765 }
4766 }
4767 }
4768
4769 @SuppressWarnings("serial")
4770 static final class ForEachTransformedValueTask<K,V,U>
4771 extends BulkTask<K,V,Void> {
4772 final Fun<? super V, ? extends U> transformer;
4773 final Action<? super U> action;
4774 ForEachTransformedValueTask
4775 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4776 Fun<? super V, ? extends U> transformer, Action<? super U> action) {
4777 super(p, b, i, f, t);
4778 this.transformer = transformer; this.action = action;
4779 }
4780 public final void compute() {
4781 final Fun<? super V, ? extends U> transformer;
4782 final Action<? super U> action;
4783 if ((transformer = this.transformer) != null &&
4784 (action = this.action) != null) {
4785 for (int i = baseIndex, f, h; batch > 0 &&
4786 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4787 addToPendingCount(1);
4788 new ForEachTransformedValueTask<K,V,U>
4789 (this, batch >>>= 1, baseLimit = h, f, tab,
4790 transformer, action).fork();
4791 }
4792 for (Node<K,V> p; (p = advance()) != null; ) {
4793 U u;
4794 if ((u = transformer.apply(p.val)) != null)
4795 action.apply(u);
4796 }
4797 propagateCompletion();
4798 }
4799 }
4800 }
4801
4802 @SuppressWarnings("serial")
4803 static final class ForEachTransformedEntryTask<K,V,U>
4804 extends BulkTask<K,V,Void> {
4805 final Fun<Map.Entry<K,V>, ? extends U> transformer;
4806 final Action<? super U> action;
4807 ForEachTransformedEntryTask
4808 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4809 Fun<Map.Entry<K,V>, ? extends U> transformer, Action<? super U> action) {
4810 super(p, b, i, f, t);
4811 this.transformer = transformer; this.action = action;
4812 }
4813 public final void compute() {
4814 final Fun<Map.Entry<K,V>, ? extends U> transformer;
4815 final Action<? super U> action;
4816 if ((transformer = this.transformer) != null &&
4817 (action = this.action) != null) {
4818 for (int i = baseIndex, f, h; batch > 0 &&
4819 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4820 addToPendingCount(1);
4821 new ForEachTransformedEntryTask<K,V,U>
4822 (this, batch >>>= 1, baseLimit = h, f, tab,
4823 transformer, action).fork();
4824 }
4825 for (Node<K,V> p; (p = advance()) != null; ) {
4826 U u;
4827 if ((u = transformer.apply(p)) != null)
4828 action.apply(u);
4829 }
4830 propagateCompletion();
4831 }
4832 }
4833 }
4834
4835 @SuppressWarnings("serial")
4836 static final class ForEachTransformedMappingTask<K,V,U>
4837 extends BulkTask<K,V,Void> {
4838 final BiFun<? super K, ? super V, ? extends U> transformer;
4839 final Action<? super U> action;
4840 ForEachTransformedMappingTask
4841 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4842 BiFun<? super K, ? super V, ? extends U> transformer,
4843 Action<? super U> action) {
4844 super(p, b, i, f, t);
4845 this.transformer = transformer; this.action = action;
4846 }
4847 public final void compute() {
4848 final BiFun<? super K, ? super V, ? extends U> transformer;
4849 final Action<? super U> action;
4850 if ((transformer = this.transformer) != null &&
4851 (action = this.action) != null) {
4852 for (int i = baseIndex, f, h; batch > 0 &&
4853 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4854 addToPendingCount(1);
4855 new ForEachTransformedMappingTask<K,V,U>
4856 (this, batch >>>= 1, baseLimit = h, f, tab,
4857 transformer, action).fork();
4858 }
4859 for (Node<K,V> p; (p = advance()) != null; ) {
4860 U u;
4861 if ((u = transformer.apply(p.key, p.val)) != null)
4862 action.apply(u);
4863 }
4864 propagateCompletion();
4865 }
4866 }
4867 }
4868
4869 @SuppressWarnings("serial")
4870 static final class SearchKeysTask<K,V,U>
4871 extends BulkTask<K,V,U> {
4872 final Fun<? super K, ? extends U> searchFunction;
4873 final AtomicReference<U> result;
4874 SearchKeysTask
4875 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4876 Fun<? super K, ? extends U> searchFunction,
4877 AtomicReference<U> result) {
4878 super(p, b, i, f, t);
4879 this.searchFunction = searchFunction; this.result = result;
4880 }
4881 public final U getRawResult() { return result.get(); }
4882 public final void compute() {
4883 final Fun<? super K, ? extends U> searchFunction;
4884 final AtomicReference<U> result;
4885 if ((searchFunction = this.searchFunction) != null &&
4886 (result = this.result) != null) {
4887 for (int i = baseIndex, f, h; batch > 0 &&
4888 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4889 if (result.get() != null)
4890 return;
4891 addToPendingCount(1);
4892 new SearchKeysTask<K,V,U>
4893 (this, batch >>>= 1, baseLimit = h, f, tab,
4894 searchFunction, result).fork();
4895 }
4896 while (result.get() == null) {
4897 U u;
4898 Node<K,V> p;
4899 if ((p = advance()) == null) {
4900 propagateCompletion();
4901 break;
4902 }
4903 if ((u = searchFunction.apply(p.key)) != null) {
4904 if (result.compareAndSet(null, u))
4905 quietlyCompleteRoot();
4906 break;
4907 }
4908 }
4909 }
4910 }
4911 }
4912
4913 @SuppressWarnings("serial")
4914 static final class SearchValuesTask<K,V,U>
4915 extends BulkTask<K,V,U> {
4916 final Fun<? super V, ? extends U> searchFunction;
4917 final AtomicReference<U> result;
4918 SearchValuesTask
4919 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4920 Fun<? super V, ? extends U> searchFunction,
4921 AtomicReference<U> result) {
4922 super(p, b, i, f, t);
4923 this.searchFunction = searchFunction; this.result = result;
4924 }
4925 public final U getRawResult() { return result.get(); }
4926 public final void compute() {
4927 final Fun<? super V, ? extends U> searchFunction;
4928 final AtomicReference<U> result;
4929 if ((searchFunction = this.searchFunction) != null &&
4930 (result = this.result) != null) {
4931 for (int i = baseIndex, f, h; batch > 0 &&
4932 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4933 if (result.get() != null)
4934 return;
4935 addToPendingCount(1);
4936 new SearchValuesTask<K,V,U>
4937 (this, batch >>>= 1, baseLimit = h, f, tab,
4938 searchFunction, result).fork();
4939 }
4940 while (result.get() == null) {
4941 U u;
4942 Node<K,V> p;
4943 if ((p = advance()) == null) {
4944 propagateCompletion();
4945 break;
4946 }
4947 if ((u = searchFunction.apply(p.val)) != null) {
4948 if (result.compareAndSet(null, u))
4949 quietlyCompleteRoot();
4950 break;
4951 }
4952 }
4953 }
4954 }
4955 }
4956
4957 @SuppressWarnings("serial")
4958 static final class SearchEntriesTask<K,V,U>
4959 extends BulkTask<K,V,U> {
4960 final Fun<Entry<K,V>, ? extends U> searchFunction;
4961 final AtomicReference<U> result;
4962 SearchEntriesTask
4963 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
4964 Fun<Entry<K,V>, ? extends U> searchFunction,
4965 AtomicReference<U> result) {
4966 super(p, b, i, f, t);
4967 this.searchFunction = searchFunction; this.result = result;
4968 }
4969 public final U getRawResult() { return result.get(); }
4970 public final void compute() {
4971 final Fun<Entry<K,V>, ? extends U> searchFunction;
4972 final AtomicReference<U> result;
4973 if ((searchFunction = this.searchFunction) != null &&
4974 (result = this.result) != null) {
4975 for (int i = baseIndex, f, h; batch > 0 &&
4976 (h = ((f = baseLimit) + i) >>> 1) > i;) {
4977 if (result.get() != null)
4978 return;
4979 addToPendingCount(1);
4980 new SearchEntriesTask<K,V,U>
4981 (this, batch >>>= 1, baseLimit = h, f, tab,
4982 searchFunction, result).fork();
4983 }
4984 while (result.get() == null) {
4985 U u;
4986 Node<K,V> p;
4987 if ((p = advance()) == null) {
4988 propagateCompletion();
4989 break;
4990 }
4991 if ((u = searchFunction.apply(p)) != null) {
4992 if (result.compareAndSet(null, u))
4993 quietlyCompleteRoot();
4994 return;
4995 }
4996 }
4997 }
4998 }
4999 }
5000
5001 @SuppressWarnings("serial")
5002 static final class SearchMappingsTask<K,V,U>
5003 extends BulkTask<K,V,U> {
5004 final BiFun<? super K, ? super V, ? extends U> searchFunction;
5005 final AtomicReference<U> result;
5006 SearchMappingsTask
5007 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5008 BiFun<? super K, ? super V, ? extends U> searchFunction,
5009 AtomicReference<U> result) {
5010 super(p, b, i, f, t);
5011 this.searchFunction = searchFunction; this.result = result;
5012 }
5013 public final U getRawResult() { return result.get(); }
5014 public final void compute() {
5015 final BiFun<? super K, ? super V, ? extends U> searchFunction;
5016 final AtomicReference<U> result;
5017 if ((searchFunction = this.searchFunction) != null &&
5018 (result = this.result) != null) {
5019 for (int i = baseIndex, f, h; batch > 0 &&
5020 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5021 if (result.get() != null)
5022 return;
5023 addToPendingCount(1);
5024 new SearchMappingsTask<K,V,U>
5025 (this, batch >>>= 1, baseLimit = h, f, tab,
5026 searchFunction, result).fork();
5027 }
5028 while (result.get() == null) {
5029 U u;
5030 Node<K,V> p;
5031 if ((p = advance()) == null) {
5032 propagateCompletion();
5033 break;
5034 }
5035 if ((u = searchFunction.apply(p.key, p.val)) != null) {
5036 if (result.compareAndSet(null, u))
5037 quietlyCompleteRoot();
5038 break;
5039 }
5040 }
5041 }
5042 }
5043 }
5044
5045 @SuppressWarnings("serial")
5046 static final class ReduceKeysTask<K,V>
5047 extends BulkTask<K,V,K> {
5048 final BiFun<? super K, ? super K, ? extends K> reducer;
5049 K result;
5050 ReduceKeysTask<K,V> rights, nextRight;
5051 ReduceKeysTask
5052 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5053 ReduceKeysTask<K,V> nextRight,
5054 BiFun<? super K, ? super K, ? extends K> reducer) {
5055 super(p, b, i, f, t); this.nextRight = nextRight;
5056 this.reducer = reducer;
5057 }
5058 public final K getRawResult() { return result; }
5059 public final void compute() {
5060 final BiFun<? super K, ? super K, ? extends K> reducer;
5061 if ((reducer = this.reducer) != null) {
5062 for (int i = baseIndex, f, h; batch > 0 &&
5063 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5064 addToPendingCount(1);
5065 (rights = new ReduceKeysTask<K,V>
5066 (this, batch >>>= 1, baseLimit = h, f, tab,
5067 rights, reducer)).fork();
5068 }
5069 K r = null;
5070 for (Node<K,V> p; (p = advance()) != null; ) {
5071 K u = p.key;
5072 r = (r == null) ? u : u == null ? r : reducer.apply(r, u);
5073 }
5074 result = r;
5075 CountedCompleter<?> c;
5076 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5077 @SuppressWarnings("unchecked") ReduceKeysTask<K,V>
5078 t = (ReduceKeysTask<K,V>)c,
5079 s = t.rights;
5080 while (s != null) {
5081 K tr, sr;
5082 if ((sr = s.result) != null)
5083 t.result = (((tr = t.result) == null) ? sr :
5084 reducer.apply(tr, sr));
5085 s = t.rights = s.nextRight;
5086 }
5087 }
5088 }
5089 }
5090 }
5091
5092 @SuppressWarnings("serial")
5093 static final class ReduceValuesTask<K,V>
5094 extends BulkTask<K,V,V> {
5095 final BiFun<? super V, ? super V, ? extends V> reducer;
5096 V result;
5097 ReduceValuesTask<K,V> rights, nextRight;
5098 ReduceValuesTask
5099 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5100 ReduceValuesTask<K,V> nextRight,
5101 BiFun<? super V, ? super V, ? extends V> reducer) {
5102 super(p, b, i, f, t); this.nextRight = nextRight;
5103 this.reducer = reducer;
5104 }
5105 public final V getRawResult() { return result; }
5106 public final void compute() {
5107 final BiFun<? super V, ? super V, ? extends V> reducer;
5108 if ((reducer = this.reducer) != null) {
5109 for (int i = baseIndex, f, h; batch > 0 &&
5110 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5111 addToPendingCount(1);
5112 (rights = new ReduceValuesTask<K,V>
5113 (this, batch >>>= 1, baseLimit = h, f, tab,
5114 rights, reducer)).fork();
5115 }
5116 V r = null;
5117 for (Node<K,V> p; (p = advance()) != null; ) {
5118 V v = p.val;
5119 r = (r == null) ? v : reducer.apply(r, v);
5120 }
5121 result = r;
5122 CountedCompleter<?> c;
5123 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5124 @SuppressWarnings("unchecked") ReduceValuesTask<K,V>
5125 t = (ReduceValuesTask<K,V>)c,
5126 s = t.rights;
5127 while (s != null) {
5128 V tr, sr;
5129 if ((sr = s.result) != null)
5130 t.result = (((tr = t.result) == null) ? sr :
5131 reducer.apply(tr, sr));
5132 s = t.rights = s.nextRight;
5133 }
5134 }
5135 }
5136 }
5137 }
5138
5139 @SuppressWarnings("serial")
5140 static final class ReduceEntriesTask<K,V>
5141 extends BulkTask<K,V,Map.Entry<K,V>> {
5142 final BiFun<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer;
5143 Map.Entry<K,V> result;
5144 ReduceEntriesTask<K,V> rights, nextRight;
5145 ReduceEntriesTask
5146 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5147 ReduceEntriesTask<K,V> nextRight,
5148 BiFun<Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer) {
5149 super(p, b, i, f, t); this.nextRight = nextRight;
5150 this.reducer = reducer;
5151 }
5152 public final Map.Entry<K,V> getRawResult() { return result; }
5153 public final void compute() {
5154 final BiFun<Map.Entry<K,V>, Map.Entry<K,V>, ? extends Map.Entry<K,V>> reducer;
5155 if ((reducer = this.reducer) != null) {
5156 for (int i = baseIndex, f, h; batch > 0 &&
5157 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5158 addToPendingCount(1);
5159 (rights = new ReduceEntriesTask<K,V>
5160 (this, batch >>>= 1, baseLimit = h, f, tab,
5161 rights, reducer)).fork();
5162 }
5163 Map.Entry<K,V> r = null;
5164 for (Node<K,V> p; (p = advance()) != null; )
5165 r = (r == null) ? p : reducer.apply(r, p);
5166 result = r;
5167 CountedCompleter<?> c;
5168 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5169 @SuppressWarnings("unchecked") ReduceEntriesTask<K,V>
5170 t = (ReduceEntriesTask<K,V>)c,
5171 s = t.rights;
5172 while (s != null) {
5173 Map.Entry<K,V> tr, sr;
5174 if ((sr = s.result) != null)
5175 t.result = (((tr = t.result) == null) ? sr :
5176 reducer.apply(tr, sr));
5177 s = t.rights = s.nextRight;
5178 }
5179 }
5180 }
5181 }
5182 }
5183
5184 @SuppressWarnings("serial")
5185 static final class MapReduceKeysTask<K,V,U>
5186 extends BulkTask<K,V,U> {
5187 final Fun<? super K, ? extends U> transformer;
5188 final BiFun<? super U, ? super U, ? extends U> reducer;
5189 U result;
5190 MapReduceKeysTask<K,V,U> rights, nextRight;
5191 MapReduceKeysTask
5192 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5193 MapReduceKeysTask<K,V,U> nextRight,
5194 Fun<? super K, ? extends U> transformer,
5195 BiFun<? super U, ? super U, ? extends U> reducer) {
5196 super(p, b, i, f, t); this.nextRight = nextRight;
5197 this.transformer = transformer;
5198 this.reducer = reducer;
5199 }
5200 public final U getRawResult() { return result; }
5201 public final void compute() {
5202 final Fun<? super K, ? extends U> transformer;
5203 final BiFun<? super U, ? super U, ? extends U> reducer;
5204 if ((transformer = this.transformer) != null &&
5205 (reducer = this.reducer) != null) {
5206 for (int i = baseIndex, f, h; batch > 0 &&
5207 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5208 addToPendingCount(1);
5209 (rights = new MapReduceKeysTask<K,V,U>
5210 (this, batch >>>= 1, baseLimit = h, f, tab,
5211 rights, transformer, reducer)).fork();
5212 }
5213 U r = null;
5214 for (Node<K,V> p; (p = advance()) != null; ) {
5215 U u;
5216 if ((u = transformer.apply(p.key)) != null)
5217 r = (r == null) ? u : reducer.apply(r, u);
5218 }
5219 result = r;
5220 CountedCompleter<?> c;
5221 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5222 @SuppressWarnings("unchecked") MapReduceKeysTask<K,V,U>
5223 t = (MapReduceKeysTask<K,V,U>)c,
5224 s = t.rights;
5225 while (s != null) {
5226 U tr, sr;
5227 if ((sr = s.result) != null)
5228 t.result = (((tr = t.result) == null) ? sr :
5229 reducer.apply(tr, sr));
5230 s = t.rights = s.nextRight;
5231 }
5232 }
5233 }
5234 }
5235 }
5236
5237 @SuppressWarnings("serial")
5238 static final class MapReduceValuesTask<K,V,U>
5239 extends BulkTask<K,V,U> {
5240 final Fun<? super V, ? extends U> transformer;
5241 final BiFun<? super U, ? super U, ? extends U> reducer;
5242 U result;
5243 MapReduceValuesTask<K,V,U> rights, nextRight;
5244 MapReduceValuesTask
5245 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5246 MapReduceValuesTask<K,V,U> nextRight,
5247 Fun<? super V, ? extends U> transformer,
5248 BiFun<? super U, ? super U, ? extends U> reducer) {
5249 super(p, b, i, f, t); this.nextRight = nextRight;
5250 this.transformer = transformer;
5251 this.reducer = reducer;
5252 }
5253 public final U getRawResult() { return result; }
5254 public final void compute() {
5255 final Fun<? super V, ? extends U> transformer;
5256 final BiFun<? super U, ? super U, ? extends U> reducer;
5257 if ((transformer = this.transformer) != null &&
5258 (reducer = this.reducer) != null) {
5259 for (int i = baseIndex, f, h; batch > 0 &&
5260 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5261 addToPendingCount(1);
5262 (rights = new MapReduceValuesTask<K,V,U>
5263 (this, batch >>>= 1, baseLimit = h, f, tab,
5264 rights, transformer, reducer)).fork();
5265 }
5266 U r = null;
5267 for (Node<K,V> p; (p = advance()) != null; ) {
5268 U u;
5269 if ((u = transformer.apply(p.val)) != null)
5270 r = (r == null) ? u : reducer.apply(r, u);
5271 }
5272 result = r;
5273 CountedCompleter<?> c;
5274 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5275 @SuppressWarnings("unchecked") MapReduceValuesTask<K,V,U>
5276 t = (MapReduceValuesTask<K,V,U>)c,
5277 s = t.rights;
5278 while (s != null) {
5279 U tr, sr;
5280 if ((sr = s.result) != null)
5281 t.result = (((tr = t.result) == null) ? sr :
5282 reducer.apply(tr, sr));
5283 s = t.rights = s.nextRight;
5284 }
5285 }
5286 }
5287 }
5288 }
5289
5290 @SuppressWarnings("serial")
5291 static final class MapReduceEntriesTask<K,V,U>
5292 extends BulkTask<K,V,U> {
5293 final Fun<Map.Entry<K,V>, ? extends U> transformer;
5294 final BiFun<? super U, ? super U, ? extends U> reducer;
5295 U result;
5296 MapReduceEntriesTask<K,V,U> rights, nextRight;
5297 MapReduceEntriesTask
5298 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5299 MapReduceEntriesTask<K,V,U> nextRight,
5300 Fun<Map.Entry<K,V>, ? extends U> transformer,
5301 BiFun<? super U, ? super U, ? extends U> reducer) {
5302 super(p, b, i, f, t); this.nextRight = nextRight;
5303 this.transformer = transformer;
5304 this.reducer = reducer;
5305 }
5306 public final U getRawResult() { return result; }
5307 public final void compute() {
5308 final Fun<Map.Entry<K,V>, ? extends U> transformer;
5309 final BiFun<? super U, ? super U, ? extends U> reducer;
5310 if ((transformer = this.transformer) != null &&
5311 (reducer = this.reducer) != null) {
5312 for (int i = baseIndex, f, h; batch > 0 &&
5313 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5314 addToPendingCount(1);
5315 (rights = new MapReduceEntriesTask<K,V,U>
5316 (this, batch >>>= 1, baseLimit = h, f, tab,
5317 rights, transformer, reducer)).fork();
5318 }
5319 U r = null;
5320 for (Node<K,V> p; (p = advance()) != null; ) {
5321 U u;
5322 if ((u = transformer.apply(p)) != null)
5323 r = (r == null) ? u : reducer.apply(r, u);
5324 }
5325 result = r;
5326 CountedCompleter<?> c;
5327 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5328 @SuppressWarnings("unchecked") MapReduceEntriesTask<K,V,U>
5329 t = (MapReduceEntriesTask<K,V,U>)c,
5330 s = t.rights;
5331 while (s != null) {
5332 U tr, sr;
5333 if ((sr = s.result) != null)
5334 t.result = (((tr = t.result) == null) ? sr :
5335 reducer.apply(tr, sr));
5336 s = t.rights = s.nextRight;
5337 }
5338 }
5339 }
5340 }
5341 }
5342
5343 @SuppressWarnings("serial")
5344 static final class MapReduceMappingsTask<K,V,U>
5345 extends BulkTask<K,V,U> {
5346 final BiFun<? super K, ? super V, ? extends U> transformer;
5347 final BiFun<? super U, ? super U, ? extends U> reducer;
5348 U result;
5349 MapReduceMappingsTask<K,V,U> rights, nextRight;
5350 MapReduceMappingsTask
5351 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5352 MapReduceMappingsTask<K,V,U> nextRight,
5353 BiFun<? super K, ? super V, ? extends U> transformer,
5354 BiFun<? super U, ? super U, ? extends U> reducer) {
5355 super(p, b, i, f, t); this.nextRight = nextRight;
5356 this.transformer = transformer;
5357 this.reducer = reducer;
5358 }
5359 public final U getRawResult() { return result; }
5360 public final void compute() {
5361 final BiFun<? super K, ? super V, ? extends U> transformer;
5362 final BiFun<? super U, ? super U, ? extends U> reducer;
5363 if ((transformer = this.transformer) != null &&
5364 (reducer = this.reducer) != null) {
5365 for (int i = baseIndex, f, h; batch > 0 &&
5366 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5367 addToPendingCount(1);
5368 (rights = new MapReduceMappingsTask<K,V,U>
5369 (this, batch >>>= 1, baseLimit = h, f, tab,
5370 rights, transformer, reducer)).fork();
5371 }
5372 U r = null;
5373 for (Node<K,V> p; (p = advance()) != null; ) {
5374 U u;
5375 if ((u = transformer.apply(p.key, p.val)) != null)
5376 r = (r == null) ? u : reducer.apply(r, u);
5377 }
5378 result = r;
5379 CountedCompleter<?> c;
5380 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5381 @SuppressWarnings("unchecked") MapReduceMappingsTask<K,V,U>
5382 t = (MapReduceMappingsTask<K,V,U>)c,
5383 s = t.rights;
5384 while (s != null) {
5385 U tr, sr;
5386 if ((sr = s.result) != null)
5387 t.result = (((tr = t.result) == null) ? sr :
5388 reducer.apply(tr, sr));
5389 s = t.rights = s.nextRight;
5390 }
5391 }
5392 }
5393 }
5394 }
5395
5396 @SuppressWarnings("serial")
5397 static final class MapReduceKeysToDoubleTask<K,V>
5398 extends BulkTask<K,V,Double> {
5399 final ObjectToDouble<? super K> transformer;
5400 final DoubleByDoubleToDouble reducer;
5401 final double basis;
5402 double result;
5403 MapReduceKeysToDoubleTask<K,V> rights, nextRight;
5404 MapReduceKeysToDoubleTask
5405 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5406 MapReduceKeysToDoubleTask<K,V> nextRight,
5407 ObjectToDouble<? super K> transformer,
5408 double basis,
5409 DoubleByDoubleToDouble reducer) {
5410 super(p, b, i, f, t); this.nextRight = nextRight;
5411 this.transformer = transformer;
5412 this.basis = basis; this.reducer = reducer;
5413 }
5414 public final Double getRawResult() { return result; }
5415 public final void compute() {
5416 final ObjectToDouble<? super K> transformer;
5417 final DoubleByDoubleToDouble reducer;
5418 if ((transformer = this.transformer) != null &&
5419 (reducer = this.reducer) != null) {
5420 double r = this.basis;
5421 for (int i = baseIndex, f, h; batch > 0 &&
5422 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5423 addToPendingCount(1);
5424 (rights = new MapReduceKeysToDoubleTask<K,V>
5425 (this, batch >>>= 1, baseLimit = h, f, tab,
5426 rights, transformer, r, reducer)).fork();
5427 }
5428 for (Node<K,V> p; (p = advance()) != null; )
5429 r = reducer.apply(r, transformer.apply(p.key));
5430 result = r;
5431 CountedCompleter<?> c;
5432 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5433 @SuppressWarnings("unchecked") MapReduceKeysToDoubleTask<K,V>
5434 t = (MapReduceKeysToDoubleTask<K,V>)c,
5435 s = t.rights;
5436 while (s != null) {
5437 t.result = reducer.apply(t.result, s.result);
5438 s = t.rights = s.nextRight;
5439 }
5440 }
5441 }
5442 }
5443 }
5444
5445 @SuppressWarnings("serial")
5446 static final class MapReduceValuesToDoubleTask<K,V>
5447 extends BulkTask<K,V,Double> {
5448 final ObjectToDouble<? super V> transformer;
5449 final DoubleByDoubleToDouble reducer;
5450 final double basis;
5451 double result;
5452 MapReduceValuesToDoubleTask<K,V> rights, nextRight;
5453 MapReduceValuesToDoubleTask
5454 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5455 MapReduceValuesToDoubleTask<K,V> nextRight,
5456 ObjectToDouble<? super V> transformer,
5457 double basis,
5458 DoubleByDoubleToDouble reducer) {
5459 super(p, b, i, f, t); this.nextRight = nextRight;
5460 this.transformer = transformer;
5461 this.basis = basis; this.reducer = reducer;
5462 }
5463 public final Double getRawResult() { return result; }
5464 public final void compute() {
5465 final ObjectToDouble<? super V> transformer;
5466 final DoubleByDoubleToDouble reducer;
5467 if ((transformer = this.transformer) != null &&
5468 (reducer = this.reducer) != null) {
5469 double r = this.basis;
5470 for (int i = baseIndex, f, h; batch > 0 &&
5471 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5472 addToPendingCount(1);
5473 (rights = new MapReduceValuesToDoubleTask<K,V>
5474 (this, batch >>>= 1, baseLimit = h, f, tab,
5475 rights, transformer, r, reducer)).fork();
5476 }
5477 for (Node<K,V> p; (p = advance()) != null; )
5478 r = reducer.apply(r, transformer.apply(p.val));
5479 result = r;
5480 CountedCompleter<?> c;
5481 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5482 @SuppressWarnings("unchecked") MapReduceValuesToDoubleTask<K,V>
5483 t = (MapReduceValuesToDoubleTask<K,V>)c,
5484 s = t.rights;
5485 while (s != null) {
5486 t.result = reducer.apply(t.result, s.result);
5487 s = t.rights = s.nextRight;
5488 }
5489 }
5490 }
5491 }
5492 }
5493
5494 @SuppressWarnings("serial")
5495 static final class MapReduceEntriesToDoubleTask<K,V>
5496 extends BulkTask<K,V,Double> {
5497 final ObjectToDouble<Map.Entry<K,V>> transformer;
5498 final DoubleByDoubleToDouble reducer;
5499 final double basis;
5500 double result;
5501 MapReduceEntriesToDoubleTask<K,V> rights, nextRight;
5502 MapReduceEntriesToDoubleTask
5503 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5504 MapReduceEntriesToDoubleTask<K,V> nextRight,
5505 ObjectToDouble<Map.Entry<K,V>> transformer,
5506 double basis,
5507 DoubleByDoubleToDouble reducer) {
5508 super(p, b, i, f, t); this.nextRight = nextRight;
5509 this.transformer = transformer;
5510 this.basis = basis; this.reducer = reducer;
5511 }
5512 public final Double getRawResult() { return result; }
5513 public final void compute() {
5514 final ObjectToDouble<Map.Entry<K,V>> transformer;
5515 final DoubleByDoubleToDouble reducer;
5516 if ((transformer = this.transformer) != null &&
5517 (reducer = this.reducer) != null) {
5518 double r = this.basis;
5519 for (int i = baseIndex, f, h; batch > 0 &&
5520 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5521 addToPendingCount(1);
5522 (rights = new MapReduceEntriesToDoubleTask<K,V>
5523 (this, batch >>>= 1, baseLimit = h, f, tab,
5524 rights, transformer, r, reducer)).fork();
5525 }
5526 for (Node<K,V> p; (p = advance()) != null; )
5527 r = reducer.apply(r, transformer.apply(p));
5528 result = r;
5529 CountedCompleter<?> c;
5530 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5531 @SuppressWarnings("unchecked") MapReduceEntriesToDoubleTask<K,V>
5532 t = (MapReduceEntriesToDoubleTask<K,V>)c,
5533 s = t.rights;
5534 while (s != null) {
5535 t.result = reducer.apply(t.result, s.result);
5536 s = t.rights = s.nextRight;
5537 }
5538 }
5539 }
5540 }
5541 }
5542
5543 @SuppressWarnings("serial")
5544 static final class MapReduceMappingsToDoubleTask<K,V>
5545 extends BulkTask<K,V,Double> {
5546 final ObjectByObjectToDouble<? super K, ? super V> transformer;
5547 final DoubleByDoubleToDouble reducer;
5548 final double basis;
5549 double result;
5550 MapReduceMappingsToDoubleTask<K,V> rights, nextRight;
5551 MapReduceMappingsToDoubleTask
5552 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5553 MapReduceMappingsToDoubleTask<K,V> nextRight,
5554 ObjectByObjectToDouble<? super K, ? super V> transformer,
5555 double basis,
5556 DoubleByDoubleToDouble reducer) {
5557 super(p, b, i, f, t); this.nextRight = nextRight;
5558 this.transformer = transformer;
5559 this.basis = basis; this.reducer = reducer;
5560 }
5561 public final Double getRawResult() { return result; }
5562 public final void compute() {
5563 final ObjectByObjectToDouble<? super K, ? super V> transformer;
5564 final DoubleByDoubleToDouble reducer;
5565 if ((transformer = this.transformer) != null &&
5566 (reducer = this.reducer) != null) {
5567 double r = this.basis;
5568 for (int i = baseIndex, f, h; batch > 0 &&
5569 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5570 addToPendingCount(1);
5571 (rights = new MapReduceMappingsToDoubleTask<K,V>
5572 (this, batch >>>= 1, baseLimit = h, f, tab,
5573 rights, transformer, r, reducer)).fork();
5574 }
5575 for (Node<K,V> p; (p = advance()) != null; )
5576 r = reducer.apply(r, transformer.apply(p.key, p.val));
5577 result = r;
5578 CountedCompleter<?> c;
5579 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5580 @SuppressWarnings("unchecked") MapReduceMappingsToDoubleTask<K,V>
5581 t = (MapReduceMappingsToDoubleTask<K,V>)c,
5582 s = t.rights;
5583 while (s != null) {
5584 t.result = reducer.apply(t.result, s.result);
5585 s = t.rights = s.nextRight;
5586 }
5587 }
5588 }
5589 }
5590 }
5591
5592 @SuppressWarnings("serial")
5593 static final class MapReduceKeysToLongTask<K,V>
5594 extends BulkTask<K,V,Long> {
5595 final ObjectToLong<? super K> transformer;
5596 final LongByLongToLong reducer;
5597 final long basis;
5598 long result;
5599 MapReduceKeysToLongTask<K,V> rights, nextRight;
5600 MapReduceKeysToLongTask
5601 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5602 MapReduceKeysToLongTask<K,V> nextRight,
5603 ObjectToLong<? super K> transformer,
5604 long basis,
5605 LongByLongToLong reducer) {
5606 super(p, b, i, f, t); this.nextRight = nextRight;
5607 this.transformer = transformer;
5608 this.basis = basis; this.reducer = reducer;
5609 }
5610 public final Long getRawResult() { return result; }
5611 public final void compute() {
5612 final ObjectToLong<? super K> transformer;
5613 final LongByLongToLong reducer;
5614 if ((transformer = this.transformer) != null &&
5615 (reducer = this.reducer) != null) {
5616 long r = this.basis;
5617 for (int i = baseIndex, f, h; batch > 0 &&
5618 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5619 addToPendingCount(1);
5620 (rights = new MapReduceKeysToLongTask<K,V>
5621 (this, batch >>>= 1, baseLimit = h, f, tab,
5622 rights, transformer, r, reducer)).fork();
5623 }
5624 for (Node<K,V> p; (p = advance()) != null; )
5625 r = reducer.apply(r, transformer.apply(p.key));
5626 result = r;
5627 CountedCompleter<?> c;
5628 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5629 @SuppressWarnings("unchecked") MapReduceKeysToLongTask<K,V>
5630 t = (MapReduceKeysToLongTask<K,V>)c,
5631 s = t.rights;
5632 while (s != null) {
5633 t.result = reducer.apply(t.result, s.result);
5634 s = t.rights = s.nextRight;
5635 }
5636 }
5637 }
5638 }
5639 }
5640
5641 @SuppressWarnings("serial")
5642 static final class MapReduceValuesToLongTask<K,V>
5643 extends BulkTask<K,V,Long> {
5644 final ObjectToLong<? super V> transformer;
5645 final LongByLongToLong reducer;
5646 final long basis;
5647 long result;
5648 MapReduceValuesToLongTask<K,V> rights, nextRight;
5649 MapReduceValuesToLongTask
5650 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5651 MapReduceValuesToLongTask<K,V> nextRight,
5652 ObjectToLong<? super V> transformer,
5653 long basis,
5654 LongByLongToLong reducer) {
5655 super(p, b, i, f, t); this.nextRight = nextRight;
5656 this.transformer = transformer;
5657 this.basis = basis; this.reducer = reducer;
5658 }
5659 public final Long getRawResult() { return result; }
5660 public final void compute() {
5661 final ObjectToLong<? super V> transformer;
5662 final LongByLongToLong reducer;
5663 if ((transformer = this.transformer) != null &&
5664 (reducer = this.reducer) != null) {
5665 long r = this.basis;
5666 for (int i = baseIndex, f, h; batch > 0 &&
5667 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5668 addToPendingCount(1);
5669 (rights = new MapReduceValuesToLongTask<K,V>
5670 (this, batch >>>= 1, baseLimit = h, f, tab,
5671 rights, transformer, r, reducer)).fork();
5672 }
5673 for (Node<K,V> p; (p = advance()) != null; )
5674 r = reducer.apply(r, transformer.apply(p.val));
5675 result = r;
5676 CountedCompleter<?> c;
5677 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5678 @SuppressWarnings("unchecked") MapReduceValuesToLongTask<K,V>
5679 t = (MapReduceValuesToLongTask<K,V>)c,
5680 s = t.rights;
5681 while (s != null) {
5682 t.result = reducer.apply(t.result, s.result);
5683 s = t.rights = s.nextRight;
5684 }
5685 }
5686 }
5687 }
5688 }
5689
5690 @SuppressWarnings("serial")
5691 static final class MapReduceEntriesToLongTask<K,V>
5692 extends BulkTask<K,V,Long> {
5693 final ObjectToLong<Map.Entry<K,V>> transformer;
5694 final LongByLongToLong reducer;
5695 final long basis;
5696 long result;
5697 MapReduceEntriesToLongTask<K,V> rights, nextRight;
5698 MapReduceEntriesToLongTask
5699 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5700 MapReduceEntriesToLongTask<K,V> nextRight,
5701 ObjectToLong<Map.Entry<K,V>> transformer,
5702 long basis,
5703 LongByLongToLong reducer) {
5704 super(p, b, i, f, t); this.nextRight = nextRight;
5705 this.transformer = transformer;
5706 this.basis = basis; this.reducer = reducer;
5707 }
5708 public final Long getRawResult() { return result; }
5709 public final void compute() {
5710 final ObjectToLong<Map.Entry<K,V>> transformer;
5711 final LongByLongToLong reducer;
5712 if ((transformer = this.transformer) != null &&
5713 (reducer = this.reducer) != null) {
5714 long r = this.basis;
5715 for (int i = baseIndex, f, h; batch > 0 &&
5716 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5717 addToPendingCount(1);
5718 (rights = new MapReduceEntriesToLongTask<K,V>
5719 (this, batch >>>= 1, baseLimit = h, f, tab,
5720 rights, transformer, r, reducer)).fork();
5721 }
5722 for (Node<K,V> p; (p = advance()) != null; )
5723 r = reducer.apply(r, transformer.apply(p));
5724 result = r;
5725 CountedCompleter<?> c;
5726 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5727 @SuppressWarnings("unchecked") MapReduceEntriesToLongTask<K,V>
5728 t = (MapReduceEntriesToLongTask<K,V>)c,
5729 s = t.rights;
5730 while (s != null) {
5731 t.result = reducer.apply(t.result, s.result);
5732 s = t.rights = s.nextRight;
5733 }
5734 }
5735 }
5736 }
5737 }
5738
5739 @SuppressWarnings("serial")
5740 static final class MapReduceMappingsToLongTask<K,V>
5741 extends BulkTask<K,V,Long> {
5742 final ObjectByObjectToLong<? super K, ? super V> transformer;
5743 final LongByLongToLong reducer;
5744 final long basis;
5745 long result;
5746 MapReduceMappingsToLongTask<K,V> rights, nextRight;
5747 MapReduceMappingsToLongTask
5748 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5749 MapReduceMappingsToLongTask<K,V> nextRight,
5750 ObjectByObjectToLong<? super K, ? super V> transformer,
5751 long basis,
5752 LongByLongToLong reducer) {
5753 super(p, b, i, f, t); this.nextRight = nextRight;
5754 this.transformer = transformer;
5755 this.basis = basis; this.reducer = reducer;
5756 }
5757 public final Long getRawResult() { return result; }
5758 public final void compute() {
5759 final ObjectByObjectToLong<? super K, ? super V> transformer;
5760 final LongByLongToLong reducer;
5761 if ((transformer = this.transformer) != null &&
5762 (reducer = this.reducer) != null) {
5763 long r = this.basis;
5764 for (int i = baseIndex, f, h; batch > 0 &&
5765 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5766 addToPendingCount(1);
5767 (rights = new MapReduceMappingsToLongTask<K,V>
5768 (this, batch >>>= 1, baseLimit = h, f, tab,
5769 rights, transformer, r, reducer)).fork();
5770 }
5771 for (Node<K,V> p; (p = advance()) != null; )
5772 r = reducer.apply(r, transformer.apply(p.key, p.val));
5773 result = r;
5774 CountedCompleter<?> c;
5775 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5776 @SuppressWarnings("unchecked") MapReduceMappingsToLongTask<K,V>
5777 t = (MapReduceMappingsToLongTask<K,V>)c,
5778 s = t.rights;
5779 while (s != null) {
5780 t.result = reducer.apply(t.result, s.result);
5781 s = t.rights = s.nextRight;
5782 }
5783 }
5784 }
5785 }
5786 }
5787
5788 @SuppressWarnings("serial")
5789 static final class MapReduceKeysToIntTask<K,V>
5790 extends BulkTask<K,V,Integer> {
5791 final ObjectToInt<? super K> transformer;
5792 final IntByIntToInt reducer;
5793 final int basis;
5794 int result;
5795 MapReduceKeysToIntTask<K,V> rights, nextRight;
5796 MapReduceKeysToIntTask
5797 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5798 MapReduceKeysToIntTask<K,V> nextRight,
5799 ObjectToInt<? super K> transformer,
5800 int basis,
5801 IntByIntToInt reducer) {
5802 super(p, b, i, f, t); this.nextRight = nextRight;
5803 this.transformer = transformer;
5804 this.basis = basis; this.reducer = reducer;
5805 }
5806 public final Integer getRawResult() { return result; }
5807 public final void compute() {
5808 final ObjectToInt<? super K> transformer;
5809 final IntByIntToInt reducer;
5810 if ((transformer = this.transformer) != null &&
5811 (reducer = this.reducer) != null) {
5812 int r = this.basis;
5813 for (int i = baseIndex, f, h; batch > 0 &&
5814 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5815 addToPendingCount(1);
5816 (rights = new MapReduceKeysToIntTask<K,V>
5817 (this, batch >>>= 1, baseLimit = h, f, tab,
5818 rights, transformer, r, reducer)).fork();
5819 }
5820 for (Node<K,V> p; (p = advance()) != null; )
5821 r = reducer.apply(r, transformer.apply(p.key));
5822 result = r;
5823 CountedCompleter<?> c;
5824 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5825 @SuppressWarnings("unchecked") MapReduceKeysToIntTask<K,V>
5826 t = (MapReduceKeysToIntTask<K,V>)c,
5827 s = t.rights;
5828 while (s != null) {
5829 t.result = reducer.apply(t.result, s.result);
5830 s = t.rights = s.nextRight;
5831 }
5832 }
5833 }
5834 }
5835 }
5836
5837 @SuppressWarnings("serial")
5838 static final class MapReduceValuesToIntTask<K,V>
5839 extends BulkTask<K,V,Integer> {
5840 final ObjectToInt<? super V> transformer;
5841 final IntByIntToInt reducer;
5842 final int basis;
5843 int result;
5844 MapReduceValuesToIntTask<K,V> rights, nextRight;
5845 MapReduceValuesToIntTask
5846 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5847 MapReduceValuesToIntTask<K,V> nextRight,
5848 ObjectToInt<? super V> transformer,
5849 int basis,
5850 IntByIntToInt reducer) {
5851 super(p, b, i, f, t); this.nextRight = nextRight;
5852 this.transformer = transformer;
5853 this.basis = basis; this.reducer = reducer;
5854 }
5855 public final Integer getRawResult() { return result; }
5856 public final void compute() {
5857 final ObjectToInt<? super V> transformer;
5858 final IntByIntToInt reducer;
5859 if ((transformer = this.transformer) != null &&
5860 (reducer = this.reducer) != null) {
5861 int r = this.basis;
5862 for (int i = baseIndex, f, h; batch > 0 &&
5863 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5864 addToPendingCount(1);
5865 (rights = new MapReduceValuesToIntTask<K,V>
5866 (this, batch >>>= 1, baseLimit = h, f, tab,
5867 rights, transformer, r, reducer)).fork();
5868 }
5869 for (Node<K,V> p; (p = advance()) != null; )
5870 r = reducer.apply(r, transformer.apply(p.val));
5871 result = r;
5872 CountedCompleter<?> c;
5873 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5874 @SuppressWarnings("unchecked") MapReduceValuesToIntTask<K,V>
5875 t = (MapReduceValuesToIntTask<K,V>)c,
5876 s = t.rights;
5877 while (s != null) {
5878 t.result = reducer.apply(t.result, s.result);
5879 s = t.rights = s.nextRight;
5880 }
5881 }
5882 }
5883 }
5884 }
5885
5886 @SuppressWarnings("serial")
5887 static final class MapReduceEntriesToIntTask<K,V>
5888 extends BulkTask<K,V,Integer> {
5889 final ObjectToInt<Map.Entry<K,V>> transformer;
5890 final IntByIntToInt reducer;
5891 final int basis;
5892 int result;
5893 MapReduceEntriesToIntTask<K,V> rights, nextRight;
5894 MapReduceEntriesToIntTask
5895 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5896 MapReduceEntriesToIntTask<K,V> nextRight,
5897 ObjectToInt<Map.Entry<K,V>> transformer,
5898 int basis,
5899 IntByIntToInt reducer) {
5900 super(p, b, i, f, t); this.nextRight = nextRight;
5901 this.transformer = transformer;
5902 this.basis = basis; this.reducer = reducer;
5903 }
5904 public final Integer getRawResult() { return result; }
5905 public final void compute() {
5906 final ObjectToInt<Map.Entry<K,V>> transformer;
5907 final IntByIntToInt reducer;
5908 if ((transformer = this.transformer) != null &&
5909 (reducer = this.reducer) != null) {
5910 int r = this.basis;
5911 for (int i = baseIndex, f, h; batch > 0 &&
5912 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5913 addToPendingCount(1);
5914 (rights = new MapReduceEntriesToIntTask<K,V>
5915 (this, batch >>>= 1, baseLimit = h, f, tab,
5916 rights, transformer, r, reducer)).fork();
5917 }
5918 for (Node<K,V> p; (p = advance()) != null; )
5919 r = reducer.apply(r, transformer.apply(p));
5920 result = r;
5921 CountedCompleter<?> c;
5922 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5923 @SuppressWarnings("unchecked") MapReduceEntriesToIntTask<K,V>
5924 t = (MapReduceEntriesToIntTask<K,V>)c,
5925 s = t.rights;
5926 while (s != null) {
5927 t.result = reducer.apply(t.result, s.result);
5928 s = t.rights = s.nextRight;
5929 }
5930 }
5931 }
5932 }
5933 }
5934
5935 @SuppressWarnings("serial")
5936 static final class MapReduceMappingsToIntTask<K,V>
5937 extends BulkTask<K,V,Integer> {
5938 final ObjectByObjectToInt<? super K, ? super V> transformer;
5939 final IntByIntToInt reducer;
5940 final int basis;
5941 int result;
5942 MapReduceMappingsToIntTask<K,V> rights, nextRight;
5943 MapReduceMappingsToIntTask
5944 (BulkTask<K,V,?> p, int b, int i, int f, Node<K,V>[] t,
5945 MapReduceMappingsToIntTask<K,V> nextRight,
5946 ObjectByObjectToInt<? super K, ? super V> transformer,
5947 int basis,
5948 IntByIntToInt reducer) {
5949 super(p, b, i, f, t); this.nextRight = nextRight;
5950 this.transformer = transformer;
5951 this.basis = basis; this.reducer = reducer;
5952 }
5953 public final Integer getRawResult() { return result; }
5954 public final void compute() {
5955 final ObjectByObjectToInt<? super K, ? super V> transformer;
5956 final IntByIntToInt reducer;
5957 if ((transformer = this.transformer) != null &&
5958 (reducer = this.reducer) != null) {
5959 int r = this.basis;
5960 for (int i = baseIndex, f, h; batch > 0 &&
5961 (h = ((f = baseLimit) + i) >>> 1) > i;) {
5962 addToPendingCount(1);
5963 (rights = new MapReduceMappingsToIntTask<K,V>
5964 (this, batch >>>= 1, baseLimit = h, f, tab,
5965 rights, transformer, r, reducer)).fork();
5966 }
5967 for (Node<K,V> p; (p = advance()) != null; )
5968 r = reducer.apply(r, transformer.apply(p.key, p.val));
5969 result = r;
5970 CountedCompleter<?> c;
5971 for (c = firstComplete(); c != null; c = c.nextComplete()) {
5972 @SuppressWarnings("unchecked") MapReduceMappingsToIntTask<K,V>
5973 t = (MapReduceMappingsToIntTask<K,V>)c,
5974 s = t.rights;
5975 while (s != null) {
5976 t.result = reducer.apply(t.result, s.result);
5977 s = t.rights = s.nextRight;
5978 }
5979 }
5980 }
5981 }
5982 }
5983
5984 /* ---------------- Counters -------------- */
5985
5986 // Adapted from LongAdder and Striped64.
5987 // See their internal docs for explanation.
5988
5989 // A padded cell for distributing counts
5990 static final class CounterCell {
5991 volatile long p0, p1, p2, p3, p4, p5, p6;
5992 volatile long value;
5993 volatile long q0, q1, q2, q3, q4, q5, q6;
5994 CounterCell(long x) { value = x; }
5995 }
5996
5997 /**
5998 * Holder for the thread-local hash code determining which
5999 * CounterCell to use. The code is initialized via the
6000 * counterHashCodeGenerator, but may be moved upon collisions.
6001 */
6002 static final class CounterHashCode {
6003 int code;
6004 }
6005
6006 /**
6007 * Generates initial value for per-thread CounterHashCodes.
6008 */
6009 static final AtomicInteger counterHashCodeGenerator = new AtomicInteger();
6010
6011 /**
6012 * Increment for counterHashCodeGenerator. See class ThreadLocal
6013 * for explanation.
6014 */
6015 static final int SEED_INCREMENT = 0x61c88647;
6016
6017 /**
6018 * Per-thread counter hash codes. Shared across all instances.
6019 */
6020 static final ThreadLocal<CounterHashCode> threadCounterHashCode =
6021 new ThreadLocal<CounterHashCode>();
6022
6023
6024 final long sumCount() {
6025 CounterCell[] as = counterCells; CounterCell a;
6026 long sum = baseCount;
6027 if (as != null) {
6028 for (int i = 0; i < as.length; ++i) {
6029 if ((a = as[i]) != null)
6030 sum += a.value;
6031 }
6032 }
6033 return sum;
6034 }
6035
6036 // See LongAdder version for explanation
6037 private final void fullAddCount(long x, CounterHashCode hc,
6038 boolean wasUncontended) {
6039 int h;
6040 if (hc == null) {
6041 hc = new CounterHashCode();
6042 int s = counterHashCodeGenerator.addAndGet(SEED_INCREMENT);
6043 h = hc.code = (s == 0) ? 1 : s; // Avoid zero
6044 threadCounterHashCode.set(hc);
6045 }
6046 else
6047 h = hc.code;
6048 boolean collide = false; // True if last slot nonempty
6049 for (;;) {
6050 CounterCell[] as; CounterCell a; int n; long v;
6051 if ((as = counterCells) != null && (n = as.length) > 0) {
6052 if ((a = as[(n - 1) & h]) == null) {
6053 if (cellsBusy == 0) { // Try to attach new Cell
6054 CounterCell r = new CounterCell(x); // Optimistic create
6055 if (cellsBusy == 0 &&
6056 U.compareAndSwapInt(this, CELLSBUSY, 0, 1)) {
6057 boolean created = false;
6058 try { // Recheck under lock
6059 CounterCell[] rs; int m, j;
6060 if ((rs = counterCells) != null &&
6061 (m = rs.length) > 0 &&
6062 rs[j = (m - 1) & h] == null) {
6063 rs[j] = r;
6064 created = true;
6065 }
6066 } finally {
6067 cellsBusy = 0;
6068 }
6069 if (created)
6070 break;
6071 continue; // Slot is now non-empty
6072 }
6073 }
6074 collide = false;
6075 }
6076 else if (!wasUncontended) // CAS already known to fail
6077 wasUncontended = true; // Continue after rehash
6078 else if (U.compareAndSwapLong(a, CELLVALUE, v = a.value, v + x))
6079 break;
6080 else if (counterCells != as || n >= NCPU)
6081 collide = false; // At max size or stale
6082 else if (!collide)
6083 collide = true;
6084 else if (cellsBusy == 0 &&
6085 U.compareAndSwapInt(this, CELLSBUSY, 0, 1)) {
6086 try {
6087 if (counterCells == as) {// Expand table unless stale
6088 CounterCell[] rs = new CounterCell[n << 1];
6089 for (int i = 0; i < n; ++i)
6090 rs[i] = as[i];
6091 counterCells = rs;
6092 }
6093 } finally {
6094 cellsBusy = 0;
6095 }
6096 collide = false;
6097 continue; // Retry with expanded table
6098 }
6099 h ^= h << 13; // Rehash
6100 h ^= h >>> 17;
6101 h ^= h << 5;
6102 }
6103 else if (cellsBusy == 0 && counterCells == as &&
6104 U.compareAndSwapInt(this, CELLSBUSY, 0, 1)) {
6105 boolean init = false;
6106 try { // Initialize table
6107 if (counterCells == as) {
6108 CounterCell[] rs = new CounterCell[2];
6109 rs[h & 1] = new CounterCell(x);
6110 counterCells = rs;
6111 init = true;
6112 }
6113 } finally {
6114 cellsBusy = 0;
6115 }
6116 if (init)
6117 break;
6118 }
6119 else if (U.compareAndSwapLong(this, BASECOUNT, v = baseCount, v + x))
6120 break; // Fall back on using base
6121 }
6122 hc.code = h; // Record index for next time
6123 }
6124
6125 // Unsafe mechanics
6126 private static final sun.misc.Unsafe U;
6127 private static final long SIZECTL;
6128 private static final long TRANSFERINDEX;
6129 private static final long TRANSFERORIGIN;
6130 private static final long BASECOUNT;
6131 private static final long CELLSBUSY;
6132 private static final long CELLVALUE;
6133 private static final long ABASE;
6134 private static final int ASHIFT;
6135
6136 static {
6137 try {
6138 U = getUnsafe();
6139 Class<?> k = ConcurrentHashMapV8.class;
6140 SIZECTL = U.objectFieldOffset
6141 (k.getDeclaredField("sizeCtl"));
6142 TRANSFERINDEX = U.objectFieldOffset
6143 (k.getDeclaredField("transferIndex"));
6144 TRANSFERORIGIN = U.objectFieldOffset
6145 (k.getDeclaredField("transferOrigin"));
6146 BASECOUNT = U.objectFieldOffset
6147 (k.getDeclaredField("baseCount"));
6148 CELLSBUSY = U.objectFieldOffset
6149 (k.getDeclaredField("cellsBusy"));
6150 Class<?> ck = CounterCell.class;
6151 CELLVALUE = U.objectFieldOffset
6152 (ck.getDeclaredField("value"));
6153 Class<?> ak = Node[].class;
6154 ABASE = U.arrayBaseOffset(ak);
6155 int scale = U.arrayIndexScale(ak);
6156 if ((scale & (scale - 1)) != 0)
6157 throw new Error("data type scale not a power of two");
6158 ASHIFT = 31 - Integer.numberOfLeadingZeros(scale);
6159 } catch (Exception e) {
6160 throw new Error(e);
6161 }
6162 }
6163
6164 /**
6165 * Returns a sun.misc.Unsafe. Suitable for use in a 3rd party package.
6166 * Replace with a simple call to Unsafe.getUnsafe when integrating
6167 * into a jdk.
6168 *
6169 * @return a sun.misc.Unsafe
6170 */
6171 private static sun.misc.Unsafe getUnsafe() {
6172 try {
6173 return sun.misc.Unsafe.getUnsafe();
6174 } catch (SecurityException tryReflectionInstead) {}
6175 try {
6176 return java.security.AccessController.doPrivileged
6177 (new java.security.PrivilegedExceptionAction<sun.misc.Unsafe>() {
6178 public sun.misc.Unsafe run() throws Exception {
6179 Class<sun.misc.Unsafe> k = sun.misc.Unsafe.class;
6180 for (java.lang.reflect.Field f : k.getDeclaredFields()) {
6181 f.setAccessible(true);
6182 Object x = f.get(null);
6183 if (k.isInstance(x))
6184 return k.cast(x);
6185 }
6186 throw new NoSuchFieldError("the Unsafe");
6187 }});
6188 } catch (java.security.PrivilegedActionException e) {
6189 throw new RuntimeException("Could not initialize intrinsics",
6190 e.getCause());
6191 }
6192 }
6193 }