Parent Directory
| 
Revision Log
Revision 1.33 - (view) (download) (as text)
| 1 : | dl | 1.1 | <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN"> |
| 2 : | <html> <head> | ||
| 3 : | <title>Concurrency Utilities</title> | ||
| 4 : | jsr166 | 1.32 | <!-- |
| 5 : | Written by Doug Lea with assistance from members of JCP JSR-166 | ||
| 6 : | Expert Group and released to the public domain, as explained at | ||
| 7 : | http://creativecommons.org/licenses/publicdomain | ||
| 8 : | --> | ||
| 9 : | dl | 1.1 | </head> |
| 10 : | |||
| 11 : | <body> | ||
| 12 : | |||
| 13 : | dl | 1.3 | <p> Utility classes commonly useful in concurrent programming. This |
| 14 : | package includes a few small standardized extensible frameworks, as | ||
| 15 : | well as some classes that provide useful functionality and are | ||
| 16 : | otherwise tedious or difficult to implement. Here are brief | ||
| 17 : | descriptions of the main components. See also the <tt>locks</tt> and | ||
| 18 : | <tt>atomic</tt> packages. | ||
| 19 : | dl | 1.1 | |
| 20 : | <h2>Executors</h2> | ||
| 21 : | |||
| 22 : | dl | 1.12 | <b>Interfaces.</b> {@link java.util.concurrent.Executor} is a simple |
| 23 : | standardized interface for defining custom thread-like subsystems, | ||
| 24 : | including thread pools, asynchronous IO, and lightweight task | ||
| 25 : | frameworks. Depending on which concrete Executor class is being used, | ||
| 26 : | tasks may execute in a newly created thread, an existing | ||
| 27 : | task-execution thread, or the thread calling <tt>execute()</tt>, and | ||
| 28 : | may execute sequentially or concurrently. {@link | ||
| 29 : | java.util.concurrent.ExecutorService} provides a more complete | ||
| 30 : | asynchronous task execution framework. An ExecutorService manages | ||
| 31 : | dl | 1.14 | queuing and scheduling of tasks, and allows controlled shutdown. The |
| 32 : | dl | 1.12 | {@link java.util.concurrent.ScheduledExecutorService} subinterface |
| 33 : | dl | 1.19 | and associated interfaces add support for delayed and periodic task execution. |
| 34 : | dl | 1.12 | ExecutorServices provide methods arranging asynchronous execution of |
| 35 : | any function expressed as {@link java.util.concurrent.Callable}, the | ||
| 36 : | result-bearing analog of {@link java.lang.Runnable}. A {@link | ||
| 37 : | java.util.concurrent.Future} returns the results of a function, allows | ||
| 38 : | dl | 1.16 | determination of whether execution has completed, and provides a means to |
| 39 : | jsr166 | 1.20 | cancel execution. A {@link java.util.concurrent.RunnableFuture} is |
| 40 : | dl | 1.19 | a Future that possesses a <tt>run</tt> method that upon execution, |
| 41 : | sets its results. | ||
| 42 : | dl | 1.12 | |
| 43 : | <p> | ||
| 44 : | |||
| 45 : | dl | 1.16 | <b>Implementations.</b> Classes {@link |
| 46 : | dl | 1.12 | java.util.concurrent.ThreadPoolExecutor} and {@link |
| 47 : | dl | 1.16 | java.util.concurrent.ScheduledThreadPoolExecutor} provide tunable, |
| 48 : | flexible thread pools. The {@link java.util.concurrent.Executors} | ||
| 49 : | class provides factory methods for the most common kinds and | ||
| 50 : | configurations of Executors, as well as a few utility methods for | ||
| 51 : | using them. Other utilities based on Executors include the concrete | ||
| 52 : | class {@link java.util.concurrent.FutureTask} providing a common | ||
| 53 : | extensible implementation of Futures, and {@link | ||
| 54 : | dl | 1.12 | java.util.concurrent.ExecutorCompletionService}, that assists in |
| 55 : | coordinating the processing of groups of asynchronous tasks. | ||
| 56 : | dl | 1.1 | |
| 57 : | <h2>Queues</h2> | ||
| 58 : | |||
| 59 : | dl | 1.6 | The java.util.concurrent {@link |
| 60 : | dl | 1.1 | java.util.concurrent.ConcurrentLinkedQueue} class supplies an |
| 61 : | dl | 1.6 | efficient scalable thread-safe non-blocking FIFO queue. Five |
| 62 : | implementations in java.util.concurrent support the extended {@link | ||
| 63 : | java.util.concurrent.BlockingQueue} interface, that defines blocking | ||
| 64 : | versions of put and take: {@link | ||
| 65 : | dl | 1.1 | java.util.concurrent.LinkedBlockingQueue}, {@link |
| 66 : | java.util.concurrent.ArrayBlockingQueue}, {@link | ||
| 67 : | java.util.concurrent.SynchronousQueue}, {@link | ||
| 68 : | java.util.concurrent.PriorityBlockingQueue}, and {@link | ||
| 69 : | dl | 1.2 | java.util.concurrent.DelayQueue}. The different classes cover the most |
| 70 : | common usage contexts for producer-consumer, messaging, parallel | ||
| 71 : | dl | 1.17 | tasking, and related concurrent designs. The {@link |
| 72 : | java.util.concurrent.BlockingDeque} interface extends | ||
| 73 : | <tt>BlockingQueue</tt> to support both FIFO and LIFO (stack-based) | ||
| 74 : | operations. Class {@link java.util.concurrent.LinkedBlockingDeque} | ||
| 75 : | provides an implementation. | ||
| 76 : | dl | 1.1 | |
| 77 : | |||
| 78 : | <h2>Timing</h2> | ||
| 79 : | |||
| 80 : | The {@link java.util.concurrent.TimeUnit} class provides multiple | ||
| 81 : | dl | 1.2 | granularities (including nanoseconds) for specifying and controlling |
| 82 : | jsr166 | 1.24 | time-out based operations. Most classes in the package contain |
| 83 : | dl | 1.9 | operations based on time-outs in addition to indefinite waits. In all |
| 84 : | cases that time-outs are used, the time-out specifies the minimum time | ||
| 85 : | that the method should wait before indicating that it | ||
| 86 : | jsr166 | 1.24 | timed-out. Implementations make a "best effort" to detect |
| 87 : | time-outs as soon as possible after they occur. However, an indefinite | ||
| 88 : | dl | 1.9 | amount of time may elapse between a time-out being detected and a |
| 89 : | jsr166 | 1.24 | thread actually executing again after that time-out. All methods |
| 90 : | dl | 1.18 | that accept timeout parameters treat values less than or equal to |
| 91 : | jsr166 | 1.24 | zero to mean not to wait at all. To wait "forever", you can use |
| 92 : | dl | 1.18 | a value of <tt>Long.MAX_VALUE</tt>. |
| 93 : | dholmes | 1.4 | |
| 94 : | dl | 1.1 | <h2>Synchronizers</h2> |
| 95 : | |||
| 96 : | dl | 1.13 | Four classes aid common special-purpose synchronization idioms. |
| 97 : | dl | 1.10 | {@link java.util.concurrent.Semaphore} is a classic concurrency tool. |
| 98 : | dl | 1.16 | {@link java.util.concurrent.CountDownLatch} is a very simple yet very |
| 99 : | dl | 1.11 | common utility for blocking until a given number of signals, events, |
| 100 : | or conditions hold. A {@link java.util.concurrent.CyclicBarrier} is a | ||
| 101 : | dl | 1.13 | resettable multiway synchronization point useful in some styles of |
| 102 : | dl | 1.11 | parallel programming. An {@link java.util.concurrent.Exchanger} allows |
| 103 : | dl | 1.13 | two threads to exchange objects at a rendezvous point, and is useful |
| 104 : | in several pipeline designs. | ||
| 105 : | dl | 1.1 | |
| 106 : | <h2>Concurrent Collections</h2> | ||
| 107 : | |||
| 108 : | dl | 1.17 | Besides Queues, this package supplies Collection implementations |
| 109 : | designed for use in multithreaded contexts: | ||
| 110 : | {@link java.util.concurrent.ConcurrentHashMap}, | ||
| 111 : | {@link java.util.concurrent.ConcurrentSkipListMap}, | ||
| 112 : | {@link java.util.concurrent.ConcurrentSkipListSet}, | ||
| 113 : | {@link java.util.concurrent.CopyOnWriteArrayList}, and | ||
| 114 : | {@link java.util.concurrent.CopyOnWriteArraySet}. | ||
| 115 : | When many threads are expected to access a given collection, | ||
| 116 : | a <tt>ConcurrentHashMap</tt> is normally preferable to | ||
| 117 : | a synchronized <tt>HashMap</tt>, and a | ||
| 118 : | <tt>ConcurrentSkipListMap</tt> is normally preferable | ||
| 119 : | to a synchronized <tt>TreeMap</tt>. A | ||
| 120 : | <tt>CopyOnWriteArrayList</tt> is preferable to | ||
| 121 : | a synchronized <tt>ArrayList</tt> when the expected number of reads | ||
| 122 : | and traversals greatly outnumber the number of updates to a list. | ||
| 123 : | dl | 1.1 | |
| 124 : | dl | 1.6 | <p>The "Concurrent" prefix used with some classes in this package is a |
| 125 : | shorthand indicating several differences from similar "synchronized" | ||
| 126 : | dl | 1.1 | classes. For example <tt>java.util.Hashtable</tt> and |
| 127 : | <tt>Collections.synchronizedMap(new HashMap())</tt> are | ||
| 128 : | dl | 1.6 | synchronized. But {@link java.util.concurrent.ConcurrentHashMap} is |
| 129 : | "concurrent". A concurrent collection is thread-safe, but not | ||
| 130 : | governed by a single exclusion lock. In the particular case of | ||
| 131 : | ConcurrentHashMap, it safely permits any number of concurrent reads as | ||
| 132 : | dl | 1.9 | well as a tunable number of concurrent writes. "Synchronized" classes |
| 133 : | can be useful when you need to prevent all access to a collection via | ||
| 134 : | a single lock, at the expense of poorer scalability. In other cases in | ||
| 135 : | which multiple threads are expected to access a common collection, | ||
| 136 : | "concurrent" versions are normally preferable. And unsynchronized | ||
| 137 : | collections are preferable when either collections are unshared, or | ||
| 138 : | are accessible only when holding other locks. | ||
| 139 : | dl | 1.1 | |
| 140 : | <p> Most concurrent Collection implementations (including most Queues) | ||
| 141 : | also differ from the usual java.util conventions in that their Iterators | ||
| 142 : | provide <em>weakly consistent</em> rather than fast-fail traversal. A | ||
| 143 : | weakly consistent iterator is thread-safe, but does not necessarily | ||
| 144 : | freeze the collection while iterating, so it may (or may not) reflect | ||
| 145 : | any updates since the iterator was created. | ||
| 146 : | |||
| 147 : | brian | 1.26 | <a name="MemoryVisibility"> |
| 148 : | dl | 1.21 | <h2> Memory Consistency Properties </h2> |
| 149 : | |||
| 150 : | <a href="http://java.sun.com/docs/books/jls/third_edition/html/memory.html"> | ||
| 151 : | Chapter 17 of the Java Language Specification</a> defines the | ||
| 152 : | <i>happens-before</i> relation on memory operations such as reads and | ||
| 153 : | writes of shared variables. The results of a write by one thread are | ||
| 154 : | guaranteed to be visible to a read by another thread only if the write | ||
| 155 : | operation <i>happens-before</i> the read operation. The | ||
| 156 : | jsr166 | 1.31 | {@code synchronized} and {@code volatile} constructs, as well as the |
| 157 : | {@code Thread.start()} and {@code Thread.join()} methods, can form | ||
| 158 : | dl | 1.21 | <i>happens-before</i> relationships. In particular: |
| 159 : | |||
| 160 : | <ul> | ||
| 161 : | brian | 1.26 | <li>Each action in a thread <i>happens-before</i> every action in that |
| 162 : | dl | 1.21 | thread that comes later in the program's order. |
| 163 : | |||
| 164 : | jsr166 | 1.31 | <li>An unlock ({@code synchronized} block or method exit) of a |
| 165 : | monitor <i>happens-before</i> every subsequent lock ({@code synchronized} | ||
| 166 : | dl | 1.23 | block or method entry) of that same monitor. And because |
| 167 : | the <i>happens-before</i> relation is transitive, all actions | ||
| 168 : | brian | 1.27 | of a thread prior to unlocking <i>happen-before</i> all actions |
| 169 : | dl | 1.23 | subsequent to any thread locking that monitor. |
| 170 : | dl | 1.21 | |
| 171 : | jsr166 | 1.31 | <li>A write to a {@code volatile} field <i>happens-before</i> every |
| 172 : | dl | 1.23 | subsequent read of that same field. Writes and reads of |
| 173 : | jsr166 | 1.31 | {@code volatile} fields have similar memory consistency effects |
| 174 : | dl | 1.23 | as entering and exiting monitors, but do <em>not</em> entail |
| 175 : | mutual exclusion locking. | ||
| 176 : | dl | 1.21 | |
| 177 : | jsr166 | 1.31 | <li>A call to {@code start} on a thread <i>happens-before</i> any action in the |
| 178 : | dl | 1.21 | started thread. |
| 179 : | |||
| 180 : | brian | 1.26 | <li>All actions in a thread <i>happen-before</i> any other thread |
| 181 : | jsr166 | 1.31 | successfully returns from a {@code join} on that thread. |
| 182 : | dl | 1.21 | |
| 183 : | </ul> | ||
| 184 : | |||
| 185 : | dl | 1.23 | |
| 186 : | jsr166 | 1.31 | The methods of all classes in {@code java.util.concurrent} and its |
| 187 : | dl | 1.21 | subpackages extend these guarantees to higher-level |
| 188 : | synchronization. In particular: | ||
| 189 : | |||
| 190 : | <ul> | ||
| 191 : | |||
| 192 : | jsr166 | 1.31 | <li>Actions in a thread prior to placing an object into any concurrent |
| 193 : | collection <i>happen-before</i> actions subsequent to the access or | ||
| 194 : | removal of that element from the collection in another thread. | ||
| 195 : | |||
| 196 : | <li>Actions in a thread prior to the submission of a {@code Runnable} | ||
| 197 : | to an {@code Executor} <i>happen-before</i> its execution begins. | ||
| 198 : | Similarly for {@code Callables} submitted to an {@code ExecutorService}. | ||
| 199 : | |||
| 200 : | <li>Actions taken by the asynchronous computation represented by a | ||
| 201 : | {@code Future} <i>happen-before</i> actions subsequent to the | ||
| 202 : | retrieval of the result via {@code Future.get()} in another thread. | ||
| 203 : | dl | 1.21 | |
| 204 : | <li>Actions prior to "releasing" synchronizer methods such as | ||
| 205 : | jsr166 | 1.31 | {@code Lock.unlock}, {@code Semaphore.release}, and |
| 206 : | {@code CountDownLatch.countDown} <i>happen-before</i> actions | ||
| 207 : | subsequent to a successful "acquiring" method such as | ||
| 208 : | {@code Lock.lock}, {@code Semaphore.acquire}, | ||
| 209 : | {@code Condition.await}, and {@code CountDownLatch.await} on the | ||
| 210 : | same synchronizer object in another thread. | ||
| 211 : | dl | 1.21 | |
| 212 : | brian | 1.27 | <li>For each pair of threads that successfully exchange objects via |
| 213 : | jsr166 | 1.31 | an {@code Exchanger}, actions prior to the {@code exchange()} |
| 214 : | brian | 1.27 | in each thread <i>happen-before</i> those subsequent to the |
| 215 : | jsr166 | 1.31 | corresponding {@code exchange()} in another thread. |
| 216 : | brian | 1.26 | |
| 217 : | jsr166 | 1.31 | <li>Actions prior to calling {@code CyclicBarrier.await} |
| 218 : | brian | 1.26 | <i>happen-before</i> actions performed by the barrier action, and |
| 219 : | actions performed by the barrier action <i>happen-before</i> actions | ||
| 220 : | jsr166 | 1.31 | subsequent to a successful return from the corresponding {@code await} |
| 221 : | brian | 1.27 | in other threads. |
| 222 : | dl | 1.21 | |
| 223 : | </ul> | ||
| 224 : | |||
| 225 : | dl | 1.15 | @since 1.5 |
| 226 : | |||
| 227 : | dl | 1.1 | </body> </html> |
| Doug Lea | ViewVC Help |
| Powered by ViewVC 1.0.8 |