写在前面
这个包对象是JDK 1.5,1.6 ,1.8 之后,新增的,大概2009年,新增了 ,枚举定义,基本类型包装,注解,加强for循环,等新特性,提高了Java对并发编程的支持,也扩大了Java的市场,前景
JDK 1.8 下 JUC 下 共有 17(atomic包下) + 10(lock包下) + 58 个对象组成;
在后续的Java 高版本中,有更改 17(atomic包下) + 10(lock包下) + 61个对象组成,多了的 3 个对象是
- java.util.concurrent.Flow
- java.util.concurrent.Helpers
- java.util.concurrent.SubmissionPublisher
一、lock 包下
这个包下共 10 个对象(接口、类)
如图,
下面简单介绍,这10 个对象
1.1、AbstractOwnableSynchronizer
可参考文章,介绍了AbstractOwnableSynchronizer和AbstractQueuedSynchronizer,链接
1.2、AbstractQueuedLongSynchronizer
1.3、AbstractQueuedSynchronizer
这个就是 简称 AQS 的东西,
java.util.concurrent包中很多类都依赖于这个类所提供队列式同步器,比如说常用的ReentranLock,Semaphore和CountDownLatch
代码示例 RoboVM(java 创建 IOS APP框架)
/** * Acquires in shared mode, aborting if interrupted. Implemented * by first checking interrupt status, then invoking at least once * {@link #tryAcquireShared}, returning on success. Otherwise the * thread is queued, possibly repeatedly blocking and unblocking, * invoking {@link #tryAcquireShared} until success or the thread * is interrupted. * @param arg the acquire argument. * This value is conveyed to {@link #tryAcquireShared} but is * otherwise uninterpreted and can represent anything * you like. * @throws InterruptedException if the current thread is interrupted */
public final void acquireSharedInterruptibly(int arg)
throws InterruptedException {
if (Thread.interrupted())
throw new InterruptedException();
if (tryAcquireShared(arg) < 0)
doAcquireSharedInterruptibly(arg);
}
1.4、Condition
在使用Lock之前,我们使用的最多的同步方式应该是synchronized关键字来实现同步方式了。配合Object的wait()、notify()系列方法可以实现等待/通知模式。Condition接口也提供了类似Object的监视器方法,与Lock配合可以实现等待/通知模式
简介,可参考,链接
1.4.1、apache-druid中使用示例
private static final TimeUnit TIME_UNIT = TimeUnit.MILLISECONDS;
@VisibleForTesting
final ArrayDeque<T> objects;
private final ReentrantLock lock;
private final Condition notEnough;
private final int maxSize;
...
...
@Nullable
private T pollObject(long timeoutMs) throws InterruptedException
{
long nanos = TIME_UNIT.toNanos(timeoutMs);
final ReentrantLock lock = this.lock;
lock.lockInterruptibly();
try {
while (objects.isEmpty()) {
if (nanos <= 0) {
return null;
}
nanos = notEnough.awaitNanos(nanos);
}
return objects.pop();
}
finally {
lock.unlock();
}
}
1.5、Lock
Jdk中核心实现类包括一下
代码示例 apache/hive 任务调度中
private final Lock scheduleLock = new ReentrantLock()
....
private void trySchedulingPendingTasks() {
scheduleLock.lock();
try {
pendingScheduleInvocations.set(true);
scheduleCondition.signal();
} finally {
scheduleLock.unlock();
}
}
1.6、LockSupport
代码示例 h2oai/h2o-2
public boolean block() {
if (isReleasable())
return true;
else if (!timed)
LockSupport.park(this);
else if (nanos > 0)
LockSupport.parkNanos(this, nanos);
return isReleasable();
}
}
1.7、ReadWriteLock
Synchronized存在明显的一个性能问题就是读与读之间互斥,
ReadWriteLock管理一组锁,一个是只读的锁,一个是写锁。可以做到读和读互不影响,读和写互斥,写和写互斥,提高读写的效率
Java并发库中ReetrantReadWriteLock实现了ReadWriteLock接口并添加了可重入的特性。
代码示例 apache/rocketMQ
private ReadWriteLock readWriteLock = new ReentrantReadWriteLock();
...
...
public String getAllConfigsFormatString() {
try {
readWriteLock.readLock().lockInterruptibly();
try {
return getAllConfigsInternal();
} finally {
readWriteLock.readLock().unlock();
}
} catch (InterruptedException e) {
log.error("getAllConfigsFormatString lock error");
}
return null;
}
代码示例 jenkinsci/jenkins
/*protected*/ void start(WorkUnit task) {
lock.writeLock().lock();
try {
this.workUnit = task;
super.start();
started = true;
} finally {
lock.writeLock().unlock();
}
}
1.8、ReentrantLock
java除了使用关键字synchronized外,还可以使用ReentrantLock实现独占锁的功能。而且ReentrantLock相比synchronized而言功能更加丰富,使用起来更为灵活,也更适合复杂的并发场景
构造参数为 true代表公平锁,会根据入排队时间,优先获取锁
@Override
public Lock get() {
return new ReentrantLock(false);
}
});
代码示例 alibaba/druid
public void setEnable(boolean enable) {
lock.lock();
try {
this.enable = enable;
if (!enable) {
notEmpty.signalAll();
notEmptySignalCount++;
}
} finally {
lock.unlock();
}
}
1.9、ReentrantReadWriteLock
代码示例 apache / hive
private final ConcurrentMap<QueryIdentifier, ReadWriteLock> dagSpecificLocks = new ConcurrentHashMap<>();
...
...
private ReentrantReadWriteLock getDagLock(QueryIdentifier queryIdentifier) {
lock.lock();
try {
ReentrantReadWriteLock dagLock = dagSpecificLocks.get(queryIdentifier);
if (dagLock == null) {
dagLock = new ReentrantReadWriteLock();
dagSpecificLocks.put(queryIdentifier, dagLock);
}
return dagLock;
} finally {
lock.unlock();
}
}
代码示例 oracle/opengrok
private RuntimeEnvironment() {
configuration = new Configuration();
configLock = new CloseableReentrantReadWriteLock();
watchDog = new WatchDogService();
lzIndexerParallelizer = LazilyInstantiate.using(() ->
new IndexerParallelizer(this));
lzSearchExecutor = LazilyInstantiate.using(() -> newSearchExecutor());
lzRevisionExecutor = LazilyInstantiate.using(() -> newRevisionExecutor());
}
...
...
/** * Add repositories to the list. * @param repositories list of repositories */
public void addRepositories(List<RepositoryInfo> repositories) {
Lock writeLock = configLock.writeLock();
try {
writeLock.lock();
configuration.addRepositories(repositories);
} finally {
writeLock.unlock();
}
}
1.10、StampedLock
这个对象是 JDK 1.8 之后出现的对象,作为读写锁
StampedLock和ReadWriteLock相比,改进之处在于:读的过程中也允许获取写锁后写入!这样一来,我们读的数据就可能不一致,所以,需要一点额外的代码来判断读的过程中是否有写入,这种读锁是一种乐观锁。
乐观锁的意思就是乐观地估计读的过程中大概率不会有写入,因此被称为乐观锁。反过来,悲观锁则是读的过程中拒绝有写入,也就是写入必须等待。显然乐观锁的并发效率更高,但一旦有小概率的写入导致读取的数据不一致,需要能检测出来,再读一遍就行。
代码示例 apache/pulsar
private final StampedLock rwLock = new StampedLock();
...
private final ArrayList<Item> heap = Lists.newArrayList();
...
public boolean isEmpty() {
long stamp = rwLock.tryOptimisticRead();
boolean isEmpty = heap.isEmpty();
if (!rwLock.validate(stamp)) {
// Fallback to read lock
stamp = rwLock.readLock();
try {
isEmpty = heap.isEmpty();
} finally {
rwLock.unlockRead(stamp);
}
}
return isEmpty;
}
二、atomic 包下 17 个对象
待补充
这里只简单总结了相关 java 并发下的常用对象和使用示例