学习JUC源码(2)——自定义同步组件
前言
在之前的博文(学习JUC源码(1)——AQS同步队列(源码分析结合图文理解))中,已经介绍了AQS同步队列的相关原理与概念,这里为了再加深理解ReentranLock等源码,模仿构造同步组件的基本模式,编写不可重入的互斥锁Mutex与指定共享线程数量的共享锁。MySharedLock。
主要参考资料《Java并发编程艺术》(有需要的小伙伴可以找我,我这里只有电子PDF)同时结合ReentranLock、AQS等源码。
一、构造同步组件的模式
丛概念方层面,在中,我们知道锁与同步器的相关概念:
- 同步器是实现锁的关键,在锁的实现中聚合同步器,利用同步器实现锁的语义;
- 锁是面向使用者的,提供锁交互的实现;
- 同步器是面向锁的实现者,简化了锁的实现方式,屏蔽了同步状态管理、线程排队、等待/唤醒等底层操作。
从代码层面,同步器是基于模板模式实现的,可以通过可重写的方法中的随便一个窥探:
/** * 模板方法: * protected关键字 * 没有任何实现 * @param arg * @return */ protected boolean tryAcquire(int arg) { throw new UnsupportedOperationException(); }
也就是需要进行以下几步:
1)继承同步器重写指定方法(idea中extends AQS点击快捷键ctrl+O即可显示)
- tryAcquire(int arg):独占式获取同步状态;
- tryRelease(int arg):独占式释放同步状态;
- tryAcquireShared(int arg):共享式获取同步状态,返回大于0的值表示获取成功,否则失败
- tryReleaseShared(int arg):共享式释放锁
- isHeldExclusively():当前线程是否在独占模式下被线程占用,一般该方法表示是否被当前线程占用
2)随后将同步器组合在自定义同步组件的实现中,即定义内部类Syn继承AQS,在Syn中重写AQS方法:
public class Sync extends AbstractQueuedSynchronizer{ @Override protected boolean tryAcquire(int arg) { final Thread current = Thread.currentThread(); if (compareAndSetState(0, 1)) { // 获取成功之后,当前线程是该锁的持有者,不需要再可重入数 setExclusiveOwnerThread(current); return true; } return false; } @Override protected boolean tryRelease(int arg) { if (getState() == 0) { throw new IllegalMonitorStateException(); } setExclusiveOwnerThread(null); setState(0); return true; } @Override protected boolean isHeldExclusively() { return getState() == 1; } // 返回Condition,每个Condition都包含了一个队列 Condition newCondition() { return new ConditionObject(); } }
3)最后调用同步器提供的模板方法,即同步组件类实现Lock方法之后,在lock/unlock方法中调用内部类Syn的方法acquire(int arg)等方法
public class Mutex implements Lock { ........ @Override public void lock() { sync.acquire(1); } @Override public void unlock() { sync.release(1); } ........ }
具体请看下面的实践部分
二、互斥不可重入锁
在我之前写过的博文中(详解Java锁的升级与对比(1)——锁的分类与细节(结合部分源码))介绍可重入锁与不可重入锁的区别时,就写到JUC中没有不可重入锁的具体实现,但是可以类比,现在呢,我们可以做到实现了,具体看下面代码,模式完全符合依赖Lock与AQS构造同步组件模式。
(1)Mutex代码实现(核心关键实现已经在代码中注释)
public class Mutex implements Lock { private final Sync sync = new Sync(); public class Sync extends AbstractQueuedSynchronizer{ @Override protected boolean tryAcquire(int arg) { final Thread current = Thread.currentThread(); if (compareAndSetState(0, 1)) { // 获取成功之后,当前线程是该锁的持有者,不需要再可重入数 setExclusiveOwnerThread(current); return true; } return false; } @Override protected boolean tryRelease(int arg) { if (getState() == 0) { throw new IllegalMonitorStateException(); } setExclusiveOwnerThread(null); setState(0); return true; } @Override protected boolean isHeldExclusively() { return getState() == 1; } // 返回Condition,每个Condition都包含了一个队列 Condition newCondition() { return new ConditionObject(); } } @Override public void lock() { sync.acquire(1); } @Override public void unlock() { sync.release(1); } @Override public void lockInterruptibly() throws InterruptedException { } @Override public boolean tryLock() { return false; } @Override public boolean tryLock(long time, TimeUnit unit) throws InterruptedException { return false; } @Override public Condition newCondition() { return null; } }
其中核心代码就是重写的两个方法:
- tryAcquire(int arg)方法:主要是设置同独占式更新同步状态,CAS实现state+1
- tryRelease(int arg)方法:独占式释放同步状态,释放锁持有
(2)测试Demo
public class MutexDemo { @Test public void test(){ final Mutex lock = new Mutex(); class Worker extends Thread { @Override public void run() { // 一直不停在获取锁 while (true) { lock.lock(); try { System.out.println(Thread.currentThread().getName() +" hold lock, "+new Date()); Thread.sleep(1000); } catch (InterruptedException e) { e.printStackTrace(); } finally { lock.unlock(); System.out.println(Thread.currentThread().getName() +" release lock, "+new Date()); } } } } for (int i = 0; i < 10; i++) { Worker worker = new Worker(); // 以守护进程运行,VM退出不影响运行,这里只是为了一个打印效果,去掉注释一直打印 worker.setDaemon(true); worker.start(); } // 每隔一秒换行 for (int j = 0; j < 10; j++) { try { Thread.sleep(1000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(); } } }
(3)运行结果
Thread-0 hold lock, Tue Dec 08 16:26:42 CST 2020 Thread-0 release lock, Tue Dec 08 16:26:43 CST 2020 Thread-1 hold lock, Tue Dec 08 16:26:43 CST 2020 Thread-2 hold lock, Tue Dec 08 16:26:44 CST 2020 Thread-1 release lock, Tue Dec 08 16:26:44 CST 2020 Thread-3 hold lock, Tue Dec 08 16:26:45 CST 2020 Thread-2 release lock, Tue Dec 08 16:26:45 CST 2020 Thread-3 release lock, Tue Dec 08 16:26:46 CST 2020 Thread-4 hold lock, Tue Dec 08 16:26:46 CST 2020 Thread-4 release lock, Tue Dec 08 16:26:47 CST 2020 Thread-6 hold lock, Tue Dec 08 16:26:47 CST 2020 Thread-7 hold lock, Tue Dec 08 16:26:48 CST 2020 Thread-6 release lock, Tue Dec 08 16:26:48 CST 2020 Thread-7 release lock, Tue Dec 08 16:26:49 CST 2020 Thread-5 hold lock, Tue Dec 08 16:26:49 CST 2020 Thread-8 hold lock, Tue Dec 08 16:26:50 CST 2020 Thread-5 release lock, Tue Dec 08 16:26:50 CST 2020 Thread-8 release lock, Tue Dec 08 16:26:51 CST 2020 Thread-9 hold lock, Tue Dec 08 16:26:51 CST 2020
(4)结果分析
互斥锁的核心就是同一个同步状态只能被一个线程持有,其它线程等待持有线程释放才能竞争获取。截图一开始的运行结果分析:
Thread-0 hold lock, Tue Dec 08 16:26:42 CST 2020 Thread-0 release lock, Tue Dec 08 16:26:43 CST 2020 Thread-1 hold lock, Tue Dec 08 16:26:43 CST 2020 Thread-2 hold lock, Tue Dec 08 16:26:44 CST 2020 Thread-1 release lock, Tue Dec 08 16:26:44 CST 2020
10个线程不断竞争锁,一开始Thread-0在08 16:26:42获取到锁,持有锁1秒后在释放16:26:43时释放,同时Thread-1立马获取到锁,1秒后于16:26:44释放锁,同时Thread-2立马获取到了锁......
根据输出结果来说,完全符合Mutex作为互斥锁这个功能:同一时刻只有一个线程持有锁(同步状态),其它线程等待释放后才能获取。
三、指定共享线程数目的共享锁
(1)代码实现(核心关键实现已经在代码中注释)
public class MyShareLock implements Lock { // 可以看到共享等待队列中的线程 public Collection<Thread> getSharedQueuedThreads(){ return syn.getSharedQueuedThreads(); } private final Syn syn = new Syn(2); private static final class Syn extends AbstractQueuedSynchronizer{ int newShareCount=0; Syn(int shareCount){ if (shareCount <= 0) { throw new IllegalArgumentException("share count must large than zero"); } // 设置初始共享同步状态 setState(shareCount); } /** * 共享锁指定数目 * @param reduceShareCount * @return */ @Override protected int tryAcquireShared(int reduceShareCount) { for (;;){ int currentShareCount = getState(); newShareCount = currentShareCount- reduceShareCount; if (newShareCount < 0 || compareAndSetState(currentShareCount,newShareCount)) { // newShareCount大于等于0才说明获取锁成功 if (newShareCount >= 0) { // System.out.println(Thread.currentThread().getName()+" hold lock, current share count is "+newShareCount+", "+new Date()); } // newShareCount小于0表示获取失败所以需要返回 // compareAndSetState(currentShareCount,newShareCount)为true自然表示成功需要返回 return newShareCount; } } } @Override protected boolean tryReleaseShared(int returnShareCount) { for (;;){ int currentShareCount = getState(); newShareCount = currentShareCount + returnShareCount; if (compareAndSetState(currentShareCount,newShareCount)) { // System.out.println(Thread.currentThread().getName() +" release lock, current share count is "+newShareCount+", "+new Date()); return true; } } } protected int getShareCount(){ return getState(); } } /** * 调用内部同步器Syn的acquireShare方法 */ @Override public void lock() { syn.acquireShared(1); } /** * 调用内部同步器Syn的releaseShared方法 */ @Override public void unlock() { syn.releaseShared(1); } @Override public void lockInterruptibly() throws InterruptedException { if (Thread.interrupted()) { throw new IllegalStateException(); } syn.acquireInterruptibly(1); } @Override public boolean tryLock() { return false; } @Override public boolean tryLock(long time, TimeUnit unit) throws InterruptedException { return false; } @Override public Condition newCondition() { return null; } }
(2)测试Demo
public class ShareLockTest { @Test public void test(){ final MyShareLock lock = new MyShareLock(); class Worker extends Thread { @Override public void run() { // 一直不停在获取锁 while (true) { lock.lock(); try { System.out.println(Thread.currentThread().getName() +" hold lock, "+new Date()); // System.out.println(lock.getSharedQueuedThreads()); Thread.sleep(1000); } catch (InterruptedException e) { e.printStackTrace(); } finally { lock.unlock(); System.out.println(Thread.currentThread().getName() +" release lock, "+new Date()); } } } } for (int i = 0; i < 10; i++) { Worker worker = new Worker(); // 以守护进程运行,VM退出不影响运行,这里只是为了一个打印效果,去掉注释一直打印 worker.setDaemon(true); worker.start(); } // 每隔一秒换行 for (int j = 0; j < 10; j++) { try { Thread.sleep(1000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println(); } } }
(3)运行结果(结果可能不同)
Thread-1 hold lock, Tue Dec 08 16:36:05 CST 2020 Thread-0 hold lock, Tue Dec 08 16:36:05 CST 2020 Thread-0 release lock, Tue Dec 08 16:36:06 CST 2020 Thread-4 hold lock, Tue Dec 08 16:36:06 CST 2020 Thread-1 release lock, Tue Dec 08 16:36:06 CST 2020 Thread-2 hold lock, Tue Dec 08 16:36:06 CST 2020 Thread-4 release lock, Tue Dec 08 16:36:07 CST 2020 Thread-2 release lock, Tue Dec 08 16:36:07 CST 2020 Thread-5 hold lock, Tue Dec 08 16:36:07 CST 2020 Thread-8 hold lock, Tue Dec 08 16:36:07 CST 2020 Thread-8 release lock, Tue Dec 08 16:36:08 CST 2020 Thread-3 hold lock, Tue Dec 08 16:36:08 CST 2020 Thread-9 hold lock, Tue Dec 08 16:36:08 CST 2020 Thread-5 release lock, Tue Dec 08 16:36:08 CST 2020 Thread-6 hold lock, Tue Dec 08 16:36:09 CST 2020 Thread-7 hold lock, Tue Dec 08 16:36:09 CST 2020 Thread-3 release lock, Tue Dec 08 16:36:09 CST 2020 Thread-9 release lock, Tue Dec 08 16:36:09 CST 2020 Thread-6 release lock, Tue Dec 08 16:36:10 CST 2020 Thread-1 hold lock, Tue Dec 08 16:36:10 CST 2020 Thread-0 hold lock, Tue Dec 08 16:36:10 CST 2020 Thread-7 release lock, Tue Dec 08 16:36:10 CST 2020 Thread-1 release lock, Tue Dec 08 16:36:11 CST 2020 Thread-2 hold lock, Tue Dec 08 16:36:11 CST 2020 Thread-0 release lock, Tue Dec 08 16:36:11 CST 2020 Thread-4 hold lock, Tue Dec 08 16:36:11 CST 2020 Thread-2 release lock, Tue Dec 08 16:36:12 CST 2020 Thread-8 hold lock, Tue Dec 08 16:36:12 CST 2020 Thread-5 hold lock, Tue Dec 08 16:36:12 CST 2020 Thread-4 release lock, Tue Dec 08 16:36:12 CST 2020 Thread-5 release lock, Tue Dec 08 16:36:13 CST 2020 Thread-9 hold lock, Tue Dec 08 16:36:13 CST 2020 Thread-3 hold lock, Tue Dec 08 16:36:13 CST 2020 Thread-8 release lock, Tue Dec 08 16:36:13 CST 2020 Thread-3 release lock, Tue Dec 08 16:36:14 CST 2020 Thread-7 hold lock, Tue Dec 08 16:36:14 CST 2020 Thread-9 release lock, Tue Dec 08 16:36:14 CST 2020 Thread-6 hold lock, Tue Dec 08 16:36:14 CST 2020
(4)结果分析
该指定共享线程数量N的共享锁的最终目的就是多个线程可以持有锁(同步状态),达到共享线程数量N(代码中默认为2)时,其它线程将进入Queue等待获取同步结果,同一时刻只能最多有N个线程持有锁。
同样地,我们分析开头运行结果:
Thread-1 hold lock, Tue Dec 08 16:36:05 CST 2020 Thread-0 hold lock, Tue Dec 08 16:36:05 CST 2020 Thread-0 release lock, Tue Dec 08 16:36:06 CST 2020 Thread-4 hold lock, Tue Dec 08 16:36:06 CST 2020 Thread-1 release lock, Tue Dec 08 16:36:06 CST 2020 Thread-2 hold lock, Tue Dec 08 16:36:06 CST 2020
10个线程不停竞争锁,一开始Thread-0与Thread-1在16:36:05时刻同时获取到了锁,此时已经达到共享数量的最大值,即N,之后持有锁1秒,Thread-0与Thread-1在16:36:06时刻立马释放锁,同时Thread-4与Thread-2立马退出等待队列立马竞争持有锁。
从结果来看,完全是符合ShareLock共享锁功能的:同一时刻最多允许N个线程持有锁,其它线程等待持有线程释放锁!