问题:如何实现死锁。
关键:
1 两个线程ta、tb
2 两个对象a、b
3 ta拥有a的锁,同时在这个锁定的过程中,需要b的锁;tb拥有b的锁,同时在这个锁定的过程中,需要a的锁;
关键的实现难点是3, —— 所以说,死锁也不是那么容易出现的吧。。
实现方式synchronized、Lock 等等
死锁例子1 采用了不同类的两个对象。 原理是: 两个线程尝试进入同一个需要对象锁的方法
package basic.thread; public class DL { public static void main(String[] args) { dla a = new dla(); a.start(); dlb b = new dlb(); b.start(); } } class dla extends Thread{ @Override public void run() { System.out.println("dla-------------"); Res.getInstace().aa(); } } class dlb extends Thread{ @Override public void run() { System.out.println("dlb-----------"); Res2.getInstace().bb(); //super.run(); } } class Res { private static Res res = new Res();; public static Res getInstace() { return res; } private Res() { } public synchronized void aa() { System.out.println("Res.aa()"); try { Thread.sleep(1000); Res2.getInstace().bb(); } catch (InterruptedException e) { e.printStackTrace(); } } } class Res2 { private static Res2 res = new Res2();; public static Res2 getInstace() { return res; } private Res2() { } public synchronized void bb() { System.out.println("Res2.bb()"); try { Thread.sleep(1000); Res.getInstace().aa(); } catch (InterruptedException e) { e.printStackTrace(); } } }
死锁例子2 原理是: 两个线程尝试进入同一个需要对象锁的方法。 这就说明的,只要是一个线程ta正在执行synchronized的方法ma(没有返回,这一点相当关键),下次别的线程进入同一个类的同一方法也好,别的synchronized方法也好,都是需要等待ta在ma的返回,就是要等ta在ma执行完了后才行,简单说就是不能进入。。—— 说起来好拗口。。。
package basic.thread; public class DL { public static void main(String[] args) { dla a = new dla(); a.start(); dlb b = new dlb(); b.start(); } } class dla extends Thread{ @Override public void run() { System.out.println("dla-------------"); Res.getInstace().aa(); } } class dlb extends Thread{ @Override public void run() { System.out.println("dlb-----------"); Res2.getInstace().bb(); //super.run(); } } class Res { private static Res res = new Res();; public static Res getInstace() { return res; } private Res() { } public synchronized void aa() { System.out.println("Res.aa()"); try { Thread.sleep(1000); Res2.getInstace().bbb(); } catch (InterruptedException e) { e.printStackTrace(); } } public synchronized void aaa() { System.out.println("Res.aaa()"); } } class Res2 { private static Res2 res = new Res2();; public static Res2 getInstace() { return res; } private Res2() { } public synchronized void bb() { System.out.println("Res2.bb() + " + Thread.currentThread().getName()); try { Thread.sleep(1000); Res.getInstace().aaa(); } catch (InterruptedException e) { e.printStackTrace(); } } public synchronized void bbb() { System.out.println("Res2.bbb()"); } }
死锁例子3 采用了同一个类的两个对象。(自旋锁? 这个例子相当不好,因为可能出现死循环! 看来还是两个不同类来实现死锁比较容易)
package basic.thread; public class DL { public static void main(String[] args) { Res res = new Res(); Res another = new Res(); res.setAnother(another); another.setAnother(res); dlb b1 = new dlb(res); dlb b2 = new dlb(another); b1.start(); b2.start(); } } class dlb extends Thread{ Res res; public dlb(Res res) { this.res = res; } @Override public void run() { System.out.println("dlb-----------"); res.aa(); } } class Res { Res another; public void setAnother(Res another) { this.another = another; } public synchronized void aa() { System.out.println("Res.aa()"); try { Thread.sleep(1000); another.aa(); } catch (InterruptedException e) { e.printStackTrace(); } } }
lock trylock
上例中把 synchronized 改成lock,则发生死锁,一样的效果。但是改成trylock ,则不会发生死锁!
package basic.thread; import java.util.concurrent.locks.Lock; import java.util.concurrent.locks.ReentrantLock; public class DL { public static void main(String[] args) { Res res = new Res(); Res another = new Res(); res.setAnother(another); another.setAnother(res); dlb b1 = new dlb(res); dlb b2 = new dlb(another); b1.start(); b2.start(); } } class dlb extends Thread{ Res res; public dlb(Res res) { this.res = res; } @Override public void run() { System.out.println("dlb-----------"); res.aa(); } } class Res { Res another; public void setAnother(Res another) { this.another = another; } ReentrantLock lock = new ReentrantLock(); public void aa() { lock.lock(); System.out.println("Res.aa()"); try { Thread.sleep(1000); another.aa(); } catch (InterruptedException e) { e.printStackTrace(); } lock.unlock(); } }
lock时候堆栈为:
"Thread-1" prio=6 tid=0x022eec00 nid=0x1e74 waiting on condition [0x0487f000..0x0487fae8] java.lang.Thread.State: WAITING (parking) at sun.misc.Unsafe.park(Native Method) - parking to wait for <0x22e75958> (a java.util.concurrent.locks.ReentrantLock$NonfairSync) at java.util.concurrent.locks.LockSupport.park(LockSupport.java:158) at java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:7 47) at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireQueued(AbstractQueuedSynchronizer.java:778) at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquire(AbstractQueuedSynchronizer.java:1114) at java.util.concurrent.locks.ReentrantLock$NonfairSync.lock(ReentrantLock.java:186) at java.util.concurrent.locks.ReentrantLock.lock(ReentrantLock.java:262) at basic.thread.Res.aa(DL.java:57) at basic.thread.Res.aa(DL.java:61) at basic.thread.dlb.run(DL.java:44) "Thread-0" prio=6 tid=0x022ee400 nid=0x2c50 waiting on condition [0x047ef000..0x047efb68] java.lang.Thread.State: WAITING (parking) at sun.misc.Unsafe.park(Native Method) - parking to wait for <0x22e75a50> (a java.util.concurrent.locks.ReentrantLock$NonfairSync) at java.util.concurrent.locks.LockSupport.park(LockSupport.java:158) at java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:7 47) at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireQueued(AbstractQueuedSynchronizer.java:778) at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquire(AbstractQueuedSynchronizer.java:1114) at java.util.concurrent.locks.ReentrantLock$NonfairSync.lock(ReentrantLock.java:186) at java.util.concurrent.locks.ReentrantLock.lock(ReentrantLock.java:262) at basic.thread.Res.aa(DL.java:57) at basic.thread.Res.aa(DL.java:61) at basic.thread.dlb.run(DL.java:44) ----
Found one Java-level deadlock:
=============================
"Thread-1":
waiting for ownable synchronizer 0x22e75958, (a java.util.concurrent.locks.ReentrantLock$NonfairSync), —— 注意这里的NonfairSync,非公平同步锁。
which is held by "Thread-0"
"Thread-0":
waiting for ownable synchronizer 0x22e75a50, (a java.util.concurrent.locks.ReentrantLock$NonfairSync),
which is held by "Thread-1"
这个时候的死锁,其实跟synchronized时候的锁是不同的,但是效果是一样的。
打印:
dlb-----------
Res.aa()
dlb-----------
Res.aa()
之后就一直‘卡住’了
tryLock的时候,打印堆栈发现:
C:\Users\Administrator>jstack 16640 2014-07-03 15:48:34 Full thread dump Java HotSpot(TM) Client VM (11.2-b01 mixed mode, sharing): "DestroyJavaVM" prio=6 tid=0x01edc800 nid=0x3a54 waiting on condition [0x00000000..0x01eafd20] java.lang.Thread.State: RUNNABLE "Thread-1" prio=6 tid=0x01f93800 nid=0x35a8 waiting on condition [0x0480f000..0x0480fce8] java.lang.Thread.State: TIMED_WAITING (sleeping) at java.lang.Thread.sleep(Native Method) at basic.thread.Res.aa(DL.java:60) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.dlb.run(DL.java:44) "Thread-0" prio=6 tid=0x01f90c00 nid=0x2abc waiting on condition [0x0477f000..0x0477fd68] java.lang.Thread.State: TIMED_WAITING (sleeping) at java.lang.Thread.sleep(Native Method) at basic.thread.Res.aa(DL.java:60) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.Res.aa(DL.java:61) at basic.thread.dlb.run(DL.java:44) "Low Memory Detector" daemon prio=6 tid=0x01f68000 nid=0x48d0 runnable [0x00000000..0x00000000] java.lang.Thread.State: RUNNABLE "CompilerThread0" daemon prio=10 tid=0x01f63c00 nid=0x3a0c waiting on condition [0x00000000..0x045cf714] java.lang.Thread.State: RUNNABLE "Attach Listener" daemon prio=10 tid=0x01f95c00 nid=0x3728 waiting on condition [0x00000000..0x00000000] java.lang.Thread.State: RUNNABLE "Signal Dispatcher" daemon prio=10 tid=0x01f95000 nid=0x365c runnable [0x00000000..0x00000000] java.lang.Thread.State: RUNNABLE "Finalizer" daemon prio=8 tid=0x01f1b800 nid=0x258c in Object.wait() [0x01cdf000..0x01cdfc68] java.lang.Thread.State: WAITING (on object monitor) at java.lang.Object.wait(Native Method) - waiting on <0x22960b28> (a java.lang.ref.ReferenceQueue$Lock) at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:116) - locked <0x22960b28> (a java.lang.ref.ReferenceQueue$Lock) at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:132) at java.lang.ref.Finalizer$FinalizerThread.run(Finalizer.java:159) "Reference Handler" daemon prio=10 tid=0x01f17000 nid=0x4584 in Object.wait() [0x003cf000..0x003cfce8] java.lang.Thread.State: WAITING (on object monitor) at java.lang.Object.wait(Native Method) - waiting on <0x22960a30> (a java.lang.ref.Reference$Lock) at java.lang.Object.wait(Object.java:485) at java.lang.ref.Reference$ReferenceHandler.run(Reference.java:116) - locked <0x22960a30> (a java.lang.ref.Reference$Lock) "VM Thread" prio=10 tid=0x01f15800 nid=0x4388 runnable "VM Periodic Task Thread" prio=10 tid=0x01f7f000 nid=0x4154 waiting on condition JNI global references: 596 ———— 这个时候并没有发生死锁,但是,有两个线程确确实实是阻塞了。 就效果来说一样的,但是原理不同。
而且后台打印不同:
dlb-----------
Res.aa()
dlb-----------
Res.aa()
Res.aa()
Res.aa()
Res.aa()
Res.aa()
Res.aa()
Res.aa()
Res.aa()
... 一直有打印,每隔一秒钟一次———— 死循环!!!!!。。说明,线程有一个轮询机制,一直在等待另一个线程释放锁,但是它又不释放,如是就一直“卡”在了那里。 —— 卡,其实就相当于了‘锁’!
好吧, 其实上面的例子有问题,因为,里面出现了一个死循环。。 而且,没有判断trylock 的返回值。
使用这个例子吧,但是,惊人的发现每次的结果不一样!: 有时候很快就两个end了,有时候一个end,有时候根本没有end——每次的tryLock都能返回true ,太奇怪了吧,如上例一样一直打印
package basic.thread; import java.util.concurrent.locks.Lock; import java.util.concurrent.locks.ReentrantLock; public class DL { public static void main(String[] args) { Res res = new Res(); Res another = new Res(); res.setAnother(another); another.setAnother(res); dlb b1 = new dlb(res); dlb b2 = new dlb(another); b1.start(); b2.start(); } } class dlb extends Thread{ Res res; public dlb(Res res) { this.res = res; } @Override public void run() { System.out.println("dlb-----------"); res.aa(); } } class Res { Res another; public void setAnother(Res another) { this.another = another; } ReentrantLock lock = new ReentrantLock(); public void aa() { boolean boo = lock.tryLock(); if(boo){ System.out.println("Res.aa()aaaaaaaa"); try { Thread.sleep(1000); another.aa(); } catch (InterruptedException e) { e.printStackTrace(); } lock.unlock(); } else { System.out.println("Res.aa() end !! "); } } }
好吧,其实上面的例子都不太好。
总结:
1 一般使用lock即可。
2 使用tryLock的话,需要判断其返回值,否则就达不到锁的效果。 应该如下,注意 unlock 必须要放在if里面
boolean tryLock = lock.tryLock(); System.out.println(" trylock : " + tryLock); if(tryLock) { do sth. lock.unlock(); ---- unlock 必须要放在if里面,否则如果tryLock为false的话,报错java.lang.IllegalMonitorStateException。 因为如果false,则lock 的holder 并非当前线程。。。 很明显的道理吧! }
3 tryLock方法无阻塞,立即返回(true/false), 所以,使用tryLock是不会造成死锁的!!! 但是,正因为如此,可能到不到我们想要的目的,比如,有时候时候我就是想锁定某个资源,否则就不能继续操作。 ———— 所以说tryLock适合无阻塞的场景!
4 java-tryLock可轮询可中断可定时的锁 —— 好奇怪的名字。。
参考http://aguang520.iteye.com/blog/815575 http://zfzaizheli.iteye.com/blog/1394823 (表示看不懂。。)等
5 注意到 ReentrantLock 的 lock() 返回值为void、 而tryLock返回boolean。
6 synchronized不能和lock等混用。 synchronized的用法见我另外一篇博客
7 ReentrantLock 是接口Lock的实现,其功能、底层原理远比synchronized复杂,
8 synchronized 可以加在方法前, 代码块前。 而且可以是静态的!
9 ReentrantReadWriteLock 的用法一般是:
ReentrantReadWriteLock lock = new ReentrantReadWriteLock();
ReadLock readLock = lock.readLock(); —— 用来对get,即查询操作加锁
WriteLock writeLock = lock.writeLock();—— 用来对add/update/delete,即新增修改删除操作加锁
—— 问题是,有必要分的这么细吗? 直接一个ReentrantLock 不就行了吗?
参见 http://www.cnblogs.com/sunwei2012/archive/2010/10/09/1846358.html
http://www.cnblogs.com/GnagWang/archive/2011/02/27/1966606.html ———— 写得非常非常非常详细。。
http://www.cnblogs.com/violinn/archive/2013/05/21/multiThread1.html 非常非常非常详细而且全面的关于java锁、同步、多线程的介绍
另外注意区别FileLock 类中也有 lock 、tryLock 方法,不过,当然,用法不同。