Java线程之如何分析死锁及避免死锁

共 27196字,需浏览 55分钟

 ·

2021-03-28 10:11

点击上方蓝色字体,选择“标星公众号”

优质文章,第一时间送达

什么是死锁

java中的死锁是一种编程情况,其中两个或多个线程被永久阻塞,Java死锁情况出现至少两个线程和两个或更多资源。

在这里,我们将写了一个简单的程序,它将导致java死锁场景,然后我们将分析它。


怎么实现死锁

下面我们一起看一个简单的死锁事例,在主线程中每隔5秒钟启动一个线程,每个线程的执行时间是3秒钟

package com.lkf.mulithread;

public class ThreadDeadlock {
    public static void main(String[] args) throws InterruptedException {
        Object obj1 = new Object();
        Object obj2 = new Object();
        Object obj3 = new Object();

        Thread t1 = new Thread(new SyncThread(obj1, obj2), "t1");
        Thread t2 = new Thread(new SyncThread(obj2, obj3), "t2");
        Thread t3 = new Thread(new SyncThread(obj3, obj1), "t3");

        t1.start();
        Thread.sleep(5000);
        t2.start();
        Thread.sleep(5000);
        t3.start();

    }

}

class SyncThread implements Runnable{
    private Object obj1;
    private Object obj2;

    public SyncThread(Object o1, Object o2){
        this.obj1=o1;
        this.obj2=o2;
    }
    @Override
    public void run() {
        String name = Thread.currentThread().getName();
        System.out.println(name + " acquiring lock on "+obj1);
        synchronized (obj1) {
            System.out.println(name + " acquired lock on "+obj1);
            work();
            System.out.println(name + " acquiring lock on "+obj2);
            synchronized (obj2) {
                System.out.println(name + " acquired lock on "+obj2);
                work();
            }
            System.out.println(name + " released lock on "+obj2);
        }
        System.out.println(name + " released lock on "+obj1);
        System.out.println(name + " finished execution.");
    }
    private void work() {
        try {
            Thread.sleep(30000);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
    }
}



运行结果:

t1 acquiring lock on java.lang.Object@25d849fd
t1 acquired lock on java.lang.Object@25d849fd
t2 acquiring lock on java.lang.Object@30298d1f
t2 acquired lock on java.lang.Object@30298d1f
t3 acquiring lock on java.lang.Object@11d32fc4
t3 acquired lock on java.lang.Object@11d32fc4
t1 acquiring lock on java.lang.Object@30298d1f
t2 acquiring lock on java.lang.Object@11d32fc4
t3 acquiring lock on java.lang.Object@25d849fd


可以看到每个线程中都能成功获取第一个对象的锁,最终却阻塞在了获取第二个对象,造成了线程之间的互相等待,形成了死锁


怎么排查死锁

我们通过jstack拿到该应用程序死锁的线程转储文件如下(获取线程转储文件方法):


2018-03-31 15:33:34
Full thread dump Java HotSpot(TM) 64-Bit Server VM (25.161-b12 mixed mode):

"RMI TCP Connection(2)-192.168.1.101" #22 daemon prio=9 os_prio=31 tid=0x00007fe2c81cc000 nid=0xa503 runnable [0x000070000ba7c000]
   java.lang.Thread.State: RUNNABLE
        at java.net.SocketInputStream.socketRead0(Native Method)
        at java.net.SocketInputStream.socketRead(SocketInputStream.java:116)
        at java.net.SocketInputStream.read(SocketInputStream.java:171)
        at java.net.SocketInputStream.read(SocketInputStream.java:141)
        at java.io.BufferedInputStream.fill(BufferedInputStream.java:246)
        at java.io.BufferedInputStream.read(BufferedInputStream.java:265)
        - locked <0x0000000797612b30> (a java.io.BufferedInputStream)
        at java.io.FilterInputStream.read(FilterInputStream.java:83)
        at sun.rmi.transport.tcp.TCPTransport.handleMessages(TCPTransport.java:550)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run0(TCPTransport.java:826)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.lambda$run$0(TCPTransport.java:683)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler$$Lambda$6/556824457.run(Unknown Source)
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run(TCPTransport.java:682)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - <0x0000000797612c78> (a java.util.concurrent.ThreadPoolExecutor$Worker)

"JMX server connection timeout 21" #21 daemon prio=9 os_prio=31 tid=0x00007fe2c8968800 nid=0xa703 in Object.wait() [0x000070000b97a000]
   java.lang.Thread.State: TIMED_WAITING (on object monitor)
        at java.lang.Object.wait(Native Method)
        at com.sun.jmx.remote.internal.ServerCommunicatorAdmin$Timeout.run(ServerCommunicatorAdmin.java:168)
        - locked <0x0000000797600720> (a [I)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"RMI Scheduler(0)" #20 daemon prio=9 os_prio=31 tid=0x00007fe2c81bc800 nid=0xa903 waiting on condition [0x000070000b877000]
   java.lang.Thread.State: TIMED_WAITING (parking)
        at sun.misc.Unsafe.park(Native Method)
        - parking to wait for  <0x0000000797618010> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
        at java.util.concurrent.locks.LockSupport.parkNanos(LockSupport.java:215)
        at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.awaitNanos(AbstractQueuedSynchronizer.java:2078)
        at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:1093)
        at java.util.concurrent.ScheduledThreadPoolExecutor$DelayedWorkQueue.take(ScheduledThreadPoolExecutor.java:809)
        at java.util.concurrent.ThreadPoolExecutor.getTask(ThreadPoolExecutor.java:1074)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1134)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"RMI TCP Connection(1)-192.168.1.101" #19 daemon prio=9 os_prio=31 tid=0x00007fe2c884e000 nid=0x5903 runnable [0x000070000b773000]
   java.lang.Thread.State: RUNNABLE
        at java.net.SocketInputStream.socketRead0(Native Method)
        at java.net.SocketInputStream.socketRead(SocketInputStream.java:116)
        at java.net.SocketInputStream.read(SocketInputStream.java:171)
        at java.net.SocketInputStream.read(SocketInputStream.java:141)
        at java.io.BufferedInputStream.fill(BufferedInputStream.java:246)
        at java.io.BufferedInputStream.read(BufferedInputStream.java:265)
        - locked <0x0000000797600eb8> (a java.io.BufferedInputStream)
        at java.io.FilterInputStream.read(FilterInputStream.java:83)
        at sun.rmi.transport.tcp.TCPTransport.handleMessages(TCPTransport.java:550)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run0(TCPTransport.java:826)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.lambda$run$0(TCPTransport.java:683)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler$$Lambda$6/556824457.run(Unknown Source)
        at java.security.AccessController.doPrivileged(Native Method)
        at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run(TCPTransport.java:682)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - <0x0000000797600ff8> (a java.util.concurrent.ThreadPoolExecutor$Worker)

"RMI TCP Accept-0" #18 daemon prio=9 os_prio=31 tid=0x00007fe2c8855000 nid=0x5703 runnable [0x000070000b671000]
   java.lang.Thread.State: RUNNABLE
        at java.net.PlainSocketImpl.socketAccept(Native Method)
        at java.net.AbstractPlainSocketImpl.accept(AbstractPlainSocketImpl.java:409)
        at java.net.ServerSocket.implAccept(ServerSocket.java:545)
        at java.net.ServerSocket.accept(ServerSocket.java:513)
        at sun.management.jmxremote.LocalRMIServerSocketFactory$1.accept(LocalRMIServerSocketFactory.java:52)
        at sun.rmi.transport.tcp.TCPTransport$AcceptLoop.executeAcceptLoop(TCPTransport.java:400)
        at sun.rmi.transport.tcp.TCPTransport$AcceptLoop.run(TCPTransport.java:372)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"DestroyJavaVM" #16 prio=5 os_prio=31 tid=0x00007fe2c90cf800 nid=0x2803 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"t3" #14 prio=5 os_prio=31 tid=0x00007fe2c7865000 nid=0x5503 waiting for monitor entry [0x000070000b56e000]
   java.lang.Thread.State: BLOCKED (on object monitor)
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x0000000797618188> (a java.lang.Object)
        - locked <0x0000000797618198> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"Attach Listener" #15 daemon prio=9 os_prio=31 tid=0x00007fe2c884d800 nid=0xb07 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"t2" #13 prio=5 os_prio=31 tid=0x00007fe2c884b800 nid=0x3e03 waiting for monitor entry [0x000070000b368000]
   java.lang.Thread.State: BLOCKED (on object monitor)
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x0000000797618198> (a java.lang.Object)
        - locked <0x00000007976181e8> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"t1" #12 prio=5 os_prio=31 tid=0x00007fe2c7885000 nid=0x4103 waiting for monitor entry [0x000070000b265000]
   java.lang.Thread.State: BLOCKED (on object monitor)
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x00000007976181e8> (a java.lang.Object)
        - locked <0x0000000797618188> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)

   Locked ownable synchronizers:
        - None

"Service Thread" #11 daemon prio=9 os_prio=31 tid=0x00007fe2c884c800 nid=0x4503 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"C1 CompilerThread2" #10 daemon prio=9 os_prio=31 tid=0x00007fe2c906c000 nid=0x4603 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"C2 CompilerThread1" #9 daemon prio=9 os_prio=31 tid=0x00007fe2c884a800 nid=0x3c03 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"C2 CompilerThread0" #8 daemon prio=9 os_prio=31 tid=0x00007fe2c806c800 nid=0x3a03 waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"JDWP Command Reader" #7 daemon prio=10 os_prio=31 tid=0x00007fe2c7811000 nid=0x4903 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"JDWP Event Helper Thread" #6 daemon prio=10 os_prio=31 tid=0x00007fe2c7810000 nid=0x3803 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"JDWP Transport Listener: dt_socket" #5 daemon prio=10 os_prio=31 tid=0x00007fe2c780f800 nid=0x4a07 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"Signal Dispatcher" #4 daemon prio=9 os_prio=31 tid=0x00007fe2c801d000 nid=0x3507 runnable [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

   Locked ownable synchronizers:
        - None

"Finalizer" #3 daemon prio=8 os_prio=31 tid=0x00007fe2c7820800 nid=0x3003 in Object.wait() [0x000070000a847000]
   java.lang.Thread.State: WAITING (on object monitor)
        at java.lang.Object.wait(Native Method)
        at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:143)
        - locked <0x0000000797618850> (a java.lang.ref.ReferenceQueue$Lock)
        at java.lang.ref.ReferenceQueue.remove(ReferenceQueue.java:164)
        at java.lang.ref.Finalizer$FinalizerThread.run(Finalizer.java:209)

   Locked ownable synchronizers:
        - None

"Reference Handler" #2 daemon prio=10 os_prio=31 tid=0x00007fe2c8017000 nid=0x2e03 in Object.wait() [0x000070000a744000]
   java.lang.Thread.State: WAITING (on object monitor)
        at java.lang.Object.wait(Native Method)
        at java.lang.Object.wait(Object.java:502)
        at java.lang.ref.Reference.tryHandlePending(Reference.java:191)
        - locked <0x0000000797618890> (a java.lang.ref.Reference$Lock)
        at java.lang.ref.Reference$ReferenceHandler.run(Reference.java:153)

   Locked ownable synchronizers:
        - None

"VM Thread" os_prio=31 tid=0x00007fe2c8014800 nid=0x5303 runnable 

"GC task thread#0 (ParallelGC)" os_prio=31 tid=0x00007fe2c8011800 nid=0x1b07 runnable 

"GC task thread#1 (ParallelGC)" os_prio=31 tid=0x00007fe2c8012000 nid=0x1d03 runnable 

"GC task thread#2 (ParallelGC)" os_prio=31 tid=0x00007fe2c8012800 nid=0x2a03 runnable 

"GC task thread#3 (ParallelGC)" os_prio=31 tid=0x00007fe2c9010000 nid=0x5403 runnable 

"VM Periodic Task Thread" os_prio=31 tid=0x00007fe2c781a000 nid=0x4303 waiting on condition 

JNI global references: 5038


Found one Java-level deadlock:
=============================
"t3":
  waiting to lock monitor 0x00007fe2c91c7198 (object 0x0000000797618188, a java.lang.Object),
  which is held by "t1"
"t1":
  waiting to lock monitor 0x00007fe2c9046568 (object 0x00000007976181e8, a java.lang.Object),
  which is held by "t2"
"t2":
  waiting to lock monitor 0x00007fe2c91c5b98 (object 0x0000000797618198, a java.lang.Object),
  which is held by "t3"

Java stack information for the threads listed above:
===================================================
"t3":
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x0000000797618188> (a java.lang.Object)
        - locked <0x0000000797618198> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)
"t1":
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x00000007976181e8> (a java.lang.Object)
        - locked <0x0000000797618188> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)
"t2":
        at com.lkf.mulithread.SyncThread.run(ThreadDeadlock.java:40)
        - waiting to lock <0x0000000797618198> (a java.lang.Object)
        - locked <0x00000007976181e8> (a java.lang.Object)
        at java.lang.Thread.run(Thread.java:748)

Found 1 deadlock.



为了分析死锁,我们需要注意那些被阻塞的线程,以及等待锁定的资源,每个资源有一个惟一的ID,我们可以找到哪个线程已经持有对象上的锁,比如:线程t3正在等待获取对象0x00007fe2c91c7198的锁,但是该对象已经被线程t1锁定。


怎么避免死锁

上边,我们分析了死锁的情况,发现了导致死锁的线程,我们需要修改代码以避免死锁的情况,那么怎么避免死锁呢


1、避免嵌套锁

这是死锁最常见的原因,如果您已经持有一个资源,请避免锁定另一个资源。如果只使用一个对象锁,则几乎不可能出现死锁情况,比如以下代码对上边的循环嵌套部分进行修改,则避免了死锁的情况:

public void run() {
        String name = Thread.currentThread().getName();
        System.out.println(name + " acquiring lock on " + obj1);
        synchronized (obj1) {
            System.out.println(name + " acquired lock on " + obj1);
            work();
        }
        System.out.println(name + " released lock on " + obj1);
        System.out.println(name + " acquiring lock on " + obj2);
        synchronized (obj2) {
            System.out.println(name + " acquired lock on " + obj2);
            work();
        }
        System.out.println(name + " released lock on " + obj2);

        System.out.println(name + " finished execution.");
    }


2、只锁需要的部分

只获对需要的资源加锁,例如在上面的程序中,我们锁定了完整的对象资源,但是如果我们只需要其中一个字段,那么我们应该只锁定那个特定的字段而不是完整的对象


3、避免无限期等待

如果两个线程使用 thread join 无限期互相等待也会造成死锁,我们可以设定等待的最大时间来避免这种情况。

————————————————

版权声明:本文为CSDN博主「IT码客」的原创文章,遵循CC 4.0 BY-SA版权协议,转载请附上原文出处链接及本声明。

原文链接:

https://blog.csdn.net/u010647035/article/details/79770534




锋哥最新SpringCloud分布式电商秒杀课程发布

👇👇👇

👆长按上方微信二维码 2 秒





感谢点赞支持下哈 

浏览 15
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报
评论
图片
表情
推荐
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报