Java线程的6种状态分析
想起来写一下Java线程状态,还是源起与最近的一次问题定位,当时碰到一个偶先超时的问题,使用jstack命令打印出堆栈信息之后,例如
"transport-vert.x-eventloop-thread-11" #37 prio=5 os_prio=0 tid=0x00007f628d0f8800 nid=0x2a32 runnable [0x00007f62631f8000]
java.lang.Thread.State: RUNNABLE
at sun.nio.ch.EPollArrayWrapper.epollWait(Native Method)
at sun.nio.ch.EPollArrayWrapper.poll(EPollArrayWrapper.java:269)
at sun.nio.ch.EPollSelectorImpl.doSelect(EPollSelectorImpl.java:93)
at sun.nio.ch.SelectorImpl.lockAndDoSelect(SelectorImpl.java:86)
- locked <0x00000006f3cb4310> (a io.netty.channel.nio.SelectedSelectionKeySet)
- locked <0x00000006f3cb43d8> (a java.util.Collections$UnmodifiableSet)
- locked <0x00000006f3cb42c8> (a sun.nio.ch.EPollSelectorImpl)
at sun.nio.ch.SelectorImpl.select(SelectorImpl.java:97)
at io.netty.channel.nio.SelectedSelectionKeySetSelector.select(SelectedSelectionKeySetSelector.java:62)
at io.netty.channel.nio.NioEventLoop.select(NioEventLoop.java:755)
at io.netty.channel.nio.NioEventLoop.run(NioEventLoop.java:410)
at io.netty.util.concurrent.SingleThreadEventExecutor$5.run(SingleThreadEventExecutor.java:884)
at io.netty.util.concurrent.FastThreadLocalRunnable.run(FastThreadLocalRunnable.java:30)
at java.lang.Thread.run(Thread.java:748)
通过定位线程的状态,找到了错误的原因,也稍微费了些劲,所以想到把线程的一些状态再这里面总结一下
线程的6种状态
Java线程Thread
在package java.lang;
中可以找到,通过源码,我们可以看到其状态有如下6种
- NEW
- RUNNABLE
- BLOCKED
- WAITING
- TIMED_WAITING
- TERMINATED
下面分别解释一下各种状态
NEW
顾名思义,这个状态,只存在于线程刚创建,未start
之前,例如
MyThread thread = new MyThread();
System.out.println(thread.getState());
此时打印出来的状态就是NEW
RUNNABLE
这个状态的线程,其正在JVM中执行,但是这个"执行",不一定是真的在运行, 也有可能是在等待CPU资源。所以,在网上,有人把这个状态区分为READY和RUNNING两个,一个表示的start了,资源一到位随时可以执行,另一个表示真正的执行中,例如
MyThread thread = new MyThread(lock);
thread.start();
System.out.println(thread.getState());
BLOCKED
这个状态,一般是线程等待获取一个锁,来继续执行下一步的操作,比较经典的就是synchronized
关键字,这个关键字修饰的代码块或者方法,均需要获取到对应的锁,在未获取之前,其线程的状态就一直未BLOCKED,如果线程长时间处于这种状态下,我们就是当心看是否出现死锁的问题了。例如
public class MyThread extends Thread {
private byte[] lock = new byte[0];
public MyThread(byte[] lock) {
this.lock = lock;
}
@Override
public void run() {
synchronized (lock){
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.println("done");
}
}
}
public static void main(String[] args) throws InterruptedException {
byte[] lock = new byte[0];
MyThread thread1 = new MyThread(lock);
thread1.start();
MyThread thread2 = new MyThread(lock);
thread2.start();
Thread.sleep(1000);//等一会再检查状态
System.out.println(thread2.getState());
}
此时我们看到的输出的第二个线程的状态就是BLOCKED
WAITING
一个线程会进入这个状态,一定是执行了如下的一些代码,例如
- Object.wait()
- Thread.join()
- LockSupport.park()
当一个线程执行了Object.wait()的时候,它一定在等待另一个线程执行Object.notify()或者Object.notifyAll()。
或者一个线程thread,其在主线程中被执行了thread.join()的时候,主线程即会等待该线程执行完成。当一个线程执行了LockSupport.park()的时候,其在等待执行LockSupport.unpark(thread)。当该线程处于这种等待的时候,其状态即为WAITING。需要关注的是,这边的等待是没有时间限制的,当发现有这种状态的线程的时候,若其长时间处于这种状态,也需要关注下程序内部有无逻辑异常。例如
LockSupport.park()
public class MyThread extends Thread {
private byte[] lock = new byte[0];
public MyThread(byte[] lock) {
this.lock = lock;
}
@Override
public void run() {
LockSupport.park();
}
}
public static void main(String[] args) throws InterruptedException {
byte[] lock = new byte[0];
MyThread thread1 = new MyThread(lock);
thread1.start();
Thread.sleep(100);
System.out.println(thread1.getState());
LockSupport.unpark(thread1);
Thread.sleep(100);
System.out.println(thread1.getState());
}
输出WAITING和TERMINATED
Object.wait()
public class MyThread extends Thread {
private byte[] lock = new byte[0];
public MyThread(byte[] lock) {
this.lock = lock;
}
@Override
public void run() {
synchronized (lock){
try {
lock.wait(); //wait并允许其他线程同步lock
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
}
public static void main(String[] args)
throws InterruptedException {
byte[] lock = new byte[0];
MyThread thread1 = new MyThread(lock);
thread1.start();
Thread.sleep(100);
System.out.println(thread1.getState()); //这时候线程状态应为WAITING
synchronized (lock){
lock.notify(); //notify通知wait的线程
}
Thread.sleep(100);
System.out.println(thread1.getState());
}
输出WAITING和TERMINATED
Thread.join()
public class MyThread extends Thread {
private byte[] lock = new byte[0];
public MyThread(byte[] lock) {
this.lock = lock;
}
@Override
public void run() {
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public class MyThread1 extends Thread {
Thread thread;
public MyThread1(Thread thread) {
this.thread = thread;
}
@Override
public void run() {
try {
thread.join();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
public class Main {
public static void main(String[] args)
throws InterruptedException {
byte[] lock = new byte[0];
MyThread thread = new MyThread(lock);
thread.start();
MyThread1 thread1 = new MyThread1(thread);
thread1.start();
Thread.sleep(100);
System.out.println(thread1.getState());
}
}
输出为WAITING
TIMED_WAITING
这个状态和WAITING状态的区别就是,这个状态的等待是有一定时效的,即可以理解为WAITING状态等待的时间是永久的,即必须等到某个条件符合才能继续往下走,否则线程不会被唤醒。但是TIMED_WAITING,等待一段时间之后,会唤醒线程去重新获取锁。当执行如下代码的时候,对应的线程会进入到TIMED_WAITING状态
- Thread.sleep(long)
- Object.wait(long)
- Thread.join(long)
- LockSupport.parkNanos()
- LockSupport.parkUntil()
代码示例
Thread.sleep
public class MyThread3 extends Thread {
@Override
public void run() {
try {
Thread.sleep(10000);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
Thread thread = new MyThread3();
thread.start();
Thread.sleep(100);
System.out.println(thread.getState());
输出为TIMED_WAITING
Object.wait
public class MyThread4 extends Thread {
private Object lock;
public MyThread4(Object lock) {
this.lock = lock;
}
@Override
public void run() {
synchronized (lock){
try {
lock.wait(1000);//注意,此处1s之后线程醒来,会重新尝试去获取锁,如果拿不到,后面的代码也不执行
} catch (InterruptedException e) {
e.printStackTrace();
}
System.out.println("lock end");
}
}
}
byte[] lock = new byte[0];
MyThread4 thread = new MyThread4(lock);
thread.start();
Thread.sleep(100);
System.out.println(thread.getState());
Thread.sleep(2000);
System.out.println(thread.getState());
输出
TIMED_WAITING
lock end
TERMINATED
其余方法类似
TERMINATED
这个状态很好理解,即为线程执行结束之后的状态
状态之间的切换
以上这张图,能够较好的说明线程之前的状态切换