线程安全的本质体现在两个方面,
A变量安全:多线程同时运行一段代码
B线程同步:一个线程还没执行完,另一个线程又进来接着执行。
看个简单的例子。
Java代码
public class ThreadSafe implements java.lang.Runnable { int num = 1; public void run() { for (int i = 0; i < 3; i++) { num = num + 1; try { Thread.sleep(2000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println("num is value +==="+Thread.currentThread().getName()+"---------" + num); } } }
TestMan.java 写道
package com.java.thread.test; public class TestMan { public static void main(String[] args) { Runnable safe=new ThreadSafe(); Thread thread1=new Thread(safe,"thread1"); Thread thread2=new Thread(safe,"thread2"); thread1.start(); thread2.start(); } }
运行结果
num is value +===thread2---------3 num is value +===thread1---------4 num is value +===thread2---------5 num is value +===thread1---------6 num is value +===thread1---------7 num is value +===thread2---------7
很明显是错误的,应为两个线程共享同一个变量。这里就是变量的安全问题。
解决办法:
1抛弃单实例,多线程的方式,用多实例,多线程的方式,这样就和单线程是一个样了,不会出错,但是是最接近传统的编程模式
2不要用类的实例变量,经可能把变量封装到方法内部。
1类的解决办法的代码。
Java代码
public class TestMan { public static void main(String[] args) { Runnable safe=new ThreadSafe(); Runnable safe2=new ThreadSafe(); Thread thread1=new Thread(safe,"thread1"); Thread thread2=new Thread(safe2,"thread2"); thread1.start(); thread2.start(); } }
运行结果
num is value +===thread1---------2 num is value +===thread2---------2 num is value +===thread1---------3 num is value +===thread2---------3 num is value +===thread1---------4 num is value +===thread2---------4
2类解决办法的代码
Java代码
public class ThreadSafe implements java.lang.Runnable { public void run() { int num = 1; for (int i = 0; i < 3; i++) { num = num + 1; try { Thread.sleep(2000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println("num is value +==="+Thread.currentThread().getName()+"---------" + num); } } }
Java代码
public class TestMan { public static void main(String[] args) { Runnable safe=new ThreadSafe(); Thread thread1=new Thread(safe,"thread1"); Thread thread2=new Thread(safe,"thread2"); thread1.start(); thread2.start(); } }
运行结果
num is value +===thread2---------2 num is value +===thread1---------2 num is value +===thread1---------3 num is value +===thread2---------3 num is value +===thread1---------4 num is value +===thread2---------4
这两种办法,比较推荐适用第二个办法,就是把变量经可能的封装到风发内部,这样他们就是线程的私有变量了。另外,从jdk1.2后,推出了 threadlocal 对象,它作为线程的一个局部变量,可以为每个线程创建一个副本,用来保存每个线程的属性,各是各的,互不干扰。单每个 threadlocal变量只能保存一个变量,假如有多个变量要保存,那么就要写多个threadlocal对象。
我们把代码改写一下。
Java代码
public class ThreadSafe implements java.lang.Runnable { ThreadLocal local=new ThreadLocal(); public void run() { for (int i = 0; i < 3; i++) { if(local.get()==null){ local.set(new Integer(1)); } int num=local.get().intValue(); num=num+1; local.set(new Integer(num)); try { Thread.sleep(2000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println("num is value +==="+Thread.currentThread().getName()+"---------" + local.get().intValue()); } } }
Java代码
public class TestMan { public static void main(String[] args) { Runnable safe=new ThreadSafe(); Thread thread1=new Thread(safe,"thread1"); Thread thread2=new Thread(safe,"thread2"); thread1.start(); thread2.start(); } }
运行结果
num is value +===thread2---------2 num is value +===thread1---------2 num is value +===thread1---------3 num is value +===thread2---------3 num is value +===thread1---------4 num is value +===thread2---------4
结果是一样的,所以这里变量安全有3个办法可以解决。
然后在说说线程的同步的问题。我们看上面的运行结果。
num is value +===thread2---------2 num is value +===thread1---------2 num is value +===thread1---------3 num is value +===thread2---------3 num is value +===thread1---------4 num is value +===thread2---------4
就 可以看出他们不是线程同步的,是thread1和thread2在交替执行的。在有些情况下,要求一段代码在运行的过程中是一个不可分割的实体,就是原子的。就是说当已经有线程在执行这段代码的时候,其他的线程必须等待他执行完毕后才能竟来执行,这就是所谓的线程同步。
java通过同步锁来执行线程的同步和等待,也就是说,要不间断执行的代码需要放在synchronized关键字标识的代码块中。可以用来修饰代码块,也可以修饰方法。
Java代码
public class ThreadSafe implements java.lang.Runnable{ int num = 1; public void run() { for (int i = 0; i < 3; i++) { synchronized (this) { num = num + 1; System.out.println("num is value +==="+Thread.currentThread().getName()+"---------" + num); } try { Thread.sleep(1000); } catch (InterruptedException e) { e.printStackTrace(); } } } }
Java代码
public class TestMan { public static void main(String[] args) { Runnable safe=new ThreadSafe(); Thread thread1=new Thread(safe,"thread1"); Thread thread2=new Thread(safe,"thread2"); thread1.start(); thread2.start(); } }
运行结果
num is value +===thread1---------2 num is value +===thread2---------3 num is value +===thread2---------4 num is value +===thread1---------5 num is value +===thread2---------6 num is value +===thread1---------7
Java代码
public class ThreadSafe implements java.lang.Runnable { public void run() { int num = 1; synchronized (this) { for (int i = 0; i < 3; i++) { num = num + 1; try { Thread.sleep(2000); } catch (InterruptedException e) { e.printStackTrace(); } System.out.println("num is value +===" + Thread.currentThread().getName() + "---------" + num); } } } }
ThreadLocal使用场合主要解决多线程中数据数据因并发产生不一致问题。ThreadLocal为每个线程的中并发访问的数据提供一个副本,通过访问副本来运行业务,这样的结果是耗费了内存,单大大减少了线程同步所带来性能消耗,也减少了线程并发控制的复杂度。
ThreadLocal不能使用原子类型,只能使用Object类型。ThreadLocal的使用比synchronized要简单得多。
ThreadLocal和Synchonized都用于解决多线程并发访问。但是ThreadLocal与synchronized有本质的区别。synchronized是利用锁的机制,使变量或代码块在某一时该只能被一个线程访问。而ThreadLocal为每一个线程都提供了变量的副本,使得每个线程在某一时间访问到的并不是同一个对象,这样就隔离了多个线程对数据的数据共享。而Synchronized却正好相反,它用于在多个线程间通信时能够获得数据共享。
Synchronized用于线程间的数据共享,而ThreadLocal则用于线程间的数据隔离。
当然ThreadLocal并不能替代synchronized,它们处理不同的问题域。Synchronized用于实现同步机制,比ThreadLocal更加复杂