volatile实现可见性但不保证原子性
volatile关键字:
- 能够保证volatile变量的可见性
- 不能保证volatile变量复合操作的原子性
volatile如何实现内存可见性:
深入来说:通过加入内存屏障和禁止重排序优化来实现的。
- 对volatile变量执行写操作时,会在写操作后加入一条store屏障指令
- 对volatile变量执行读操作时,会在读操作前加入一条load屏障指令
通俗地讲:volatile变量在每次被线程访问时,都强迫从主内存中重读该变量的值,而当该变量发生变化时,又会强迫线程将最新的值刷新到主内存。这样任何时刻,不同的线程总能看到该变量的最新值。
线程写volatile变量的过程:
- 改变线程工作内存中volatile变量副本的值
- 将改变后的副本的值从工作内存刷新到主内存
线程读volatile变量的过程:
- 从主内存中读取volatile变量的最新值到线程的工作内存中
- 从工作内存中读取volatile变量的副本
volatile不能保证volatile变量复合操作的原子性:
private int number = 0; number++; //不是原子操作
它分为三步:
读取number的值
将number的值加1
写入最新的number的值
保证number自增操作的原子性:
- 使用synchronized关键字
- 使用ReentrantLock
- 使用AtomicInteger
使用synchronized关键字
import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; /** * @author InJavaWeTrust */ public class TestSyn implements Runnable { private int number = 0; public int getNumber() { return this.number; } public void run() { increase(); } public void increase() { synchronized (this) { this.number++; } } public static void main(String[] args) { ExecutorService exec = Executors.newFixedThreadPool(1000); TestSyn syn = new TestSyn(); for (int i = 0; i < 1000; i++) { exec.submit(syn); } System.out.println("number : " + syn.getNumber()); exec.shutdown(); } }
使用ReentrantLock
import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.locks.Lock; import java.util.concurrent.locks.ReentrantLock; /** * @author InJavaWeTrust */ public class TestRee implements Runnable { private Lock lock = new ReentrantLock(); private int number = 0; public int getNumber() { return this.number; } public void run() { increase(); } public void increase() { lock.lock(); try { this.number++; } finally { lock.unlock(); } } public static void main(String[] args) { TestRee ree = new TestRee(); ExecutorService exec = Executors.newFixedThreadPool(1000); for (int i = 0; i < 1000; i++) { exec.submit(ree); } System.out.println("number : " + ree.getNumber()); exec.shutdown(); } }
使用AtomicInteger
import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors; import java.util.concurrent.atomic.AtomicInteger; /** * @author InJavaWeTrust */ public class TestAtomic implements Runnable { private static AtomicInteger number = new AtomicInteger(0); public void run() { increase(); } public void increase() { number.getAndAdd(1); } public static void main(String[] args) { TestAtomic ato = new TestAtomic(); ExecutorService exec = Executors.newFixedThreadPool(1000); for (int i = 0; i < 1000; i++) { exec.submit(ato); } System.out.println("number : " + number.get()); exec.shutdown(); } }