曾經(jīng)有面試官問過這樣一個問題，如何讓兩個線程交替打印奇數(shù)和偶數(shù)？如何讓多個線程交替執(zhí)行？
回答這個問題之前需要先了解java多線程的運(yùn)行機(jī)制，線程間通信機(jī)制，線程同步問題。
這個問題我知道的實現(xiàn)方案有兩種，一種是基于synchronized和wait/notify，另外一種是基于Lock和Condition.
1.基于synchronized和wait/notify
package org.stathry.jdkdeep.concurrent;
/**
* 線程間通信-基于Object的wait/notify
* @date 2017年12月21日
*/
public class ThreadTurnTest1 {
private Object lock = new Object();
private boolean RUN0 = true;
private static final int LIMIT = 1000;
public static void main(String[] args) throws InterruptedException {
final ThreadTurnTest1 o = new ThreadTurnTest1();
new Thread(new Runnable() {
@Override
public void run() {
try {
o.m0();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}, "t0").start();
new Thread(new Runnable() {
@Override
public void run() {
try {
o.m1();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}, "t1").start();
Thread.sleep(10 * 1000);
}
private void m1() throws InterruptedException {
for (int i = 1; i < LIMIT; i += 2) {
synchronized (lock) {
if (RUN0) {
lock.wait();
}
System.out.println(Thread.currentThread().getName() + "___" + i);
RUN0 = true;
lock.notify();
}
}
}
private void m0() throws InterruptedException {
for (int i = 0; i < LIMIT; i += 2) {
synchronized (lock) {
if (!RUN0) {
lock.wait();
}
System.out.println(Thread.currentThread().getName() + "___" + i);
RUN0 = false;
lock.notify();
}
}
}
}
執(zhí)行結(jié)果如下：
t0___0
t1___1
t0___2
t1___3
t0___4
t1___5
t0___6
t1___7
t0___8
t1___9
t0___10
...
2.基于Lock和Condition
基于Lock和Condition可以實現(xiàn)兩個以上線程的通信。
package org.stathry.jdkdeep.concurrent;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.LinkedBlockingQueue;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.locks.Condition;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReentrantLock;
/**
* 線程間通信-基于lock的condition
* @date 2017年12月21日
*/
public class ThreadTurnTest2 {
private Lock lock = new ReentrantLock();
private Condition c0 = lock.newCondition();
private Condition c1 = lock.newCondition();
private Condition c2 = lock.newCondition();
private int status = 0;
private static final int LIMIT = 10000;
public static void main(String[] args) throws InterruptedException {
ExecutorService exec = new ThreadPoolExecutor(3, 30, 60, TimeUnit.SECONDS, new LinkedBlockingQueue(300),
Executors.defaultThreadFactory(), new ThreadPoolExecutor.AbortPolicy());
final ThreadTurnTest2 o = new ThreadTurnTest2();
exec.submit(o.createTask(0));
exec.submit(o.createTask(1));
exec.submit(o.createTask(2));
exec.shutdown();
exec.awaitTermination(5, TimeUnit.MINUTES);
}
/**
* @param name
* @param i
*/
private Runnable createTask(final int i) {
return new Runnable() {
@Override
public void run() {
try {
m(i);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
};
}
private void m(final int n) throws InterruptedException {
for (int i = n; i < LIMIT; i += 3) {
lock.lock();
try {
switch (n) {
case 0:
if (status != 0) {
c0.await();
}
System.out.println(Thread.currentThread().getName() + "___" + i);
status = 1;
c1.signal();
break;
case 1:
if (status != 1) {
c1.await();
}
System.out.println(Thread.currentThread().getName() + "___" + i);
status = 2;
c2.signal();
break;
case 2:
if (status != 2) {
c2.await();
}
System.out.println(Thread.currentThread().getName() + "___" + i);
status = 0;
c0.signal();
break;
default:
break;
}
} finally {
lock.unlock();
}
}
}
}
執(zhí)行結(jié)果如下：
pool-1-thread-1___0
pool-1-thread-2___1
pool-1-thread-3___2
pool-1-thread-1___3
pool-1-thread-2___4
pool-1-thread-3___5
pool-1-thread-1___6
pool-1-thread-2___7
pool-1-thread-3___8
pool-1-thread-1___9
pool-1-thread-2___10
pool-1-thread-3___11
pool-1-thread-1___12
pool-1-thread-2___13
pool-1-thread-3___14
...
看完上面兩種方式你學(xué)會了嗎？
其實這兩種方式思路都是類似的，大致可以分三步：
循環(huán)內(nèi)加鎖
判斷共享狀態(tài)變量，如果狀態(tài)值表示還沒輪到當(dāng)前線程執(zhí)行，則調(diào)用調(diào)用鎖對象的wait方法
等待狀態(tài)變化的線程被喚醒，執(zhí)行任務(wù)，然后改變狀態(tài)，然后調(diào)用鎖對象的notify方法
再多問幾個問題，為什么要加鎖呢？因為只有加鎖才能保證多線程交替有序執(zhí)行，否則線程的執(zhí)行是由操作系統(tǒng)隨機(jī)調(diào)度的，那么執(zhí)行順序自然是亂序的。為什么要對同一個對象加鎖呢？因為只有對同一對象加鎖，才能保證訪問該鎖的兩個線程之間相互通信(不同的對象調(diào)用notify不會喚醒處于wait狀態(tài)的線程)，否則即使線程已經(jīng)獲取到鎖了，但可能會因為沒輪到自己執(zhí)行而一直處于wait狀態(tài)。
如果覺得寫的不錯，記得點擊關(guān)注，如果寫的不好歡迎批評指正，讓我們一起進(jìn)步！
