wait()、notify()和notifyAll()是Object类中的native final方法(除重载),这两个方法是为了解决多线程环境下的竞态环境,必须在同步方法块内调用
notify,notifyAll 唤起其他等待锁的线程,但并不会立刻切换到其他线程,而是等到执行完同步代码块之后再进行切换,使用notify会唤起最先等待的线程,notifyAll会唤起所有等待的线程
wait 使当前线程阻塞,释放所持有的锁(不必等到同步代码块结束)
不建议直接使用notify wait进行多线程编程
模拟使用notify的场景:两个线程竞争同一资源
//two threads competing for one resource
public class CompetingThread {
Thread[] threads;
CompetingThread(Thread[] threads){
this.threads= threads;
}
public void get() {
Thread t1 = threads[0], t2 = threads[1];
System.out.println("Ready to Get");
synchronized (this){
System.out.println(Thread.currentThread().getName()+" Got the lock");
System.out.println("Now, "+t1.getName()+" is "+t1.getState()+", "+t2.getName()+" is "+t2.getState());
System.out.println("notify() to release the lock after all done");
System.out.println("Do sth. before notify is just the same as after");
notify();
System.out.println("Do sth. after notify is just the same as before");
System.out.println(Thread.currentThread().getName()+" is done.");
}
}
public static void main(String[] args) {
Thread[] threads = new Thread[2];
CompetingThread competingThread = new CompetingThread(threads);
threads[0] = new Thread(()->competingThread.get());
threads[1] = new Thread(()->competingThread.get());
threads[0].start();
threads[1].start();
}
}
输出
//output
Ready to Get
Ready to Get
Thread-0 Got the lock
Now, Thread-0 is RUNNABLE, Thread-1 is BLOCKED
notify() to release the lock after all done
Do sth. before notify is just the same as after
Do sth. after notify is just the same as before
Thread-0 is done.
Thread-1 Got the lock
Now, Thread-0 is TERMINATED, Thread-1 is RUNNABLE
notify() to release the lock after all done
Do sth. before notify is just the same as after
Do sth. after notify is just the same as before
Thread-1 is done.
模拟使用wait和notify(notifyAll)的场景:顾客在餐厅点餐,服务员等候(BLOCKING),点餐完毕,顾客等候上菜(wait()),服务员上菜,告知顾客菜来了(notify())
public class WaitNotifyExample {
public static Object menu = new Object();
public static void main(String[] args) {
Thread customer = new Thread() {
@Override
public void run() {
synchronized (menu) {
System.out.println("Customer: I'm reading menu, please wait");
try {
Thread.sleep(1000);
System.out.println("Customer: These are my orders.(waiting for the meal)");
menu.wait();
}catch (InterruptedException e){}
System.out.println("Customer: Oh, these are my foods!");
}
}
};
Thread waiter = new Thread(){
@Override
public void run(){
System.out.println("Waiter: (Waiting for customer's orders)");
synchronized (menu){
System.out.println("Waiter: Gonna prepare you meal, sir.");
try {
Thread.sleep(1000);
menu.notify();//notifyAll if there are more than 1 customer
System.out.println("Waiter: Your foods are ready.");
}catch (InterruptedException e){}
}
}
};
customer.start();
waiter.start();
}
}
输出
//output
Customer: I'm reading menu, please wait
Waiter: (Waiting for customer's orders)
Customer: These are my orders.(waiting for the meal)
Waiter: Gonna prepare you meal, sir.
Waiter: Your foods are ready.
Customer: Oh, these are my foods!
一个线程安全的倒计时器,在倒计时为0前,令线程等待(getState():WAITING)
使用步骤:
CountDownLatch latch = new CountDownLatch(10)
latch.await()
,等待相关线程进行倒计时latch.countDown()
方法,进行倒计时ReentrantLock(Re-entrantLock)是一个基于AQS(AbstractQueuedSynchronizer)高性能工具,支持线程在未释放锁的情况下重复获取锁
方法 | 解释 |
---|---|
lock | holdCount+1,并给当前线程加锁 |
unlock | holdCount-1,holdCount为0时,释放资源 |
tryLock | 如果资源锁不在其他线程中,返回true,holdCount+1,否则返回false,不过线程并不会阻塞 |
lockInterruptibly() | 锁空闲的情况下正常获取,但是允许被其他线程的请求中断 |
getHoldCount | 获取holdCount,注意如果持有锁的不是本线程,则直接返回0 |
ReentrantLock的可重入功能基于AQS的同步状态:state(可通过getHoldCount获取)。当某一线程获取锁后,holdCount+1,并记录下当前持有锁的线程,再有线程来获取锁时,判断这个线程与持有锁的线程是否是同一个线程,如果是,holdCount+1,如果不是,阻塞线程。 当线程释放锁时,holdCount-1,holdCount为0时,唤醒其他线程,使其重新竞争锁。
解决synchronized在竞争激烈场景下的性能问题
在锁竞争不激烈的时候,多数情况下锁会停留在偏向锁和轻量级锁阶段,这两个阶段下,synchronized性能很好,但当存在大量线程竞争锁时,可能会膨胀为重量级锁,性能下降,ReentrantLock的性能会优于synchronized。不过在JDK1.6之后,synchronized进行了优化,大多数场景下性能与ReentrantLock所差无几
填补其他synchronized的缺陷
一旦线程卡在等待锁的阶段,就有可能出现死锁
不可响应中断
不能尝试获取锁
对多线程操作支持情况和实现方式(如果支持的话)
项目\锁 | synchronized | ReentrantLock |
---|---|---|
公平性 | 只支持非公平锁 | 支持公平锁和非公平锁 |
是否支持尝试获取锁 | 不支持 | 支持 tryLock(time, TimeUnit) |
是否可响应中断 | 不支持 | 支持 lockInterruptibly |
等待条件 | 支持 notify wait | 支持 Condition |
tryLock,lock,lockInterruptibly的对比测试,runnableWork是一个耗时操作,目的是让线程状态保持在RUNNABLE
public class ReentrantLockAPI {
public static void runnableWork(){
for (int i = 0; i < 1000; i++) {
for (int j = 0; j < 1000; j++) {
for (int k = 0; k < 1000; k++) {
}
}
}
}
public static void lockInterruptibly_tryLock_API(){
final ReentrantLock lock = new ReentrantLock();
Thread interruptibleThread = new Thread(()->{
try {
runnableWork();
lock.lockInterruptibly();
System.out.println("ReentrantLock is locked by "+Thread.currentThread().getName());
}catch (InterruptedException e){
System.out.println(Thread.currentThread().getName()+" was interrupted while getting lock");
}finally {
if(lock.isHeldByCurrentThread()) {
System.out.println("ReentrantLock is released by "+Thread.currentThread().getName());
lock.unlock();
}
}
});
Thread uninterruptibleThread = new Thread(()->{
System.out.println("ReentrantLock is locked by "+Thread.currentThread().getName());
try {
runnableWork();
lock.lock();
}catch (Exception e){
System.out.println(Thread.currentThread().getName()+" was interrupted.");
}finally {
System.out.println("ReentrantLock is released by "+Thread.currentThread().getName());
lock.unlock();
}
});
Thread interruptThread = new Thread(()->{
//uninterruptibleThread.interrupt();
interruptibleThread.interrupt();
});
interruptibleThread.start();
// uninterruptibleThread.start();
interruptThread.start();
}
public static void lock_tryLock_API(){
ReentrantLock lock = new ReentrantLock();
Thread t1 = new Thread(()->{
System.out.println("ReentrantLock is locked by "+Thread.currentThread().getName());
lock.lock();
try {
Thread.sleep(1000);
}catch (Exception e){}
System.out.println("ReentrantLock is released by "+Thread.currentThread().getName());
lock.unlock();
});
Thread t2 = new Thread(()->{
try {
System.out.println("Try to get the lock, if not, wait for sometime");
boolean res = lock.tryLock(500l, TimeUnit.MILLISECONDS);//if timeout >1000, result will be true, because t1 released the lock after 1000ms
System.out.println("tryLock returns: "+res+" ,and "+Thread.currentThread().getName()+" is "+Thread.currentThread().getState());
} catch (InterruptedException e) { }
});
t1.start();
t2.start();
}
public static void main(String[] args) {
lock_tryLock_API();
lockInterruptibly_tryLock_API();
}
}
lockInterruptibly_tryLock_API
尝试打断interruptibleThread,输出如下
//output when interrupt interruptibleThread thread
Thread-0 was interrupted while getting lock
尝试打断uninterruptibleThread,输出如下
//output when interrupt uninterruptibleThread thread
ReentrantLock is locked by Thread-1
ReentrantLock is released by Thread-1
lock_tryLock_API
调整tryLock的时间,可以观察到给定足够尝试时间的情况下才能获取到锁
//output when tryLock time is NOT enough
Try to get the lock, if not, wait for sometime
ReentrantLock is locked by Thread-0
tryLock returns: false ,and Thread-1 is RUNNABLE
ReentrantLock is released by Thread-0
//output when tryLock time is enough
ReentrantLock is locked by Thread-0
Try to get the lock, if not, wait for sometime
ReentrantLock is released by Thread-0
tryLock returns: true ,and Thread-1 is RUNNABLE
使用tryLock解决死锁问题
一个典型的死锁案例,两个对象锁分别被两个线程持有,两个线程经过一段时间后各请求对方所持有的锁,自然就产生了死锁
public class DeadLock {
public static final Object lock1 = new Object();
public static final Object lock2 = new Object();
public static void main(String[] a) {
Thread t1 = new Thread1();
Thread t2 = new Thread2();
t1.start();
t2.start();
}
private static class Thread1 extends Thread {
public void run() {
synchronized (lock1) {
try {
Thread.sleep(10);
} catch (InterruptedException ignored) {}
synchronized (lock2) {
}
}
}
}
private static class Thread2 extends Thread {
public void run() {
synchronized (lock2) {
try {
Thread.sleep(10);
} catch (InterruptedException ignored) {}
synchronized (lock1) {
}
}
}
}
}
使用tryLock解决,我们的目的是使两个线程都能获取到两个锁,并且分别处理,那么我们可以把获取锁写在一个逻辑里
public class DeadLock {
public static final ReentrantLock reentrantLock1 = new ReentrantLock();
public static final ReentrantLock reentrantLock2 = new ReentrantLock();
private static boolean getLocks(long milliseconds){
boolean getLock1 = false, getLock2 = false;
try{
getLock1 = reentrantLock1.tryLock(milliseconds, TimeUnit.MILLISECONDS);
getLock2 = reentrantLock2.tryLock(milliseconds,TimeUnit.MILLISECONDS);
}catch (Exception e){
e.printStackTrace();
} finally {
if(getLock1 && getLock2) return true;
else if(getLock1) reentrantLock1.unlock();
else if(getLock2) reentrantLock2.unlock();
return false;
}
}
public static void main(String[] a) {
Thread t1 = new Thread1();
Thread t2 = new Thread2();
t1.start();
t2.start();
}
private static class Thread1 extends Thread {
public void run() {
boolean res = getLocks(1000);
if(res){
System.out.println(Thread.currentThread().getName()+" Got lock1?"+reentrantLock1.isHeldByCurrentThread());
System.out.println(Thread.currentThread().getName()+" Got lock2?"+reentrantLock2.isHeldByCurrentThread());
try{
//..do sth. with locked resources
Thread.sleep(100);
}catch (Exception e){
e.printStackTrace();
} finally {
reentrantLock1.unlock();
reentrantLock2.unlock();
}
}
}
}
private static class Thread2 extends Thread {
public void run() {
boolean res = getLocks(1000);
if(res){
System.out.println(Thread.currentThread().getName()+" Got lock1?"+reentrantLock1.isHeldByCurrentThread());
System.out.println(Thread.currentThread().getName()+" Got lock2?"+reentrantLock2.isHeldByCurrentThread());
try{
//..do other things. with locked resources
Thread.sleep(100);
}catch (Exception e){
e.printStackTrace();
} finally {
reentrantLock1.unlock();
reentrantLock2.unlock();
}
}
}
}
}
我们模拟两个线程,可以看到输出中,两个线程分别获得了两个锁,从而解决了死锁的问题
//output
Thread-1 Got lock1?true
Thread-1 Got lock2?true
Thread-0 Got lock1?true
Thread-0 Got lock2?true
Condition对应一个ReentrantLock,在调用时,要求线程持有该ReentrantLock的锁,我们看下Condition类的实例方法:
方法名 | 参数 | 返回 |
---|---|---|
await | void | |
awaitUninterruptibly | void | |
awaitNanos | long | void |
await | long,TimeUnit | boolean |
awaitUntil | Date | boolean |
signal | void | |
signalAll | void |
可以看到,实际上主要实现了await和signal这两个功能,await和Object.await类似,给定等待时间,直到被通知,signal和notify类似,signal()唤醒等待在该Condition的线程,signalAll()唤醒所有等待在该Condition的线程
两个方法分别对应两个示例
第一个示例中,一个线程通过condition.signalAll唤醒另一个线程,相应的输出和使用方法与synchronized下的Object.notify,Object.await完全一致
第二个示例中,前3个线程对应一个锁的ConditionA,后3个线程对应同一个锁的ConditionB,最后一个线程唤醒前3个线程,后3个线程在等待时间截止后自动执行,从而实现了线程的分组控制
public class ReentrantLockConditionAPI {
public static void singleConditionForThreads(){
ReentrantLock lock = new ReentrantLock();
Condition condition = lock.newCondition();
Thread t1 = new Thread(){
@Override
public void run() {
try {
lock.lock();
System.out.println(Thread.currentThread().getName()+" work for 1 second");
Thread.sleep(1000);
condition.await();//Must between lock() and unlock()
System.out.println("Finished, start to wait");
System.out.println(Thread.currentThread().getName()+" another work for 1 second");
Thread.sleep(1000);
}catch (Exception e){
e.printStackTrace();
}finally {
System.out.println(Thread.currentThread().getName() + " finished");
lock.unlock();
}
}
};
Thread t2 = new Thread(){
@Override
public void run() {
try {
Thread.sleep(100);
lock.lock();
System.out.println(Thread.currentThread().getName()+" got the lock");
condition.signalAll();
}catch (Exception e){
e.printStackTrace();
}finally {
System.out.println(Thread.currentThread().getName()+" finished");
lock.unlock();
}
}
};
t1.start();
t2.start();
}
public static void multipleConditionForThreadGroups(){
ReentrantLock lock = new ReentrantLock();
Condition conditionA = lock.newCondition();
Condition conditionB = lock.newCondition();
ExecutorService executorService = Executors.newFixedThreadPool(7);
//conditionA
for (int i = 0; i < 3; i++) {
executorService.submit(new Thread(){
@Override
public void run() {
try {
lock.lock();
conditionA.await(2000,TimeUnit.MILLISECONDS);
System.out.println(Thread.currentThread().getName()+" got the lock again");
} catch (InterruptedException e) {
e.printStackTrace();
}finally {
lock.unlock();
}
}
});
}
//conditionB
for (int i = 0; i < 3; i++) {
executorService.submit(new Thread(){
@Override
public void run() {
lock.lock();
try {
conditionB.await(2000, TimeUnit.MILLISECONDS);
System.out.println(Thread.currentThread().getName()+" got the lock again");
} catch (InterruptedException e) {
e.printStackTrace();
}finally {
lock.unlock();
}
}
});
}
executorService.submit(new Thread(){
@Override
public void run() {
try {
Thread.sleep(1000);
lock.lock();
conditionA.signalAll();
System.out.println(Thread.currentThread().getName()+" signalAll conditionA threads");
} catch (InterruptedException e) {
e.printStackTrace();
}finally {
lock.unlock();
}
}
});
executorService.shutdown();
}
public static void main(String[] args) {
singleConditionForThreads();
multipleConditionForThreadGroups();
}
}
第二个示例的输出(注意时间间隔)
//output
pool-1-thread-7 signalAll conditionA threads
pool-1-thread-1 got the lock again
pool-1-thread-2 got the lock again
pool-1-thread-3 got the lock again
pool-1-thread-4 got the lock again
pool-1-thread-5 got the lock again
pool-1-thread-6 got the lock again
Semaphore提供了信号量机制,作为mutex(互斥量)与ReentrantLock类似,不同的是,ReentrantLock中的Condition要求线程必须持有锁,并且一个锁只能被一个线程持有,而Semaphore允许多线程访问同一资源。当Semaphore不允许访问资源时,线程会被阻塞直到可以获取permit
方法 | 解释 |
---|---|
Semaphore(int permits, boolean fair) | 构造方法,两个参数分别表示permit数量,是否公平锁,默认非公平锁 |
acquire() | 获取permit,否则阻塞直到可以获取 |
release() | 释放permit,注意这里可以超过构造方法中允许的permit数量,也就是可以不用acquire直接release |
availablePermits() | 查看目前可用的permit |
drainPermits() | 把permit置0,返回所有permit的数量 |
Semaphore不关心获取锁和释放锁的对象,并且允许多线程同时访问同一资源,acquire和release只是发放和收回permit(许可),是一种相对高级的同步机制
因为是高级API,所以没啥特别的,示例中一个容量为5的信号量跑20个线程,可以防止死锁,互斥执行
public class SemaphoreAPI {
public static void main(String[] args) {
Semaphore semaphore = new Semaphore(5);
ExecutorService executorService = Executors.newFixedThreadPool(20);
for (int i = 0; i < 20; i++) {
executorService.submit(()->{
try {
semaphore.acquire();
System.out.println(Thread.currentThread().getName()+" is doing some work");
Thread.sleep(1000);
} catch (InterruptedException e) {
e.printStackTrace();
}finally {
semaphore.release();
}
});
}
executorService.shutdown();
}
}
https://juejin.im/post/5ce91cedf265da1bca51b78a https://stackoverflow.com/questions/17683575/binary-semaphore-vs-a-reentrantlock https://www.geeksforgeeks.org/reentrant-lock-java/