LinkedBlockingQueue
看LinkedBlockingQueue之前可以参考前一篇:ArrayBlockingQueue源码解读
例子
将前面的例子改为LinkedBlockingQueue实现,我们看到,程序依然可以正常运行。
package juc;
import org.junit.Test;
import java.util.concurrent.BlockingQueue;
import java.util.concurrent.LinkedBlockingQueue;
public class LinkedBlockingQueueTest {
@Test
public void test() throws InterruptedException {
final BlockingQueue blockingQueue = new LinkedBlockingQueue(10);
// 新建两个生产者线程+一个消费者线程,(生产是比消费快的,当缓冲区满时,生产者会阻塞)
Thread t1 = new Thread(new Producer(blockingQueue), "生产者1");
Thread t2 = new Thread(new Producer(blockingQueue), "生产者2");
Thread t3 = new Thread(new Consumer(blockingQueue), "消费者1");
t1.start();
t2.start();
t3.start();
t1.join();
t2.join();
t3.join();
}
class Consumer implements Runnable {
private BlockingQueue blockingQueue;
Consumer(BlockingQueue blockingQueue) {
this.blockingQueue = blockingQueue;
}
@Override
public void run() {
for (; ; ) {
System.out.println(Thread.currentThread().getName() + "消费,当前容量:" + blockingQueue.size());
try {
Thread.sleep(1000);
blockingQueue.take();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
}
class Producer implements Runnable {
private BlockingQueue blockingQueue;
Producer(BlockingQueue blockingQueue) {
this.blockingQueue = blockingQueue;
}
@Override
public void run() {
for (; ; ) {
System.out.println(Thread.currentThread().getName() + "生产,当前容量:" + blockingQueue.size());
try {
Thread.sleep(1000);
blockingQueue.put("product");
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
}
}
执行结果
源码
那么LinkedBlockingQueue与ArrayBlockingQueue之间有什么区别呢?我们先来看看LinkedBlockingQueue吧。
/**
* Linked list node class
*/
static class Node<E> {
E item;
/**
* One of:
* - the real successor Node
* - this Node, meaning the successor is head.next
* - null, meaning there is no successor (this is the last node)
*/
Node<E> next;
Node(E x) { item = x; }
}
/** The capacity bound, or Integer.MAX_VALUE if none */
private final int capacity;
/** Current number of elements */
private final AtomicInteger count = new AtomicInteger();
/**
* Head of linked list.
* Invariant: head.item == null
*/
transient Node<E> head;
/**
* Tail of linked list.
* Invariant: last.next == null
*/
private transient Node<E> last;
相比ArrayBlockingQueue,LinkedBlockingQueue是采用了链表结构,如果不设置capacity,默认大小为Integer.MAX_VALUE。同时维护了队列头部与尾部。另外你发现,维护队列大小的count不再是int,而是AtomicInteger。
/** Lock held by take, poll, etc */
private final ReentrantLock takeLock = new ReentrantLock();
/** Wait queue for waiting takes */
private final Condition notEmpty = takeLock.newCondition();
/** Lock held by put, offer, etc */
private final ReentrantLock putLock = new ReentrantLock();
/** Wait queue for waiting puts */
private final Condition notFull = putLock.newCondition();
在并发控制方面,我们看到读和写分别维护了一把锁,实现了锁分离,感觉LinkedBlockingQueue的并发读更高了。
下面来看看是不是这样的:
先看take方法:
public E take() throws InterruptedException {
E x;
int c = -1;
final AtomicInteger count = this.count;
final ReentrantLock takeLock = this.takeLock;
// 获取takeLock锁
takeLock.lockInterruptibly();
try {
// 当队列为空,take线程阻塞,排在notEmpty阻塞队列中
while (count.get() == 0) {
notEmpty.await();
}
// 执行出队操作
x = dequeue();
// count数 -1,返回的是新增前的值
c = count.getAndDecrement();
// 如果个数>1,唤醒因take阻塞的线程去竞争takeLock锁
if (c > 1)
notEmpty.signal();
} finally {
// 释放锁
takeLock.unlock();
}
// 在释放takeLock之后判断
if (c == capacity)
// 如果队列已满,唤醒notFull条件队列中的线程执行put操作 (这里有点疑问了,为什么满了才要唤醒put操作呢?)
signalNotFull();
return x;
}
private E dequeue() {
// head 节点是不存元素的
Node<E> h = head;
// 出队的是head的下一个first节点
Node<E> first = h.next;
h.next = h; // help GC
// 让head后移一位,取出first元素,并置为null,出队
head = first;
E x = first.item;
first.item = null;
return x;
}
private void signalNotFull() {
final ReentrantLock putLock = this.putLock;
putLock.lock();
try {
notFull.signal();
} finally {
putLock.unlock();
}
}
我们先往下看,再看看put方法,看看两把锁是如何配合使用的:
public void put(E e) throws InterruptedException {
if (e == null) throw new NullPointerException();
int c = -1;
Node<E> node = new Node<E>(e);
final ReentrantLock putLock = this.putLock;
final AtomicInteger count = this.count;
// 获取putLock锁才能执行put操作
putLock.lockInterruptibly();
try {
// 当检测到count已满,put现在就阻塞在notFull条件队列中
while (count.get() == capacity) {
notFull.await();
}
// 否则就入队(这里是没有问题了,因为此时已经获取到了putLock,其他add,offer操作在此时是阻塞的,不会并发执行,所以count只会减小,不会增大,可以入队。另外是从尾部入队,也不会影响出队)
enqueue(node);
// 获取最新的count数并加1
c = count.getAndIncrement();
if (c + 1 < capacity)
// 如果至少还可以放一个元素,则唤醒一个put线程
notFull.signal();
} finally {
// 释放putLock锁
putLock.unlock();
}
if (c == 0)
// 如果元素个数为0,唤醒一个take线程(这里同样是有疑问的,为什么只有为0的时候才唤醒)
signalNotEmpty();
}
private void enqueue(Node<E> node) {
// 将node放入链表尾部。首先node置为last的next,放入链表。再将last指向node节点就可以了
last = last.next = node;
}
private void signalNotEmpty() {
final ReentrantLock takeLock = this.takeLock;
// 获取takeLock锁,进行唤醒take线程
takeLock.lock();
try {
notEmpty.signal();
} finally {
takeLock.unlock();
}
}
回答:上述当c==0才唤醒take线程,引用qq_26898645博主的理解,我觉得有道理,贴在这里:
上述put方法中,比较疑惑的地方是,为什么最后要判断,当容量为0时,需要激活notEmpty-Condition阻塞的take线程?
按道理讲,只要进行了put操作,就证明肯定队列不为空了,直接进行signalNotEmpty()不可以吗?
想了想,大概原因是这样:
直接进行signalNotEmpty()可以是可以,不过性能不是最优的,因为如果之前的队列本身就不为空,则说明没有处于因notEmpty.wait()而阻塞的take线程,自然也就无需进行唤醒动作。
另外,判断之前是否有处于阻塞的take线程的方法也非常巧妙,即通过count.getAndIncrement();的返回值获得,
因getAndIncrement返回之前旧值,自然在实现count同步自增的同时,返回了之前值。
小结
1、LinkedBlockingQueue可以指定大小,也可以不指定,不指定大小默认为int最大值。ArrayBlockingQueue必须指定大小。
2、LinkedBlockingQueue的take与put分别持有一把锁,而ArrayBlockingQueue所有操作持有的是一把锁,感觉上LinkedBlockingQueue在并发性能上更好。我测试了一下:
package juc;
import org.junit.Test;
import java.util.concurrent.ArrayBlockingQueue;
import java.util.concurrent.BlockingQueue;
import java.util.concurrent.LinkedBlockingQueue;
public class ArrayBlockingQueueAndLinkedBlockingQueueTest {
@Test
public void test() throws InterruptedException {
final BlockingQueue arrayBlockingQueue = new ArrayBlockingQueue(10);
Thread t1 = new Thread(new Taker(arrayBlockingQueue), "arrayBlockingQueueTaker");
Thread t2 = new Thread(new Puter(arrayBlockingQueue), "arrayBlockingQueuePuter");
long start1 = System.currentTimeMillis();
t1.start();
t2.start();
t1.join();
t2.join();
System.out.println("arrayBlockingQueue消耗时间:" + (System.currentTimeMillis() - start1));
final BlockingQueue linkedBlockingQueue = new LinkedBlockingQueue(10);
Thread t3 = new Thread(new Taker(linkedBlockingQueue), "linkedBlockingQueueTaker");
Thread t4 = new Thread(new Puter(linkedBlockingQueue), "linkedBlockingQueuePuter");
long start2 = System.currentTimeMillis();
t3.start();
t4.start();
t3.join();
t4.join();
System.out.println("linkedBlockingQueue消耗时间:" + (System.currentTimeMillis() - start2));
}
class Taker implements Runnable {
private BlockingQueue blockingQueue;
Taker(BlockingQueue blockingQueue) {
this.blockingQueue = blockingQueue;
}
@Override
public void run() {
for (int i = 0; i < 10000000; i++) {
try {
blockingQueue.take();
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
}
class Puter implements Runnable {
private BlockingQueue blockingQueue;
Puter(BlockingQueue blockingQueue) {
this.blockingQueue = blockingQueue;
}
@Override
public void run() {
for (int i = 0; i < 10000000; i++) {
try {
blockingQueue.put("product");
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
}
}
执行结果:
执行效率差不多,可能的原因是:虽然LinkedBlockingQueue并发程度较高,但是入队需要新增节点,创建节点对象。而ArrayBlockingQueue可以重用队列存放数组。