CountDownLatch、Semaphore和CyclicBarrier区别

wy971744839

已于 2023-06-11 11:49:27 修改

阅读量320

点赞数

分类专栏： Java基础文章标签： java jvm 开发语言

于 2023-04-05 17:34:38 首次发布

本文链接：https://blog.csdn.net/wy971744839/article/details/129973926

版权

Java基础专栏收录该内容

17 篇文章 0 订阅

订阅专栏

Java并发包中有三个类用于同步一批线程的行为，分别是CountDownLatch、Semaphore和CyclicBarrier。

一、CountDownLatch

CountDownLatch是一个计数器闭锁，通过它可以完成类似于阻塞当前线程的功能，即：一个线程或多个线程一直等待，直到其他线程执行的操作完成。

CountDownLatch用一个给定的计数器来初始化，该计数器的操作是原子操作，即同时只能有一个线程去操作该计数器。调用该类await方法的线程会一直处于阻塞状态，直到其他线程调用countDown方法使当前计数器的值变为零，每次调用countDown计数器的值减1。当计数器值减至零时，所有因调用await()方法而处于等待状态的线程就会继续往下执行。

这种现象只会出现一次，因为计数器不能被重置，如果业务上需要一个可以重置计数次数的版本，可以考虑使用CycliBarrier。

在某些业务场景中，程序执行需要等待某个条件完成后才能继续执行后续的操作；典型的应用如并行计算，当某个处理的运算量很大时，可以将该运算任务拆分成多个子任务，等待所有的子任务都完成之后，父任务再拿到所有子任务的运算结果进行汇总。

例子1：

import lombok.extern.slf4j.Slf4j;

import java.util.concurrent.CountDownLatch;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

@Slf4j
public class CountDownLatchExample1 {

    private final static int threadCount = 200;

    public static void main(String[] args) throws Exception {

        ExecutorService exec = Executors.newCachedThreadPool();

        final CountDownLatch countDownLatch = new CountDownLatch(threadCount);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    test(threadNum);
                } catch (Exception e) {
                    log.error("exception", e);
                } finally {
                    countDownLatch.countDown();
                }
            });
        }
        countDownLatch.await();
        log.info("finish");
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        Thread.sleep(100);
        log.info("{}", threadNum);
        Thread.sleep(100);
    }
}

输出：

20:18:32.917 [pool-1-thread-7] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 6
20:18:32.917 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 5
20:18:32.919 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 4
20:18:32.918 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 0
20:18:32.918 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 2
20:18:32.916 [pool-1-thread-9] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 8
20:18:32.918 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 3
20:18:32.916 [pool-1-thread-10] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 9
20:18:32.916 [pool-1-thread-8] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 7
20:18:32.917 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - 1
20:18:33.032 [main] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample1 - finish

例子2：

import java.util.concurrent.CountDownLatch;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;

@Slf4j
public class CountDownLatchExample2 {

    private final static int threadCount = 200;

    public static void main(String[] args) throws Exception {

        ExecutorService exec = Executors.newCachedThreadPool();

        final CountDownLatch countDownLatch = new CountDownLatch(threadCount);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    test(threadNum);
                } catch (Exception e) {
                    log.error("exception", e);
                } finally {
                    countDownLatch.countDown();
                }
            });
        }
        countDownLatch.await(10, TimeUnit.MILLISECONDS);
        log.info("finish");
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        Thread.sleep(100);
        log.info("{}", threadNum);
    }
}

结果：超过指定时间跳过等待

20:19:34.878 [main] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - finish
20:19:34.964 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 2
20:19:34.965 [pool-1-thread-10] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 9
20:19:34.964 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 0
20:19:34.965 [pool-1-thread-8] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 7
20:19:34.964 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 1
20:19:34.965 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 4
20:19:34.965 [pool-1-thread-7] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 6
20:19:34.964 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 3
20:19:34.965 [pool-1-thread-9] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 8
20:19:34.965 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CountDownLatchExample2 - 5

Semaphore

Semaphore与CountDownLatch相似，不同的地方在于Semaphore的值被获取到后是可以释放的，并不像CountDownLatch那样一直减到底。它也被更多地用来限制流量，类似阀门的功能。

如果限定某些资源最多有N个线程可以访问，那么超过N个主不允许再有线程来访问，同时当现有线程结束后，就会释放，然后允许新的线程进来。有点类似于锁的lock与unlock过程。相对来说他也有两个主要的方法：

用于获取权限的acquire()，其底层实现与CountDownLatch.countdown()类似；
用于释放权限的release()，其底层实现与acquire()是一个互逆的过程。

具体可参看：工藤新木：Java常用锁的使用

CyclicBarrier

CyclicBarrier也是一个同步辅助类，它允许一组线程相互等待，直到到达某个公共屏障点（common barrier point）。通过它可以完成多个线程之间相互等待，只有当每个线程都准备就绪后，才能各自继续往下执行后面的操作。

类似于CountDownLatch，它也是通过计数器来实现的。当某个线程调用await方法时，该线程进入等待状态，且计数器加1，当计数器的值达到设置的初始值时，所有因调用await进入等待状态的线程被唤醒，继续执行后续操作。

例子1：

@Slf4j
public class CyclicBarrierExample1 {

    private static CyclicBarrier barrier = new CyclicBarrier(5);

    public static void main(String[] args) throws Exception {

        ExecutorService executor = Executors.newCachedThreadPool();

        for (int i = 0; i < 10; i++) {
            final int threadNum = i;
            Thread.sleep(1000);
            executor.execute(() -> {
                try {
                    race(threadNum);
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        executor.shutdown();
    }

    private static void race(int threadNum) throws Exception {
        Thread.sleep(1000);
        log.info("{} is ready", threadNum);
        barrier.await();
        log.info("{} continue", threadNum);
    }
}

结果： ready ready .. go

20:24:34.616 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 0 is ready
20:24:35.610 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 1 is ready
20:24:36.610 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 2 is ready
20:24:37.611 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 3 is ready
20:24:38.612 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 4 is ready
20:24:38.612 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 0 continue
20:24:38.612 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 1 continue
20:24:38.612 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 4 continue
20:24:38.612 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 3 continue
20:24:38.612 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 2 continue
20:24:39.614 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 5 is ready
20:24:40.613 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 6 is ready
20:24:41.614 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 7 is ready
20:24:42.615 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 8 is ready
20:24:43.615 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 9 is ready
20:24:43.615 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 9 continue
20:24:43.615 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 5 continue
20:24:43.615 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 6 continue
20:24:43.615 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 7 continue
20:24:43.615 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample1 - 8 continue

因为CyclicBarrier在释放等待线程后可以重用，所以称为循环barrier。CyclicBarrier支持一个可选的Runnable，在计数器的值到达设定值后（但在释放所有线程之前），该Runnable运行一次。注：Runnable在每个屏障点只运行一个。

import lombok.extern.slf4j.Slf4j;

import java.util.concurrent.CyclicBarrier;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;

@Slf4j
public class CyclicBarrierExample3 {

    private static CyclicBarrier barrier = new CyclicBarrier(5, () -> {
        log.info("callback is running");
    });

    public static void main(String[] args) throws Exception {

        ExecutorService executor = Executors.newCachedThreadPool();

        for (int i = 0; i < 10; i++) {
            final int threadNum = i;
            Thread.sleep(1000);
            executor.execute(() -> {
                try {
                    race(threadNum);
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        executor.shutdown();
    }

    private static void race(int threadNum) throws Exception {
        Thread.sleep(1000);
        log.info("{} is ready", threadNum);
        barrier.await();
        log.info("{} continue", threadNum);
    }
}

结果：

20:28:32.790 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 0 is ready
20:28:33.785 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 1 is ready
20:28:34.786 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 2 is ready
20:28:35.787 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 3 is ready
20:28:36.787 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 4 is ready
20:28:36.787 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - callback is running
20:28:36.787 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 4 continue
20:28:36.788 [pool-1-thread-1] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 0 continue
20:28:36.788 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 1 continue
20:28:36.788 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 2 continue
20:28:36.788 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 3 continue
20:28:37.788 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 5 is ready
20:28:38.789 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 6 is ready
20:28:39.789 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 7 is ready
20:28:40.790 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 8 is ready
20:28:41.791 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 9 is ready
20:28:41.791 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - callback is running
20:28:41.791 [pool-1-thread-3] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 9 continue
20:28:41.791 [pool-1-thread-6] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 5 continue
20:28:41.791 [pool-1-thread-4] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 6 continue
20:28:41.818 [pool-1-thread-2] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 8 continue
20:28:41.818 [pool-1-thread-5] INFO com.mmall.concurrency.example.aqs.CyclicBarrierExample3 - 7 continue

CountDownLatch与CyclicBarrier的区别

CountDownLatch主要是实现了1个或N个线程需要等待其他线程完成某项操作之后才能继续往下执行操作，描述的是1个线程或N个线程等待其他线程的关系。
CyclicBarrier主要是实现了多个线程之间相互等待，直到所有的线程都满足了条件之后各自才能继续执行后续的操作，描述的多个线程内部相互等待的关系。
CountDownLatch是一次性的，而CyclicBarrier则可以被重置而重复使用。