Java并发之CountDownLatch、Semaphore和CyclicBarrier-低调大师

Java并发之CountDownLatch、Semaphore和CyclicBarrier

2018-08-14 347

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。

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。

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);
    }
}

结果：

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

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。

import lombok.extern.slf4j.Slf4j;

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

@Slf4j
public class SemaphoreExample1 {

    private final static int threadCount = 20;

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

        ExecutorService exec = Executors.newCachedThreadPool();
       // 每次最多三个线程获取许可
        final Semaphore semaphore = new Semaphore(3);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    semaphore.acquire(); // 获取一个许可
                    test(threadNum);
                    semaphore.release(); // 释放一个许可
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        log.info("{}", threadNum);
        Thread.sleep(1000);
    }
}

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

@Slf4j
public class SemaphoreExample2 {

    private final static int threadCount = 20;

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

        ExecutorService exec = Executors.newCachedThreadPool();

        final Semaphore semaphore = new Semaphore(3);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    semaphore.acquire(3); // 获取多个许可
                    test(threadNum);
                    semaphore.release(3); // 释放多个许可
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        log.info("{}", threadNum);
        Thread.sleep(1000);
    }
}

import lombok.extern.slf4j.Slf4j;

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

@Slf4j
public class SemaphoreExample3 {

    private final static int threadCount = 20;

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

        ExecutorService exec = Executors.newCachedThreadPool();

        final Semaphore semaphore = new Semaphore(3);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    if (semaphore.tryAcquire()) { // 尝试获取一个许可
                        test(threadNum);
                        semaphore.release(); // 释放一个许可
                    }
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        log.info("{}", threadNum);
        Thread.sleep(1000);
    }
}

import lombok.extern.slf4j.Slf4j;

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

@Slf4j
public class SemaphoreExample4 {

    private final static int threadCount = 20;

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

        ExecutorService exec = Executors.newCachedThreadPool();

        final Semaphore semaphore = new Semaphore(3);

        for (int i = 0; i < threadCount; i++) {
            final int threadNum = i;
            exec.execute(() -> {
                try {
                    if (semaphore.tryAcquire(5000, TimeUnit.MILLISECONDS)) { // 尝试获取一个许可
                        test(threadNum);
                        semaphore.release(); // 释放一个许可
                    }
                } catch (Exception e) {
                    log.error("exception", e);
                }
            });
        }
        exec.shutdown();
    }

    private static void test(int threadNum) throws Exception {
        log.info("{}", threadNum);
        Thread.sleep(1000);
    }
}

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。因为CycliBarrier在释放等待线程后可以重用，所以称为循环barrier。CycliBarrier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。

使用场景类似于CountDownLatch与CountDownLatch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。

@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

import lombok.extern.slf4j.Slf4j;

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

@Slf4j
public class CyclicBarrierExample2 {

    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);
        try {
            barrier.await(2000, TimeUnit.MILLISECONDS);
        } catch (Exception e) {
            log.warn("BarrierException", e);
        }
        log.info("{} continue", threadNum);
    }
}

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

微信关注我们

原文链接：https://yq.aliyun.com/articles/636346

转载内容版权归作者及来源网站所有！

低调大师中文资讯倾力打造互联网数据资讯、行业资源、电子商务、移动互联网、网络营销平台。持续更新报道IT业界、互联网、市场资讯、驱动更新,是最及时权威的产业资讯及硬件资讯报道平台。

手写SpringMVC实战，一切从Spring底层源码分析开始

Spring框架对于Java后端程序员来说再熟悉不过了，以前只知道它用的反射实现的，但了解之后才知道有很多巧妙的设计在里面。如果不看Spring的源码，你将会失去一次和大师学习的机会：它的代码规范，设计思想很值得学习。我们程序员大部分人都是野路子，不懂什么叫代码规范。写了一个月的代码，最后还得其他老司机花3天时间重构，相信大部分老司机都很头疼看新手的代码。废话不多说，我们进入今天的正题，在Web应用程序设计中，MVC模式已经被广泛使用。SpringMVC以DispatcherServlet为核心，负责协调和组织不同组件以完成请求处理并返回响应的工作，实现了MVC模式。点击这里学习 Spring MVC 常用注解。想要实现自己的SpringMVC框架，需要从以下几点入手：了解SpringMVC运行流程及九大组件梳理自己的SpringMVC的设计思路实现自己的SpringMVC框架一、了解SpringMVC运行流程及九大组件 1、SpringMVC的运行流程 ⑴ 用户发送请求至前端控制器DispatcherServlet ⑵ DispatcherServlet收到请求调用H...

2018-08-15

468

首先要知道 Java 中的锁主要分两类锁 , 一种是 synchronize锁 , 另外一种就是 J.U.C中提供的锁 , J.U.C里核心的锁是 ReentrantLock ReentrantLock (可重入锁)与 synchronize 的区别可重入性 ReentrantLock 字面意思就是再进入锁 , 所以称之为可重入锁 , synchronize 使用的锁也是可重入的. 它俩都是同一个线程进入一次锁的计数器就自增 1,所以要等到锁的计数器下降为 0 时才释放锁 . 锁的实现 synchronize 的锁是基于 JVM 来实现的 , ReentrantLock 是jdk 实现的. 通俗的来讲就是操作系统来控制实现和用户编码实现的区别 . 性能区别在 synchronize 关键字优化之前, 其性能比 ReentrantLock 差 , 但是优化过后 , 在两者都可以使用的情况下, 建议使用 synchronize, 主要是其写法比较容易功能 synchronize 写起来更简洁 , 它是由编译器来实现锁的加锁和释放 , 而ReentrantLock 需要我们手工...

2018-08-15

534

资源下载

更多资源

Mario

马里奥是站在游戏界顶峰的超人气多面角色。马里奥靠吃蘑菇成长，特征是大鼻子、头戴帽子、身穿背带裤，还留着胡子。与他的双胞胎兄弟路易基一起，长年担任任天堂的招牌角色。

腾讯云软件源

为解决软件依赖安装时官方源访问速度慢的问题，腾讯云为一些软件搭建了缓存服务。您可以通过使用腾讯云软件源站来提升依赖包的安装速度。为了方便用户自由搭建服务架构，目前腾讯云软件源站支持公网访问和内网访问。

Spring

Spring框架（Spring Framework）是由Rod Johnson于2002年提出的开源Java企业级应用框架，旨在通过使用JavaBean替代传统EJB实现方式降低企业级编程开发的复杂性。该框架基于简单性、可测试性和松耦合性设计理念，提供核心容器、应用上下文、数据访问集成等模块，支持整合Hibernate、Struts等第三方框架，其适用范围不仅限于服务器端开发，绝大多数Java应用均可从中受益。

Sublime Text

Sublime Text具有漂亮的用户界面和强大的功能，例如代码缩略图，Python的插件，代码段等。还可自定义键绑定，菜单和工具栏。Sublime Text 的主要功能包括：拼写检查，书签，完整的 Python API ， Goto 功能，即时项目切换，多选择，多窗口等等。Sublime Text 是一个跨平台的编辑器，同时支持Windows、Linux、Mac OS X等操作系统。