java 多线程知识梳理4

常用并发辅助类 CountDownLatch Semaphore CyclicBarrier, 都基于ReentrantLock实现。

1 Semaphores       are often used to restrict the number of threads than can
 * access some (physical or logical) resource. For example, here is
 * a class that uses a semaphore to control access to a pool of items:

Semaphores 信号量, 相对常规同步块只能有一个线程对资源进行操作,Semaphores能允许多个线程同时操作资源。

public Semaphore(int permits) {          //参数permits表示许可数目,即同时可以允许多少线程进行访问
    sync = new NonfairSync(permits);
}
public Semaphore(int permits, boolean fair) {    //这个多了一个参数fair表示是否是公平的,即等待时间越久的越先获取许可
    sync = (fair)? new FairSync(permits) : new NonfairSync(permits);
}
public void acquire() throws InterruptedException {  }     //获取一个许可
public void acquire(int permits) throws InterruptedException { }    //获取permits个许可
public void release() { }          //释放一个许可
public void release(int permits) { }    //释放permits个许可
public boolean tryAcquire() { };    //尝试获取一个许可,若获取成功,则立即返回true,若获取失败,则立即返回false
public boolean tryAcquire(long timeout, TimeUnit unit) throws InterruptedException { };  //尝试获取一个许可,若在指定的时间内获取成功,则立即返回true,否则则立即返回false
public boolean tryAcquire(int permits) { }; //尝试获取permits个许可,若获取成功,则立即返回true,若获取失败,则立即返回false
public boolean tryAcquire(int permits, long timeout, TimeUnit unit) throws InterruptedException { }; //尝试获取permits个许可,若在指定的时间内获取成功,则立即返回true,否则则立即返回false
 
 * class Pool {
 *   private static final int MAX_AVAILABLE = 100;
 *   private final Semaphore available = new Semaphore(MAX_AVAILABLE, true);
 *
 *   public Object getItem() throws InterruptedException {
 *     available.acquire();//允许前100个线程进入方法,之后的线程阻塞
 *     return getNextAvailableItem();
 *   }
 *
 *   public void putItem(Object x) {
 *     if (markAsUnused(x))
 *       available.release();//是否一个许可,之后阻塞的线程可以调用getItem方法
 *   }
 *
 *   // Not a particularly efficient data structure; just for demo
 *
 *   protected Object[] items = ... whatever kinds of items being managed
 *   protected boolean[] used = new boolean[MAX_AVAILABLE];
 *
 *   protected synchronized Object getNextAvailableItem() {
 *     for (int i = 0; i < MAX_AVAILABLE; ++i) {
 *       if (!used[i]) {
 *          used[i] = true;
 *          return items[i];
 *       }
 *     }
 *     return null; // not reached
 *   }
 *
 *   protected synchronized boolean markAsUnused(Object item) {
 *     for (int i = 0; i < MAX_AVAILABLE; ++i) {
 *       if (item == items[i]) {
 *          if (used[i]) {
 *            used[i] = false;
 *            return true;
 *          } else
 *            return false;
 *       }
 *     }
 *     return false;
 *   }
 *
 * }
 
 
2 CountDownLatch 利用它可以实现类似计数器的功能
public CountDownLatch(int count) {  };  //参数count为计数值
public void await() throws InterruptedException { };   //调用await()方法的线程会被挂起,它会等待直到count值为0才继续执行
public boolean await(long timeout, TimeUnit unit) throws InterruptedException { };  //和await()类似,只不过等待一定的时间后count值还没变为0的话就会继续执行
public void countDown() { };  //将count值减1
public class Test {
     public static void main(String[] args) {   
         final CountDownLatch latch = new CountDownLatch(2);
          
         new Thread(){
             public void run() {
                 try {
                     System.out.println("子线程"+Thread.currentThread().getName()+"正在执行");
                    Thread.sleep(3000);
                    System.out.println("子线程"+Thread.currentThread().getName()+"执行完毕");
                    latch.countDown();
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
             };
         }.start();
          
         new Thread(){
             public void run() {
                 try {
                     System.out.println("子线程"+Thread.currentThread().getName()+"正在执行");
                     Thread.sleep(3000);
                     System.out.println("子线程"+Thread.currentThread().getName()+"执行完毕");
                     latch.countDown();
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
             };
         }.start();
          
         try {
             System.out.println("等待2个子线程执行完毕...");
            latch.await();
            System.out.println("2个子线程已经执行完毕");
            System.out.println("继续执行主线程");
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
     }
}
3 CyclicBarrier 栅栏,通过它可以实现让一组线程等待至某个状态之后再全部同时执行。 CyclicBarrier可以重用,CountDownLatch不可以
public CyclicBarrier(int parties, Runnable barrierAction) {//参数parties指让多少个线程或者任务等待至barrier状态;参数barrierAction为当这些线程都达到barrier状态时会执行的内容。
}
 
public CyclicBarrier(int parties) {
}
public int await() throws InterruptedException, BrokenBarrierException { };//用来挂起当前线程,直至所有线程都到达barrier状态再同时执行后续任务;
public int await(long timeout, TimeUnit unit)throws InterruptedException,BrokenBarrierException,TimeoutException { };
 * class Solver {
 *   final int N;
 *   final float[][] data;
 *   final CyclicBarrier barrier;
 *
 *   class Worker implements Runnable {
 *     int myRow;
 *     Worker(int row) { myRow = row; }
 *     public void run() {
 *       while (!done()) {
 *         processRow(myRow);
 *
 *         try {
 *           barrier.await();//设置当前线程为barried状态,barrier内置的计数器+1,直到N
 *         } catch (InterruptedException ex) {
 *           return;
 *         } catch (BrokenBarrierException ex) {
 *           return;
 *         }
 *       }
 *     }
 *   }
 *
 *   public Solver(float[][] matrix) {
 *     data = matrix;
 *     N = matrix.length;
 *     barrier = new CyclicBarrier(N,
 *                                 new Runnable() {//当N个线程处于barried状态后,执行后续的mergeRows任务。
 *                                   public void run() {
 *                                     mergeRows(...);
 *                                   }
 *                                 });
 *     for (int i = 0; i < N; ++i)
 *       new Thread(new Worker(i)).start();
        system.out.println("cyclicBarrier重复使用");
 *     for (int i = 0; i < N; ++i)
 *       new Thread(new Worker(i)).start();
 *
 *     waitUntilDone();
 *   }
 * }

郑重声明:本站内容如果来自互联网及其他传播媒体,其版权均属原媒体及文章作者所有。转载目的在于传递更多信息及用于网络分享,并不代表本站赞同其观点和对其真实性负责,也不构成任何其他建议。