zoukankan      html  css  js  c++  java

  • Java高并发

    Java并发程序基础
    1. 线程的状态：new, runnable, blocked, waiting, timed_waiting, terminated。
    public enum State {
        NEW,
        RUNNABLE,
        BLOCKED,
        WAITING,
        TIMED_WAITING,
        TERMINATED;
    }
    1. Thread.run()和Thread.start()的区别是run不会开启新线程，在当前线程中，执行run()中的代码。 
    2. 新建线程：可以使用重载实现run()方法或者使用public Thread(Runnable runnable)来定义线程；
    3. 终止线程：不能直接使用stop()，因为在结束线程时，会直接终止线程，并会立即释放这个线程所持有的锁。而这些锁恰恰是用来维持对象的一致性。可以使用变量的方式实现：
    public class DemoThread extends Thread {
        volatile boolean stop = false; //退出线程标记
     
        public void stopThread() {
            stop = true;
        }
     
        @Override
        public void run() {
            while (true) {
                if (stop) {
                    break; //退出线程
                }
     
                synchronized (this) {
                    //执行逻辑
                }
     
                Thread.yield(); //谦让，线程退出CPU
            }
        }
    }
    1. 线程中断interrupt
    　　线程中断并不是立即退出，而是给线程发送一个通知，告知目标线程，希望退出。至于目标线程接收到通知后如何处理，则完全由目标线程自行决定。API如下：
    public void Thread.interrupt() //中断线程
    public bool Thread.isInterrupted() //判断是否中断
    public void Thread.interrupted() //判断是否被中断，并清楚当前中断状态
    　　在循环体中添加中断判断并添加中断的处理，如果在循环体中出现类似wait和sleep的操作，则只能通过中断来识别。因为sleep和wait会抛出InterruptedException中断异常。InterruptedException不是运行时异常，程序必须捕获并处理它，当线程处于sleep、状态时，如果被中断，就会触发该异常。
    public static native void sleep(long millis) throws InterruptedException;
    public class InterruptAndStopThread {
        public static void main(String args[]) throws InterruptedException {
            Thread thread = new Thread() {
                @Override
                public void run() {
                    while (true) {
                        if (Thread.currentThread().isInterrupted()) {
                            break;
                        }
     
                        try {
                            Thread.sleep(600);
                        } catch (InterruptedException e) {
                            e.printStackTrace();
                            Thread.currentThread().interrupt(); //再次执行中断，置上中断标记位。异常后会清除该标记，如果不添加处理，则在下一个循环时无法捕获该中断
                        }
     
                        System.out.println("Running!");
                        Thread.yield();
                    }
                }
            };
            thread.start();
            Thread.sleep(2000);
            thread.interrupt();//中断
        }
    }
    1. 等待wait和通知notify
    　　为了支持多线程之间的协作，JDK在Object类中提供了wait和notify接口，任何对象都可以调用该方法。调用必须包含在对应的synchronizied语句中，无论是wait或者notify都需要首先获取目标对象的一个监视器。wait通过synchronizied获取监视器。调用wait后自动释放监视器；notify通过synchronizied获取监视器，完毕后释放监视器。
    public final void wait() throws InterruptedException
    public final native void notify();
    public class SimpleWaitAndNotify {
        final static Object object = new Object();
     
        public static class Thread1 extends Thread {
            public void run() {
                synchronized (object) {
                    System.out.println(System.currentTimeMillis() + ":thread1 start !");
     
                    try {
                        System.out.println(System.currentTimeMillis() + ":thread1 wait for object !");
                        object.wait();
                    } catch (InterruptedException e) {
                        e.printStackTrace();
                    }
     
                    System.out.println(System.currentTimeMillis() + ":thread1 end!");
                }
            }
        }
     
        public static class Thread2 extends Thread {
            public void run() {
                synchronized (object) {
                    System.out.println(System.currentTimeMillis() + ":thread2 start ! notify one thread");
                    object.notify();
     
                    try {
                        Thread.sleep(2000);
                    } catch (InterruptedException e) {
                        e.printStackTrace();
                    }
     
                    System.out.println(System.currentTimeMillis() + ":thread2 end!");
                }
            }
        }
     
        public static void main(String args[]) {
            Thread thread1 = new Thread1();
            Thread thread2 = new Thread2();
            thread1.start();
            thread2.start();
        }
    }
    　　得到的结果是：thread1 start ! -> thread1 wait for object ! -> thread2 start ! notify one thread -> thread2 end! -> thread1 end!
    　　wait和sleep的区别是：wait会自动释放锁，而sleep不会。
    1. suspend和resume
    　　JDK已经废弃的接口，suspend在导致线程暂停的同时，不会去释放任何锁资源。如果resume操作意外在suspend之前执行，那么被挂起的线程可能很难有机会继续执行。
    1. 等待线程结束join和谦让yield
    　　join：线程需要等待依赖线程执行完毕。join的本质是让调用线程wait在当前线程对象实例上，线程执行完毕后，被等待的线程会在退出前notifyAll通知所有的等待线程继续执行。
    　　yield：是个静态方法，一旦执行，会使当前线程让出CPU。
    public final void join() throws InterruptedException
    public final synchronized void join(long millis) throws InterruptedException
    public static native void yield()
    public class JoinMain {
        public volatile static int i = 0;
        public static class AddThread extends Thread {
            public void run() {
                System.out.println("add!");
                for (i = 0; i < 1000000; i++) ;
                try {
                    Thread.sleep(500);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        }
        public static void main(String args[]) throws InterruptedException {
            AddThread at = new AddThread();
            at.start();
            at.join(); //主线程等待at线程执行完毕
            System.out.println(i);
        }
    }
    1. volatile
    　　为了确保变量被修改后，应用程序范围的所有线程都能够“看到”这个改动，虚拟机就必须采用特殊的手段，保证变量的可见性特点。volatile对保证操作的原子性有非常大的帮助，但是不能代替锁，无法保证一些复核操作的原子性。
    　　如果没有把握，可以直接用synchronizied。例如：
    public class ThreadVolatile extends Thread {
        public static volatile int n = 0;
        public void run() {
            for (int i = 0; i < 10; i++)
                try {
                    n = n + 1;
                    sleep(3); // 为了使运行结果更随机，延迟3毫秒
                } catch (Exception e) {
                }
        }
     
        public static void main(String[] args) throws Exception {
            Thread threads[] = new Thread[100];
            for (int i = 0; i < threads.length; i++)
                // 建立100个线程
                threads[i] = new ThreadVolatile();
            for (int i = 0; i < threads.length; i++)
                // 运行刚才建立的100个线程
                threads[i].start();
            for (int i = 0; i < threads.length; i++)
                // 100个线程都执行完后继续
                threads[i].join();
            System.out.println("n=" + ThreadVolatile.n);
        }
    }
    　　如果对n的操作是原子级别的，最后输出的结果应该为n=1000，而在执行上面代码时，很多时侯输出的n都小于1000，这说明n=n+1不是原子级别的操作。原因是声明为volatile的简单变量如果当前值由该变量以前的值相关，那么volatile关键字不起作用，也就是说如下的表达式都不是原子操作：
    n = n + 1;
    n++;
    　　如果要想使这种情况变成原子操作，需要使用synchronized关键字，如上的代码可以改成如下的形式：
    public class ThreadVolatile2 extends Thread {
        public static volatile int n = 0;
        public static synchronized void inc() {
            n++;
        }
     
        public void run() {
            for (int i = 0; i < 10; i++)
                try {
                    inc(); // n = n + 1 改成了 inc();
                    sleep(3); // 为了使运行结果更随机，延迟3毫秒
                } catch (Exception e) {
                }
        }
     
        public static void main(String[] args) throws Exception {
            Thread threads[] = new Thread[100];
            for (int i = 0; i < threads.length; i++)
                // 建立100个线程
                threads[i] = new ThreadVolatile2();
            for (int i = 0; i < threads.length; i++)
                // 运行刚才建立的100个线程
                threads[i].start();
            for (int i = 0; i < threads.length; i++)
                // 100个线程都执行完后继续
                threads[i].join();
            System.out.println("n=" + ThreadVolatile2.n);
        }
    }
    1. 线程组
    public class ThreadGroupName implements Runnable {
        @Override
        public void run() {
            String groupAndName = Thread.currentThread().getThreadGroup().getName() + "-" + Thread.currentThread().getName();
            while (true) {
                System.out.println("I am " + groupAndName);
                try {
                    Thread.sleep(3000);
                } catch (InterruptedException e) {
                    e.printStackTrace();
                }
            }
        }
     
        public static void main(String args[]) {
            ThreadGroup tg = new ThreadGroup("PrintGroup");
            Thread t1 = new Thread(tg, new ThreadGroupName(), "T1");
            Thread t2 = new Thread(tg, new ThreadGroupName(), "T2");
            t1.start();
            t2.start();
            System.out.println(tg.activeCount());
            Thread t3 = new Thread(tg, new ThreadGroupName(), "T3");
            t3.start();
            System.out.println(tg.activeCount());
            tg.list();
        }
    }
    1. 守护线程：setDaemon需要在start之前设置，否则抛异常。
    1. 线程优先级：setPriority
    /**
      * The minimum priority that a thread can have.
      */
    public final static int MIN_PRIORITY = 1;
     
    /**
      * The default priority that is assigned to a thread.
      */
    public final static int NORM_PRIORITY = 5;
     
    /**
      * The maximum priority that a thread can have.
      */
    public final static int MAX_PRIORITY = 10;
    1. 线程安全和synchronizied
    　　用法：
    　　（1）指定加锁对象：进入同步代码前要获得给定对象；
    synchronizied(instance) {}
    　　（2）直接作用于实例对象：相当于对当前实例加锁，进入同步代码前要获得当前实例的锁；
    　　注意：因为加锁于实例对象，所以创建线程时需要传入同一个对象。
     
    public class DemoThread extends Thread {
        private static DemoThread demoThread = new DemoThread();
        private  static int i = 0;
        
        private synchronized void increase() {
            i++;
        }
     
        @Override
        public void run() {
            for (int j = 0; j < 1000; j++) {
                increase();
            }
        }
        
        public static void main(String[] args) throws InterruptedException {
            Thread t1 = new Thread(demoThread);
            Thread t2 = new Thread(demoThread);
            t1.start();
            t2.start();
            t1.join();
            t2.join();
            System.out.println(i);
        }
    }
    　　（3）直接作用于静态方法：相当于当前类加锁，进入同步代码前要获得当前类的锁；
    public class DemoThread extends Thread {
        private static int i = 0;
     
        private static synchronized void increase() {
            i++;
        }
     
        @Override
        public void run() {
            for (int j = 0; j < 1000; j++) {
                increase();
            }
        }
     
        public static void main(String[] args) throws InterruptedException {
            Thread t1 = new Thread(new DemoThread());
            Thread t2 = new Thread(new DemoThread());
            t1.start();
            t2.start();
            t1.join();
            t2.join();
            System.out.println(i);
        }
    }
    1. 并发下的ArrayList和HashMap
    public class ArrayListMultiThread {
        static ArrayList<Integer> arrayList = new ArrayList<Integer>(10);
        public static class AddThread implements Runnable {
            @Override
            public void run() {
                for (int i = 0; i < 10000; i++) {
                    arrayList.add(i);
                }
            }
        }
     
        public static void main(String args[]) throws InterruptedException {
            Thread thread1 = new Thread(new AddThread());
            Thread thread2 = new Thread(new AddThread());
            thread1.start();
            thread2.start();
            thread1.join();
            thread2.join();
            System.out.println(arrayList.size());
        }
    }
    　　出现的结果可能有：
    • 正常；
    • ArrayList在扩容时，内部一致性被破坏，由于没有锁对象，另一个线程访问到了不一致的内部状态，导致越界问题；
    • 出现隐蔽错误，ArrayList的大小小于20000。
    　　最简单的方式是：使用线程安全的Vector代替ArrayList。
    public class HashMapMultiThread {
        static Map<String, String> map = new HashMap<String, String>();
        public static class AddThread implements Runnable {
            int start = 0;
            public AddThread(int start) {
                this.start = start;
            }
            @Override
            public void run() {
                for (int i = start; i < 100000; i += 2) {
                    map.put(Integer.toString(i), Integer.toBinaryString(i));
                }
            }
        }
     
        public static void main(String args[]) throws InterruptedException {
            Thread thread1 = new Thread(new HashMapMultiThread.AddThread(0));
            Thread thread2 = new Thread(new HashMapMultiThread.AddThread(1));
            thread1.start();
            thread2.start();
            thread1.join();
            thread2.join();
            System.out.println(map.size());
        }
    }
    　　结果可能是：
    • 正常；
    • 正常退出，小于10000，
    • 程序永远不能结束，当两个线程正在遍历HashMap的内部数据，put方法类似于链表的遍历，由于多线程的冲突，导致链表结构遭到破坏，当链表成环时，就出现了死循坏。死循坏在JDK8中已经不存在，但是在多线程使用HashMap依然会导致内部数据不一致。
    　　最简单的方式是：使用ConcurrentHashMap代替HashMap。
     
     JDK并发包
    1. 重入锁ReentrantLock
    　　重入锁可以完全那替代synchronized关键字。提供了支持中断响应的接口，ReentrantLock的几个重要方法整理：
    • lock()：获得锁，如果锁已经被占用，则等待；
    • lockInterruptibly：获得锁，但优先响应中断；
    • tryLock()：尝试获得锁，如果成功则返回true，失败则返回false。不等待锁；
    • tryLock(long time, TimeUnit unit)：在给定的时间尝试获得锁；
    • unlock()：释放锁；
    • ReentrantLock(bool fair)：公平锁；
    public class IntLock implements Runnable {
        public static ReentrantLock lock1 = new ReentrantLock();
        public static ReentrantLock lock2 = new ReentrantLock();
        int lock;
        public IntLock(int lock) {
            this.lock = lock;
        }
     
        @Override
        public void run() {
            try {
                if (lock == 1) {
                    lock1.lockInterruptibly();
                    Thread.sleep(500);
                    lock2.lockInterruptibly();
                } else {
                    lock2.lockInterruptibly();
                    Thread.sleep(500);
                    lock1.lockInterruptibly();
                }
            } catch (InterruptedException e) {
                e.printStackTrace();
            } finally {
                if (lock1.isHeldByCurrentThread()) {
                    lock1.unlock();
                }
                if (lock2.isHeldByCurrentThread()) {
                    lock2.unlock();
                }
                System.out.println(Thread.currentThread().getId() + ":线程退出");
            }
        }
     
        public static void main(String args[]) throws InterruptedException {
            IntLock r1 = new IntLock(1);
            IntLock r2 = new IntLock(2);
            Thread thread1 = new Thread(r1);
            Thread thread2 = new Thread(r2);
            thread1.start();
            thread2.start();
            Thread.sleep(1000);
            thread2.interrupt();
        }
    　　上述代码分析：t1和t2启动之后，t1先占用lock1，再占用lock2；t2先占用lock2，再占用lock1，容易导致t1和t2处于相互等待状态，导致死锁。最后t2调用中断，最后t1能顺利执行，t2中断退出释放资源。
    public class TryLock implements Runnable {
        public static ReentrantLock lock1 = new ReentrantLock();
        public static ReentrantLock lock2 = new ReentrantLock();
        int lock;
        public TryLock(int lock) {
            this.lock = lock;
        }
        @Override
        public void run() {
            if (lock == 1) {
                while (true) {
                    if (lock1.tryLock()) {
                        try {
                            try {
                                Thread.sleep(500);
                            } catch (InterruptedException e) {
                                e.printStackTrace();
                            }
                            if (lock2.tryLock()) {
                                try {
                                    System.out.println(Thread.currentThread().getId() + ":My Job done;");
                                    return;
                                } finally {
                                    lock2.unlock();
                                }
                            }
                        } finally {
                            lock1.unlock();
                        }
                    }
                }
            } else {
                while (true) {
                    if (lock2.tryLock()) {
                        try {
                            try {
                                Thread.sleep(500);
                            } catch (InterruptedException e) {
                                e.printStackTrace();
                            }
                            if (lock1.tryLock()) {
                                try {
                                    System.out.println(Thread.currentThread().getId() + ":My Job done;");
                                    return;
                                } finally {
                                    lock1.unlock();
                                }
                            }
                        } finally {
                            lock2.unlock();
                        }
                    }
                }
            }
        }
     
        public static void main(String args[]) {
            TryLock r1 = new TryLock(1);
            TryLock r2 = new TryLock(2);
            Thread thread1 = new Thread(r1);
            Thread thread2 = new Thread(r2);
            thread1.start();
            thread2.start();
        }
    　　上述代码分析：使用tryLock，由于线程不会傻傻等待，而是不断尝试，只要执行时间足够长，线程总是会得到所需要的资源，从而正常执行。
    1. 重入锁的好搭档：Condition条件
    • await：是当前的线程等待，同时释放当前锁，当其他线程使用signal或者signalAll方法时，线程会重新获得锁并继续执行，或者被线程中断时，也会跳出等待，和Object.wait()相似；
    • awaitUninterruptibly：和await相似，但是不会在等待过程中响应中断；
    • signal：用于唤醒一个等待中的线程，和Object.notify()类似；
    • signalAll：唤醒所有等待中的线程；
    public class ReenterLockCondition implements Runnable {
        public static ReentrantLock lock = new ReentrantLock();
        public static Condition condition = lock.newCondition();
     
        @Override
        public void run() {
            try {
                lock.lock();
                condition.await();
                System.out.println("Thread is going on");
            } catch (InterruptedException e) {
                e.printStackTrace();
            } finally {
                lock.unlock();
            }
        }
     
        public static void main(String args[]) throws InterruptedException {
            ReenterLockCondition reenterLockCondition = new ReenterLockCondition();
            Thread thread1 = new Thread(reenterLockCondition);
            thread1.start();
            Thread.sleep(2000);
            lock.lock();
            condition.signal();
            lock.unlock();
        }
    }
    　　代码分析：await调用后，线程释放锁，等待sign释放锁后，线程获得锁并执行剩下的代码。
    1. 信号量Semaphore
    　　信号量是对锁的扩展，无论是内部锁synchronized还是重入锁ReentrantLock，一次只允许一个线程访问一个资源，而信号量却可以指定多个线程，同时访问一个资源。
    public Semaphore(int permits)：
    public Semaphore(int permits)：
    public void acquire(int permits)：
    public void acquireUninterruptibly(int permits)：
    public boolean tryAcquire()：
    public boolean tryAcquire(long timeout, TimeUnit unit)：
    public void release()：
    public class SemapDemo implements Runnable {
        final Semaphore semp = new Semaphore(5);
     
        @Override
        public void run() {
            try {
                semp.acquire();
                Thread.sleep(2000);
                System.out.println(Thread.currentThread().getId() + ":done!");
                semp.release();
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
        }
     
        public static void main(String args[]) {
            ExecutorService executorService = Executors.newFixedThreadPool(20);
            final SemapDemo demo = new SemapDemo();
            for (int i = 0; i < 20; i++) {
                executorService.submit(demo);
            }
        }
    }
    1. ReadWriteLock读写锁
    　　ReadWriteLock是JDK5中提供的读写分离锁，读写分离锁有效地减少了锁竞争，以提升系统的性能。
    public class ReadWriteLockDemo {
        private static Lock lock = new ReentrantLock();
        private static ReentrantReadWriteLock reentrantReadWriteLock = new ReentrantReadWriteLock();
        private static Lock readLock = reentrantReadWriteLock.readLock();
        private static Lock writeLock = reentrantReadWriteLock.writeLock();
        private int value;
     
        public Object handleRead(Lock lock) throws InterruptedException {
            try {
                lock.lock();
                Thread.sleep(1000);
                return value;
            } finally {
                lock.unlock();
            }
        }
     
        public void handleWrite(Lock lock, int index) throws InterruptedException {
            try {
                lock.lock();
                Thread.sleep(1000);
                value = index;
            } finally {
                lock.unlock();
            }
        }
        public static void main(String args[]) {
            final ReadWriteLockDemo demo = new ReadWriteLockDemo();
            Runnable readRunnable = new Runnable() {
                @Override
                public void run() {
                    try {
                        demo.handleRead(readLock);
                    } catch (InterruptedException e) {
                        e.printStackTrace();
                    }
                }
            };
     
            Runnable writeRunnable = new Runnable() {
                @Override
                public void run() {
                    try {
                        demo.handleWrite(writeLock, new Random().nextInt(100));
                    } catch (InterruptedException e) {
                        e.printStackTrace();
                    }
                }
            };
     
            for (int i = 0; i < 18; i++) {
                new Thread(readRunnable).start();
            }
     
            for (int i = 18; i < 20; i++) {
                new Thread(writeRunnable).start();
            }
        }
    }
    　　代码分析：读线程为并行执行，写线程为串行执行。读读不堵塞，读写和写写堵塞。
    1. 倒计时器：CountDownLatch
    　　用来控制线程等待，它可以让某个线程等待直到倒计时结束，再开始执行。
    public class CountDownLatchDemo implements Runnable {
        static final CountDownLatch end = new CountDownLatch(10); //构造等待的线程数
        static final CountDownLatchDemo demo = new CountDownLatchDemo();
     
        @Override
        public void run() {
            try {
                Thread.sleep(new Random().nextInt(3) * 1000);
                System.out.println("check complete");
                end.countDown();
            } catch (InterruptedException e) {
                e.printStackTrace();
            }
        }
        public static void main(String args[]) throws InterruptedException {
            ExecutorService executorService = Executors.newFixedThreadPool(10);
     
            for (int i = 0; i < 10; i++) {
                executorService.submit(demo);
            }
     
            //等待10个线程执行完毕，再继续执行
            end.await();
            System.out.println("Fire!");
            executorService.shutdown();
        }
    }
    1. 线程池
    public static ExecutorService newFixedThreadPool(int nThreads)：固定线程数的线程池
    public static ExecutorService newSingleThreadExecutor()：单线程线程池
    public static ExecutorService newCachedThreadPool()：根据实际情况调整线程数量的线程池
    public static ScheduledExecutorService newSingleThreadScheduledExecutor()：单线程任务线程池
    public static ScheduledExecutorService newScheduledThreadPool(int corePoolSize)：固定线程数的任务线程池
    /**
    * Creates a new {@code ThreadPoolExecutor} with the given initial
    * parameters.
    *
    * @param corePoolSize the number of threads to keep in the pool, even
    *        if they are idle, unless {@code allowCoreThreadTimeOut} is set
    * @param maximumPoolSize the maximum number of threads to allow in the
    *        pool
    * @param keepAliveTime when the number of threads is greater than
    *        the core, this is the maximum time that excess idle threads
    *        will wait for new tasks before terminating.
    * @param unit the time unit for the {@code keepAliveTime} argument
    * @param workQueue the queue to use for holding tasks before they are
    *        executed.  This queue will hold only the {@code Runnable}
    *        tasks submitted by the {@code execute} method.
    * @param threadFactory the factory to use when the executor
    *        creates a new thread
    * @param handler the handler to use when execution is blocked
    *        because the thread bounds and queue capacities are reached
    * @throws IllegalArgumentException if one of the following holds:
     
    *         {@code corePoolSize < 0}
     
    *         {@code keepAliveTime < 0}
     
    *         {@code maximumPoolSize <= 0}
     
    *         {@code maximumPoolSize < corePoolSize}
    * @throws NullPointerException if {@code workQueue}
    *         or {@code threadFactory} or {@code handler} is null
    */
    public ThreadPoolExecutor(int corePoolSize,
                              int maximumPoolSize,
                              long keepAliveTime,
                              TimeUnit unit,
                              BlockingQueue<Runnable> workQueue,
                              ThreadFactory threadFactory,
                              RejectedExecutionHandler handler)
    • corePoolSize：线程数；
    • maximumPoolSize：最大线程数；
    • keepAliveTime：当线程数量大于corePoolSize时，多余的空闲线程的存活时间。
    • unit：keepAliveTime的单位时间；
    • workQueue：任务列表，被提交但尚未被执行的任务；
    • threadFactory：线程工厂，用于创建线程；
    • handler：拒绝策略，当任务太多来不及处理，如何拒绝任务。
    　　根据任务队列的功能，ThreadPoolExecutor的构造函数中可以使用以下的几种BlockingQueue：
    （1）直接提交队列：SynchronousQueue，需要设置很大的maximumPoolSize值；
    （2）有界任务队列：ArrayBlockingQueue，线程数量为0~maximumPoolSize，当线程数量大于corePoolSize时，任务存放到任务队列，当总线程数大于maximumPoolSize时，则执行拒绝策略；
    （3）无界任务队列：LinkedBlockingQueue，线程数量最大值为corePoolSize，maximumPoolSize没有实际意义，当总线程数大于corePoolSize时，被存储到任务队列中；
    （4）优先任务队列：PriorityBlockingQueue，可以根据任务自身的优先级顺序先后执行。
     
    1. 线程异常信息的封装
    public class TraceThreadPoolExecutor extends ThreadPoolExecutor {
        public TraceThreadPoolExecutor(int corePoolSize,
                                       int maximumPoolSize,
                                       long keepAliveTime,
                                       TimeUnit unit,
                                       BlockingQueue<Runnable> workQueue) {
            super(corePoolSize, maximumPoolSize, keepAliveTime, unit, workQueue);
        }
     
        public void execute(Runnable task) {
            super.execute(wrap(task, clientTrace(), Thread.currentThread().getName()));
        }
     
        private Runnable wrap(final Runnable task, final Exception clientTrace, String name) {
            return () -> {
                try {
                    task.run();
                } catch (Exception e) {
                    clientTrace.printStackTrace();
                    throw e;
                }
            };
        }
     
        private Exception clientTrace() {
            return new Exception("Client stack trace");
        }
     
        public static void main(String[] args) {
            TraceThreadPoolExecutor traceThreadPoolExecutor = new TraceThreadPoolExecutor(0, Integer.MAX_VALUE, 0L, TimeUnit.SECONDS, new SynchronousQueue<>());
     
            for (int i = 0; i< 5; i++) {
                traceThreadPoolExecutor.execute(new DivTask(100, i));
            }
        }
     
        public static class DivTask implements Runnable {
            int a, b;
     
            public DivTask(int a, int b) {
                this.a = a;
                this.b = b;
            }
     
            @Override
            public void run() {
                double re = a / b;
                System.out.println(re);
            }
        }
    }
    1. JDK并发容器
    • ConcurrentHashMap：高效并发的HashMap；
    • CopyOnWriteArrayList：和ArrayList是一族的，在读多写少的场合，List性能非常好，远远好于Vector；
    • ConcurrentLinkedQueue：高效并发队列，使用链表实现，线程安全的LinkedList；
    • BlockingQueue：堵塞队列，适合作为数据共享的通道；
    • ConcurrentSkipListMap：跳表的实现，有序Map，使用跳表的数据结构进行快速查找。
     
    锁优化
    1. 提高锁的性能
    • 减小锁持有时间：减少某个锁的占用时间，以减少线程间互斥的功能，提高系统的吞吐量；
    • 减小锁粒度：减小锁粒度是削弱多线程锁竞争的有效手段，ConcurrentHashMap添加一个新的表项，首先根据hashcode得到该表项应该存放到哪个段中，然后对该段加锁，并完成put操作。默认有16段。
    • 读写分离锁来替换独占锁：通过读写锁ReadWriteLock提高系统性能。
    • 锁分离：读写锁根据读写操作功能上的不同，进行有效的分离，使用类似的分离思想，从而对独占锁进行分离。例如LinkedBlockingQueue，take()和put()的获取和存储操作，由于是链表，两个操作分别作用于队列的前端和尾端。
    • 锁粗化：将一连串连续地对同一个锁不断进行请求和释放的操作时，便会把所有的锁操作整合成对锁的一次请求，减少对锁的请求同步次数。
    1. 无锁CAS
     
    并发模式与算法
    1. 单例模式 
    public class StaticSingleton {
        private StaticSingleton() {
            System.out.println("StaticSingle is create");
        }
     
        private static class SingletonHolder {
            private static StaticSingleton instance = new StaticSingleton();
        }
     
        public static StaticSingleton getInstance() {
            return SingletonHolder.instance;
        }
    }
    2. 不变模式 
    不变模式的使用场景需要满足条件：
    • 当对象创建之后，其内部状态和数据不再发生改变；
    • 对象需要被共享，被多线程频繁访问；
    实现不变模式的方式如下：
    • 去除setter方法以及所有修改自身属性的方法；
    • 将所有属性设置为私有，并用final标记，确保其不可修改；
    • 确保没有子类可以重载修改它的行为；
    • 有一个可以创建完整对象的构造函数；
    public final class Product {
        private final String no;
        private final String name;
        private final String price;
     
        public Product(String no, String name, String price) {
            super();
            this.no = no;
            this.name = name;
            this.price = price;
        }
     
        public String getNo() {
            return no;
        }
     
        public String getName() {
            return name;
        }
     
        public String getPrice() {
            return price;
        }
    }
    • 生产者-消费者模式
    　　使用BufferingQueue实现。
     
    • 无锁的缓存架构Disruptor
    public class LongEvent {
        private long value;
        public void set(long value) {
            this.value = value;
        }
     
        public long getValue() {
            return value;
        }
    }
     
    public class LongEventFactory implements EventFactory<LongEvent> {
        public LongEvent newInstance() {
            return new LongEvent();
        }
    }
     
    public class LongEventHandler implements EventHandler<LongEvent> {
        public void onEvent(LongEvent event, long sequence, boolean endOfBatch) {
            System.out.println("Event: " + event.getValue());
        }
    }
     
    public class LongEventProducer {
        private final RingBuffer<LongEvent> ringBuffer;
     
        public LongEventProducer(RingBuffer<LongEvent> ringBuffer) {
            this.ringBuffer = ringBuffer;
        }
     
        /* 3.0之前的方法
        public void onData(ByteBuffer bb) {
            long sequence = ringBuffer.next();
            try {
                LongEvent event = ringBuffer.get(sequence);
                event.set(bb.getLong(0));
            } finally {
                ringBuffer.publish(sequence);
            }
        }
        */
     
        private static final EventTranslatorOneArg<LongEvent, ByteBuffer> TRANSLATOR_ONE_ARG = new EventTranslatorOneArg<LongEvent, ByteBuffer>() {
            @Override
            public void translateTo(LongEvent longEvent, long l, ByteBuffer byteBuffer) {
                longEvent.set(byteBuffer.getLong(0));
            }
        };
     
        public void onData(ByteBuffer byteBuffer) {
            ringBuffer.publishEvent(TRANSLATOR_ONE_ARG, byteBuffer);
        }
    }
     
    public class Main {
        public static void main(String[] args) throws Exception {
            Disruptor<LongEvent> disruptor = new Disruptor<LongEvent>(new LongEventFactory(),
                    1024, //RingBuffer的大小
                    Executors.newCachedThreadPool(), //消费者线程池
                    ProducerType.MULTI, //支持多事件发布者
                    new BlockingWaitStrategy());
     
            RingBuffer<LongEvent> ringBuffer = disruptor.getRingBuffer();
            disruptor.handleEventsWith(
                    new LongEventHandler(),
                    new LongEventHandler(),
                    new LongEventHandler(),
                    new LongEventHandler()
            );
            
            disruptor.start();
     
            LongEventProducer producer = new LongEventProducer(ringBuffer);
     
            ByteBuffer bb = ByteBuffer.allocate(8);
            for (long l = 0; true; l++) {
                bb.putLong(0, l);
                producer.onData(bb);
                Thread.sleep(1000);
            }
        }
    }
  • 相关阅读:
    哈夫曼树及哈夫曼编码
    01背包问题
    Java IO
    Java对象的复制三种方式
    TCP三次握手和四次挥手
    轻量级Java Web框架的实现原理
    Java并发
    消息队列
    赋值、浅拷贝、深拷贝
    Python文件操作(txtxlsxcsv)及os操作
  • 原文地址：https://www.cnblogs.com/liguochun/p/8433262.html
Copyright © 2011-2022 走看看