ThreadLocal基本原理及应用

一、ThreadLocal是什么？

关于ThreadLocal是什么，先来看看官方对它的解释.

ThreadLocal并不是一个Thread，而是Thread的局部变量。当使用ThreadLocal维护变量时，ThreadLocal为每个使用该变量的线程提供独立的变量副本，所以每个线程都可以独立地改变自己的副本，而不影响其他线程所对应的副本。

ThreadLocal并不是用来并发控制访问一个共同对象，而是为了给每个线程分配一个只属于该线程的变量，顾名思义它是local variable（线程局部变量）。它的功用非常简单，就是为每一个使用该变量的线程都提供一个变量值的副本，使每一个线程都可以独立地改变自己的副本，而不会和其它线程的副本冲突，实现线程间的数据隔离。从线程的角度看，就好像每一个线程都完全拥有该变量。

ThreadLocal 的接口方法

二、基本原理

ThreadLocal是如何做到为每一个线程维护变量的副本的呢？其实实现的思路很简单：在ThreadLocal类中有一个Map，用于存储每一个线程的变量副本，Map中元素的键为线程对象，而值对应线程的变量副本。根据这种思路，可以实现一个ThreadLocal的模型。

public class SimpleThreadLocal {  
    private Map valueMap = Collections.synchronizedMap(new HashMap());  
    public void set(Object newValue) {  
        // 键为线程对象，值为本线程的变量副本  
        valueMap.put(Thread.currentThread(), newValue);  
    }  
    public Object get() {  
        Thread currentThread = Thread.currentThread();  
        //返回本线程对应的变量  
        Object o = valueMap.get(currentThread);   
        //如果在Map中不存在，放到Map中保存起来  
        if (o == null && !valueMap.containsKey(currentThread)) {  
            o = initialValue();  
            valueMap.put(currentThread, o);  
        }  
        return o;  
    }  
    public void remove() {  
        valueMap.remove(Thread.currentThread());  
    }  
    public Object initialValue() {  
        return null;  
    }  
}  

接下来从ThreadLocal源码的角度分析一下。

我们直接来看最常用的set方法

 /**
     * Sets the current thread's copy of this thread-local variable
     * to the specified value.  Most subclasses will have no need to
     * override this method, relying solely on the {@link #initialValue}
     * method to set the values of thread-locals.
     *
     * @param value the value to be stored in the current thread's copy of
     *        this thread-local.
     */
    public void set(T value) {
        Thread t = Thread.currentThread(); // 首先获取当前线程对象
        ThreadLocalMap map = getMap(t);  // 然后从当前线程对象中取出ThreadLocalMap
        if (map != null)
            map.set(this, value); // 如果当前ThreadLocalMap存在，就进行key/value的设置，key就是ThreadLocal对象，value就是线程中的ThreadLocalMap
        else
            createMap(t, value); // 如果不存在,那么就创建一个
    }

 /**
     * Get the map associated with a ThreadLocal. Overridden in
     * InheritableThreadLocal.
     *
     * @param  t the current thread
     * @return the map
     */
    ThreadLocalMap getMap(Thread t) {
        return t.threadLocals;
    }

    /**
     * Create the map associated with a ThreadLocal. Overridden in
     * InheritableThreadLocal.
     *
     * @param t the current thread
     * @param firstValue value for the initial entry of the map
     */
    void createMap(Thread t, T firstValue) {
        t.threadLocals = new ThreadLocalMap(this, firstValue);
    }

接下来看看get方法

 /**
     * Returns the value in the current thread's copy of this
     * thread-local variable.  If the variable has no value for the
     * current thread, it is first initialized to the value returned
     * by an invocation of the {@link #initialValue} method.
     *
     * @return the current thread's value of this thread-local
     */
    public T get() {
        Thread t = Thread.currentThread();
        ThreadLocalMap map = getMap(t);
        if (map != null) {
            ThreadLocalMap.Entry e = map.getEntry(this);
            if (e != null) {
                @SuppressWarnings("unchecked")
                T result = (T)e.value;
                return result;
            }
        }
        return setInitialValue();
    }

这里其实揭示了ThreadLocalMap里面的数据存储结构，从上面的代码来看，ThreadLocalMap中存放的就是Entry，Entry的KEY就是ThreadLocal，VALUE就是值。接着看ThreadLocalMap

可以看到ThreadLocalMap 中的 Entry使用了弱引用。为什么要使用弱引用呢？

      因为如果这里使用普通的key-value形式来定义存储结构，实质上就会造成节点的生命周期与线程强绑定，只要线程没有销毁，那么节点在GC分析中一直处于可达状态，没办法被回收，而程序本身也无法判断是否可以清理节点。如果一个对象没有强引用链可达，那么一般活不过下一次GC。当某个ThreadLocal已经没有强引用可达，则随着它被垃圾回收，在ThreadLocalMap里对应的Entry的键值会失效，这为ThreadLocalMap本身的垃圾清理提供了便利。

那么，这是否就意味着不会存在内存泄漏呢？

       每个thread中都存在一个map， map的类型是ThreadLocal.ThreadLocalMap。Map中的key为一个threadlocal实例. 这个Map的确使用了弱引用，不过弱引用只是针对key。 每个key都弱引用指向threadlocal， 当把threadlocal实例置为null以后，没有任何强引用指向threadlocal实例,所以threadlocal将会被gc回收. 但是,我们的value却不能回收,因为存在一条从current thread连接过来的强引用. 只有当前thread结束以后, current thread就不会存在栈中,强引用断开, Current Thread, Map, value将全部被GC回收。
       所以得出一个结论就是只要这个线程对象被gc回收，就不会出现内存泄露，但在threadLocal设为null和线程结束这段时间不会被回收的，就发生了我们认为的内存泄露。但是当线程对象不被回收的情况，这就发生了真正意义上的内存泄露。比如使用线程池的时候，线程结束是不会销毁的，会再次使用的，就可能出现内存泄露。

三、ThreadLocal的使用

在并发编程的时候，成员变量如果不做任何处理其实是线程不安全的，各个线程都在操作同一个变量，显然是不行的，并且我们也知道volatile这个关键字也是不能保证线程安全的。那么在有一种情况之下，我们需要满足这样一个条件：变量是同一个，但是每个线程都使用同一个初始值，也就是使用同一个变量的一个新的副本。这种情况之下ThreadLocal就非常使用，比如说DAO的数据库连接，我们知道DAO是单例的，那么他的属性Connection就不是一个线程安全的变量。而我们每个线程都需要使用他，并且各自使用各自的。这种情况，ThreadLocal就比较好的解决了这个问题。

应用场景：当很多线程需要多次使用同一个对象，并且需要该对象具有相同初始化值的时候最适合使用ThreadLocal。

常见ThreadLocal应用场景：以Handler为例来进行说明：Handler需要获取当前线程中的Looper对象，但是不同的线程中含有不同的Looper对象，这个时候使用ThreadLocal对Looper进行保存，那就实现了在不同的线程中读取到的Looper对象就是相应的那个线程中的。

接下来看看Looper的源码

public final class Looper {
    private static final String TAG = "Looper";


    // sThreadLocal.get() will return null unless you've called prepare().
    static final ThreadLocal<Looper> sThreadLocal = new ThreadLocal<Looper>();
    private static Looper sMainLooper;  // guarded by Looper.class


    final MessageQueue mQueue;
    final Thread mThread;
    volatile boolean mRun;


    private Printer mLogging;


     /** Initialize the current thread as a looper.
      * This gives you a chance to create handlers that then reference
      * this looper, before actually starting the loop. Be sure to call
      * {@link #loop()} after calling this method, and end it by calling
      * {@link #quit()}.
      */
    public static void prepare() {
        prepare(true);
    }


    private static void prepare(boolean quitAllowed) {
        if (sThreadLocal.get() != null) {
            throw new RuntimeException("Only one Looper may be created per thread");
        }
        sThreadLocal.set(new Looper(quitAllowed));
    }


    /**
     * Initialize the current thread as a looper, marking it as an
     * application's main looper. The main looper for your application
     * is created by the Android environment, so you should never need
     * to call this function yourself.  See also: {@link #prepare()}
     */
    public static void prepareMainLooper() {
        prepare(false);
        synchronized (Looper.class) {
            if (sMainLooper != null) {
                throw new IllegalStateException("The main Looper has already been prepared.");
            }
            sMainLooper = myLooper();
        }
    }


    /** Returns the application's main looper, which lives in the main thread of the application.
     */
    public static Looper getMainLooper() {
        synchronized (Looper.class) {
            return sMainLooper;
        }
    }


    /**
     * Run the message queue in this thread. Be sure to call
     * {@link #quit()} to end the loop.
     */
    public static void loop() {
        final Looper me = myLooper();
        if (me == null) {
            throw new RuntimeException("No Looper; Looper.prepare() wasn't called on this thread.");
        }
        final MessageQueue queue = me.mQueue;


        // Make sure the identity of this thread is that of the local process,
        // and keep track of what that identity token actually is.
        Binder.clearCallingIdentity();
        final long ident = Binder.clearCallingIdentity();


        for (;;) {
            Message msg = queue.next(); // might block
            if (msg == null) {
                // No message indicates that the message queue is quitting.
                return;
            }


            // This must be in a local variable, in case a UI event sets the logger
            Printer logging = me.mLogging;
            if (logging != null) {
                logging.println(">>>>> Dispatching to " + msg.target + " " +
                        msg.callback + ": " + msg.what);
            }


            msg.target.dispatchMessage(msg);


            if (logging != null) {
                logging.println("<<<<< Finished to " + msg.target + " " + msg.callback);
            }


            // Make sure that during the course of dispatching the
            // identity of the thread wasn't corrupted.
            final long newIdent = Binder.clearCallingIdentity();
            if (ident != newIdent) {
                Log.wtf(TAG, "Thread identity changed from 0x"
                        + Long.toHexString(ident) + " to 0x"
                        + Long.toHexString(newIdent) + " while dispatching to "
                        + msg.target.getClass().getName() + " "
                        + msg.callback + " what=" + msg.what);
            }


            msg.recycle();
        }
    }


    /**
     * Return the Looper object associated with the current thread.  Returns
     * null if the calling thread is not associated with a Looper.
     */
    public static Looper myLooper() {
        return sThreadLocal.get();
    }


    /**
     * Control logging of messages as they are processed by this Looper.  If
     * enabled, a log message will be written to <var>printer</var> 
     * at the beginning and ending of each message dispatch, identifying the
     * target Handler and message contents.
     * 
     * @param printer A Printer object that will receive log messages, or
     * null to disable message logging.
     */
    public void setMessageLogging(Printer printer) {
        mLogging = printer;
    }
    
    /**
     * Return the {@link MessageQueue} object associated with the current
     * thread.  This must be called from a thread running a Looper, or a
     * NullPointerException will be thrown.
     */
    public static MessageQueue myQueue() {
        return myLooper().mQueue;
    }


    private Looper(boolean quitAllowed) {
        mQueue = new MessageQueue(quitAllowed);
        mRun = true;
        mThread = Thread.currentThread();
    }


    /**
     * Returns true if the current thread is this looper's thread.
     * @hide
     */
    public boolean isCurrentThread() {
        return Thread.currentThread() == mThread;
    }


    /**
     * Quits the looper.
     * <p>
     * Causes the {@link #loop} method to terminate without processing any
     * more messages in the message queue.
     * </p><p>
     * Any attempt to post messages to the queue after the looper is asked to quit will fail.
     * For example, the {@link Handler#sendMessage(Message)} method will return false.
     * </p><p class="note">
     * Using this method may be unsafe because some messages may not be delivered
     * before the looper terminates.  Consider using {@link #quitSafely} instead to ensure
     * that all pending work is completed in an orderly manner.
     * </p>
     *
     * @see #quitSafely
     */
    public void quit() {
        mQueue.quit(false);
    }


    /**
     * Quits the looper safely.
     * <p>
     * Causes the {@link #loop} method to terminate as soon as all remaining messages
     * in the message queue that are already due to be delivered have been handled.
     * However pending delayed messages with due times in the future will not be
     * delivered before the loop terminates.
     * </p><p>
     * Any attempt to post messages to the queue after the looper is asked to quit will fail.
     * For example, the {@link Handler#sendMessage(Message)} method will return false.
     * </p>
     */
    public void quitSafely() {
        mQueue.quit(true);
    }


    /**
     * Posts a synchronization barrier to the Looper's message queue.
     *
     * Message processing occurs as usual until the message queue encounters the
     * synchronization barrier that has been posted.  When the barrier is encountered,
     * later synchronous messages in the queue are stalled (prevented from being executed)
     * until the barrier is released by calling {@link #removeSyncBarrier} and specifying
     * the token that identifies the synchronization barrier.
     *
     * This method is used to immediately postpone execution of all subsequently posted
     * synchronous messages until a condition is met that releases the barrier.
     * Asynchronous messages (see {@link Message#isAsynchronous} are exempt from the barrier
     * and continue to be processed as usual.
     *
     * This call must be always matched by a call to {@link #removeSyncBarrier} with
     * the same token to ensure that the message queue resumes normal operation.
     * Otherwise the application will probably hang!
     *
     * @return A token that uniquely identifies the barrier.  This token must be
     * passed to {@link #removeSyncBarrier} to release the barrier.
     *
     * @hide
     */
    public int postSyncBarrier() {
        return mQueue.enqueueSyncBarrier(SystemClock.uptimeMillis());
    }




    /**
     * Removes a synchronization barrier.
     *
     * @param token The synchronization barrier token that was returned by
     * {@link #postSyncBarrier}.
     *
     * @throws IllegalStateException if the barrier was not found.
     *
     * @hide
     */
    public void removeSyncBarrier(int token) {
        mQueue.removeSyncBarrier(token);
    }


    /**
     * Return the Thread associated with this Looper.
     */
    public Thread getThread() {
        return mThread;
    }


    /** @hide */
    public MessageQueue getQueue() {
        return mQueue;
    }


    /**
     * Return whether this looper's thread is currently idle, waiting for new work
     * to do.  This is intrinsically racy, since its state can change before you get
     * the result back.
     * @hide
     */
    public boolean isIdling() {
        return mQueue.isIdling();
    }


    public void dump(Printer pw, String prefix) {
        pw = PrefixPrinter.create(pw, prefix);
        pw.println(this.toString());
        pw.println("mRun=" + mRun);
        pw.println("mThread=" + mThread);
        pw.println("mQueue=" + ((mQueue != null) ? mQueue : "(null"));
        if (mQueue != null) {
            synchronized (mQueue) {
                long now = SystemClock.uptimeMillis();
                Message msg = mQueue.mMessages;
                int n = 0;
                while (msg != null) {
                    pw.println("  Message " + n + ": " + msg.toString(now));
                    n++;
                    msg = msg.next;
                }
                pw.println("(Total messages: " + n + ")");
            }
        }
    }


    public String toString() {
        return "Looper{" + Integer.toHexString(System.identityHashCode(this)) + "}";
    }