WeakHashMap和ThreadLocal内存泄漏中的弱引用运行原理
本文原创，如有引用，请指明出处。
WeakHashMap和ThreadLocal内存泄漏中的弱引用运行原理
WeakHashMap的内存泄漏问题
DefaultChannelPipeline中使用了WeakHashMap来作为缓存。事实上，WeakHashMap的设计理念与ThreadLocal很像。但是ThreadLocal重新设计了自己的实现，并没有直接使用WeakHashMap。同时，ThreadLocal存在着内存泄漏的问题。而网上关于WeakHashMap内存泄漏问题却谈的非常少。下面贴一个网上关于WeakHashMap内存泄漏问题的代码。出处在这里对WeakHashMap的使用不慎导致内存溢出分析
import java.util.WeakHashMap;
public class WeekTest {
public static class Locker {
private static WeakHashMap<String, Locker> lockerMap = new WeakHashMap<String, Locker>();
private final String id;
private Locker(String id) {
this.id= id;
}
public synchronized static Locker acquire(String id) {
Locker locker = lockerMap.get(id);
if (locker == null) {
locker = new Locker(id);
//lockerMap.put(id, locker); //问题代码，导致了entry.key == entry.value.id
lockerMap.put(new String(id), locker); //这是一种修改方式，保证了WeakHashMap中的key，没有被value直接或间接所引用
}
return locker;
}
public String getId() {
return this.id;
}
public static int getSize() {
return lockerMap.size();
}
}
public static void main(String[] args) {
for (int i = 0; i < 10000000; i++) {
Locker.acquire("abc" + i);
if (i % 10000 == 0) {
System.gc();
System.out.println(Locker.getSize()); //输出垃圾回收后的Map的Size
}
}
}
}
读者可以自己运行上述代码。比较结果，很容易发现内存泄漏了。事实上，这里出现了许多疑问。当一个对象仅仅被弱引用指定时，它将被回收。因此，它被用来设计ThreadLocalMap和WeakHashMap。二者原理十分类似。二者内部数据结构十分类似。ThreadLocalMap元数据设计Entry代码如下：
/**
* The entries in this hash map extend WeakReference, using
* its main ref field as the key (which is always a
* ThreadLocal object). Note that null keys (i.e. entry.get()
* == null) mean that the key is no longer referenced, so the
* entry can be expunged from table. Such entries are referred to
* as "stale entries" in the code that follows.
*/
static class Entry extends WeakReference<ThreadLocal<?>> {
/** The value associated with this ThreadLocal. */
Object value;
Entry(ThreadLocal<?> k, Object v) {
super(k);
value = v;
}
}
下面是WeakHashMap的源代码。这里只比较数据结构，会发现两个内部类十分相似。但是，ThreadLocal会引发内存泄漏，而WeakHashMap很少引发内存泄漏。除非使用错误，WeakHashMap中的key被value直接或间接所引用。在使用正确的情况下，WeakHashMap中的数据在key没有被强引用的情况下，回收器可以正确回收整个Entry的内存；ThreadLocal则必须在当前线程停止后才可以，否则回收器将仅仅回收key(Threshold)内存，value内存无法被回收。
private static class Entry<K,V> extends WeakReference<Object> implements Map.Entry<K,V> {
V value;
final int hash;
Entry<K,V> next;
/**
* Creates new entry.
*/
Entry(Object key, V value,
ReferenceQueue<Object> queue,
int hash, Entry<K,V> next) {
super(key, queue);
this.value = value;
this.hash = hash;
this.next = next;
}
//其他忽略
弱引用运行原理
为了理解这其中原因，我们需要查看三个类WeakReference，Reference，ReferenceQueue。WeakReference继承了Reference，
public class WeakReference<T> extends Reference<T> {
/**
* Creates a new weak reference that refers to the given object. The new
* reference is not registered with any queue.
*
* @param referent object the new weak reference will refer to
*/
public WeakReference(T referent) {
super(referent);
}
/**
* Creates a new weak reference that refers to the given object and is
* registered with the given queue.
*
* @param referent object the new weak reference will refer to
* @param q the queue with which the reference is to be registered,
* or <tt>null</tt> if registration is not required
*/
public WeakReference(T referent, ReferenceQueue<? super T> q) {
super(referent, q);
}
}
整个WeakReference的代码十分简单。主要的逻辑由Reference实现。WeakReference仅实现了两个构造器。Reference的实现较晦涩，但原理十分简单。我们将结合例子讲解。如果我们想实现一个字符串变成弱引用。实现如下：
String String = new String("Random"); // 强引用
ReferenceQueue<String> StringQ = new ReferenceQueue<String>();// 引用队列
WeakReference<String> StringWRefernce = new WeakReference<String>(String, StringQ);//弱引用
构造器有两个十分重要的属性。这里构造器第一个参数是new String("Random")。在Reference中，回收器线程将对这个对象进行监视，从而决定是否将Reference放入ReferenceQueue中。而这个ReferenceQueue正是传递给构造器的第二个属性。Reference对象为了自己专门定义了四个内部状态：Active-Pending-Enqueued-Inactive。当它被传进来那一刻，Reference对象的状态为Active。回收器开始运作。当可达性发生变化，Reference对象状态将被转化为Pending（这部分是由JVM实现的）。同时，Reference对象将被加入一个Pending队列（源码中中实现为链表）。Reference类还启动一个守护线程ReferenceHandle。这个线程负责将Pending队列中的Reference对象加入ReferenceQueue队列中。可以说ReferenceQueue中的Reference对象已经是无用对象。ReferenceQueue可以看作是一个Reference链表。同时加入其中的Reference对象已经不可达，自动回收。示例代码如下：（改自Reference Queues）
import java.lang.ref.ReferenceQueue;
import java.lang.ref.WeakReference;
import java.lang.reflect.Field;
public class WeekTest {
public static <T>boolean checkNotNull(T t,String fieldName){
try {
Field field=t.getClass().getDeclaredField(fieldName);
field.setAccessible(true);
Object result=field.get(t);
return result!=null;
} catch (NoSuchFieldException e) {
System.out.println("不存在该字段");
e.printStackTrace();
} catch (IllegalAccessException e) {
System.out.println("");
e.printStackTrace();
}
return false;
}
public static void main(String[] args) throws InterruptedException {
String String = new String("Random"); // a strong object
ReferenceQueue<String> StringQ = new ReferenceQueue<String>();// the ReferenceQueue
WeakReference<String> StringWRefernce = new WeakReference<String>(String, StringQ);
System.out.println("String created as a weak ref " + StringWRefernce);
System.out.println("Is refQ head not null?"+checkNotNull(StringQ,"head"));
Runtime.getRuntime().gc();
System.out.println("Any weak references in Q ? " + (StringQ.poll() != null));
String = null; // the only strong reference has been removed. The heap
// object is now only weakly reachable
System.out.println("Now to call gc...");
Runtime.getRuntime().gc(); // the object will be cleared here - finalize will be called.
System.out.println("Is refQ head not null?"+checkNotNull(StringQ,"head"));
//这里改用remove()方法是因为poll()是无阻塞方法。一个对象标记不可达到加入ReferenceQueue是分别由GC和线程ReferenceHandler分别来实现的，因此是存在延迟的。而remove()是阻塞方法。
System.out.println("Any weak references in Q ? " + (StringQ.remove() != null));
System.out.println("Does the weak reference still hold the heap object ? " + (StringWRefernce.get() != null));
System.out.println("Is the weak reference added to the ReferenceQ ? " + (StringWRefernce.isEnqueued()));
}
}
这里我们回到ThreadLocal中。ThreadLocal中的内部类Entry继承了WeakReference，传递给Reference监视的对象是ThreadLocal。value则是Entry内部的强引用。当外部ThreadLocal被设置为空时，回收器设置Reference为Pending，再加入ReferenceQueue，最后被回收。至始至终，我们传递给JVM监视并准备回收的都是ThreadLocal。Entry中的value并没有得到处理。因此如果线程一直没结束，那么就会存在Thread->ThreadLocalMap->Entry(null,value)的引用，造成内存泄漏。
为什么WeakHashMap能够回收value内存？很简单，WeakHashMap专门实现了一个方法
private void expungeStaleEntries() {
for (Object x; (x = queue.poll()) != null; ) {
synchronized (queue) {
@SuppressWarnings("unchecked")
Entry<K,V> e = (Entry<K,V>) x;
int i = indexFor(e.hash, table.length);
Entry<K,V> prev = table[i];
Entry<K,V> p = prev;
while (p != null) {
Entry<K,V> next = p.next;
if (p == e) {
if (prev == e)
table[i] = next;
else
prev.next = next;
// Must not null out e.next;
// stale entries may be in use by a HashIterator
e.value = null; // Help GC
size--;
break;
}
prev = p;
p = next;
}
}
}
}
WeakHashMap专门去翻找了ReferenceQueue。当发现了Reference中有某个key，说明它将被回收。那么e.value = null。直接将value置空，这样value也可以进行回收。
以上就是整个弱引用的运行原理。如果有兴趣，可以自己翻看jdk代码，对照本文。如果本文有误，欢迎指正。
补充
最后贴一个ReferenceQueue的使用代码
ReferenceQueue<Foo> fooQueue = new ReferenceQueue<Foo>();
class ReferenceWithCleanup extends WeakReference<Foo> {
Bar bar;
ReferenceWithCleanup(Foo foo, Bar bar) {
super(foo, fooQueue);
this.bar = bar;
}
public void cleanUp() {
bar.cleanUp();
}
}
public Thread cleanupThread = new Thread() {
public void run() {
while(true) {
ReferenceWithCleanup ref = (ReferenceWithCleanup)fooQueue.remove();
ref.cleanUp();
}
}
}
public void doStuff() {
cleanupThread.start();
Foo foo = new Foo();
Bar bar = new Bar();
ReferenceWithCleanup ref = new ReferenceWithCleanup(foo, bar);
... // From now on, once you release all non-weak references to foo,
// then at some indeterminate point in the future, bar.cleanUp() will
// be run. You can force it by calling ref.enqueue().
}
