ConcurrentLinkedQueue是Queue的一个安全实现.Queue中元素按FIFO原则进行排序.采用CAS操作,来保证元素的一致性.
数据结构为:单向链表.
变量使用volatile修改,保证内在可见性(happens-before,对变量的写操作对后续的读操作是可见的),同样也不会导致CPU指令的重排序.
private static class Node<E> {
volatile E item;
volatile Node<E> next;
/**
* Constructs a new node. Uses relaxed write because item can
* only be seen after publication via casNext.
*/
Node(E item) {
UNSAFE.putObject(this, itemOffset, item);
}
boolean casItem(E cmp, E val) {
return UNSAFE.compareAndSwapObject(this, itemOffset, cmp, val);
}
void lazySetNext(Node<E> val) {
UNSAFE.putOrderedObject(this, nextOffset, val);
}
boolean casNext(Node<E> cmp, Node<E> val) {
return UNSAFE.compareAndSwapObject(this, nextOffset, cmp, val);
}
// Unsafe mechanics
private static final sun.misc.Unsafe UNSAFE;
private static final long itemOffset;
private static final long nextOffset;
static {
try {
UNSAFE = sun.misc.Unsafe.getUnsafe();
Class k = Node.class;
itemOffset = UNSAFE.objectFieldOffset
(k.getDeclaredField("item"));
nextOffset = UNSAFE.objectFieldOffset
(k.getDeclaredField("next"));
} catch (Exception e) {
throw new Error(e);
}
}
}
/***
* Sets the value of the object field at the specified offset in the
* supplied object to the given value, with volatile store semantics.
* 设置obj对象中offset偏移地址对应的object型field的值为指定值。支持volatile store语义
*
* @param obj the object containing the field to modify.
* 包含需要修改field的对象
* @param offset the offset of the object field within <code>obj</code>.
* <code>obj</code>中object型field的偏移量
* @param value the new value of the field.
* field将被设置的新值
* @see #putObject(Object,long,Object)
*/
public native void putObjectVolatile(Object obj, long offset, Object value);
/***
* Compares the value of the object field at the specified offset
* in the supplied object with the given expected value, and updates
* it if they match. The operation of this method should be atomic,
* thus providing an uninterruptible way of updating an object field.
* 在obj的offset位置比较object field和期望的值,如果相同则更新。这个方法
* 的操作应该是原子的,因此提供了一种不可中断的方式更新object field。
*
* @param obj the object containing the field to modify.
* 包含要修改field的对象
* @param offset the offset of the object field within <code>obj</code>.
* <code>obj</code>中object型field的偏移量
* @param expect the expected value of the field.
* 希望field中存在的值
* @param update the new value of the field if it equals <code>expect</code>.
* 如果期望值expect与field的当前值相同,设置filed的值为这个新值
* @return true if the field was changed.
* 如果field的值被更改
*/
public native boolean compareAndSwapObject(Object obj, long offset,
Object expect, Object update);
/***
* Sets the value of the object field at the specified offset in the
* supplied object to the given value. This is an ordered or lazy
* version of <code>putObjectVolatile(Object,long,Object)</code>, which
* doesn't guarantee the immediate visibility of the change to other
* threads. It is only really useful where the object field is
* <code>volatile</code>, and is thus expected to change unexpectedly.
* 设置obj对象中offset偏移地址对应的object型field的值为指定值。这是一个有序或者
* 有延迟的<code>putObjectVolatile</cdoe>方法,并且不保证值的改变被其他线程立
* 即看到。只有在field被<code>volatile</code>修饰并且期望被意外修改的时候
* 使用才有用。
*
* @param obj the object containing the field to modify.
* 包含需要修改field的对象
* @param offset the offset of the object field within <code>obj</code>.
* <code>obj</code>中long型field的偏移量
* @param value the new value of the field.
* field将被设置的新值
*/
public native void putOrderedObject(Object obj, long offset, Object value);
构造head和tail都为一个空的Node
private transient volatile Node<E> head;
private transient volatile Node<E> tail;
public ConcurrentLinkedQueue() {
head = tail = new Node<E>(null);
}
因为只继承AbstractQueue,Queue所以只有add(e),remove,offer(e),poll.主要的方法.
public boolean offer(E e) {
checkNotNull(e);
final Node<E> newNode = new Node<E>(e);
for (Node<E> t = tail, p = t;;) {
Node<E> q = p.next;
if (q == null) {
// p is last node
if (p.casNext(null, newNode)) {
// Successful CAS is the linearization point
// for e to become an element of this queue,
// and for newNode to become "live".
if (p != t) // hop two nodes at a time
casTail(t, newNode); // Failure is OK.
return true;
}
// Lost CAS race to another thread; re-read next
}
else if (p == q)
// We have fallen off list. If tail is unchanged, it
// will also be off-list, in which case we need to
// jump to head, from which all live nodes are always
// reachable. Else the new tail is a better bet.
p = (t != (t = tail)) ? t : head;
else
// Check for tail updates after two hops.
p = (p != t && t != (t = tail)) ? t : q;
}
}
放入到Queue整个过程.
1.获取尾结点t,设置p=t,q为t的下一个结点,判断q是否为空.如果为空,说明为最后的结点,使用CAS更新p的next node,
如果成功,返回true;如果失败说明next node已经被更新,那么执行循环从1开始执行;如果成功执行2.
2.判断p和q是否为同一对象,如果不是使用CAS更新tail(并发时可能出现),更新失败说明已经有tail了.不管成功失败都返回true.
3.如果q不为null 并且p == q,如果tail没有改变,那么jump to head,重新检查,继续执行1
4.如果q不为null 并且p <> q,也就是说此时尾节点后面还有元素,那么就需要把尾节点往后移,继续执行1.
public E poll() {
restartFromHead:
for (;;) {
for (Node<E> h = head, p = h, q;;) {
E item = p.item;
if (item != null && p.casItem(item, null)) {
// Successful CAS is the linearization point
// for item to be removed from this queue.
if (p != h) // hop two nodes at a time
updateHead(h, ((q = p.next) != null) ? q : p);
return item;
}
else if ((q = p.next) == null) {
updateHead(h, p);
return null;
}
else if (p == q)
continue restartFromHead;
else
p = q;
}
}
}
从Queue取item整个过程。
1.p = head,如果p.item不过空,那么通过CAS将p.item更新为null,如果更新成功返回item。
2.如果p.item为空,判断p.next是否为空,如果为空更新head为p。
3.如果p.next不为空,那么p = q(p.next).重新执行1。
4.如果p.next不为空,当p==q说明1中更新没有成功,重新执行1。
public int size() {
int count = 0;
for (Node<E> p = first(); p != null; p = succ(p))
if (p.item != null)
// Collection.size() spec says to max out
if (++count == Integer.MAX_VALUE)
break;
return count;
}
获取队列大小的过程,由于没有一个计数器来对队列大小计数,所以获取队列的大小只能通过从头到尾完整的遍历队列,代价是很大的。所以通常情况下ConcurrentLinkedQueue需要和一个AtomicInteger搭配来获取队列大小。
需要说明的是,对尾节点的tail的操作需要换成临时变量t和s,一方面是为了去掉volatile变量的可变性,另一方面是为了减少volatile的性能影响。