Collections—ConcurrentHashMap

取出数组中的值 return e.val;

让出线程资源Thread.yield();

失败重试

binCount = 2;

判断红黑树else if (eh < 0)

设置扩容阈值  sizeCtl = sc;

初始化数组tab = initTable();

ek = e.key) == key || (ek != null && key.equals(ek)

链表长度大于等于8binCount >= TREEIFY_THRESHOLD

binCount = 0)失败重试

false

true底层数组还未初始化

true当前数组不在扩容，且不是树，是链表

true

hash运算，让值尽可能分布均匀int hash = spread(key.hashCode());

false循环失败重试

遍历链表，向其中添加值binCount = 1;binCount ++;

sizeCtl最开始为0，表示还未初始化if ((sc = sizeCtl) < 0)

while ((tab = table) == null || tab.length == 0)循环失败重试机制

开始

return null

回到开始循环位置进行重试

else if (f instanceof TreeBin)

if (fh >= 0)

true进行底层数组初始化操作

退出循环break；

 concurrentHashMap.get(key)

判断hash冲突eh = e.hash) == h

算出key的hash值int h = spread(key.hashCode());

key或者value为null时抛出空指针异常 if (key == null || value == null) throw new NullPointerException();

返回初始化好的数组 return tab;

判断节点上的hash属性为-1else if ((fh = f.hash) == MOVED)

if (binCount != 0)

 synchronized (f)hash冲突时对数组的添加操作加同步锁

if (tab == null || (n = tab.length) == 0)

循环失败重试

扩容transfer()

值添加成功退出循环break;

break；退出循环

返回链表中的值while ((e = e.next) != null)if (e.hash == h &&                    ((ek = e.key) == key || (ek != null && key.equals(ek))))                    return e.val;