JDK1.7的实现
ConcurrentHashMap的数据结构是由一个Segment数组和多个HashEntry组成,如图:
Segment数组的意义就是将一个大的table分割成多个小的table来进行加锁,segment的结构和HashMap类似,是一种数组和链表的结构,一个segment里包含一个HashEntry数组,每个HashEntry是一个链表结构的元素,每个segment守护者一个HashEntry数组里的元素,当对HashEntry数组的数据进行修改,必须首先获得与它对应的segment锁
1.初始化
int sshift = 0;
int ssize = 1;
while (ssize < concurrencyLevel) {
++sshift;
ssize <<= 1;
}
int segmentShift = 32 - sshift;
int segmentMask = ssize - 1;
segments数组的长度ssize是通过concurrencyLevel计算得出的,ssize用位与运算来计算,so,segments数组的长度是2的n次方,concurrencyLevel的最大值为65536,这意味着segments数组的长度最大值为65536,对应的二进制为16位
sshift等于ssize从1向左移位的次数,默认concurrencyLevel等于16,1需要向左移位4次,so,是shift等于4;segmentShift用于定位参与散列运算的位数,这里等于28,因为ConcurrentHashMap里的hash方法输出的最大数是32位的,so,32-4=28;segmentMask是散列运算的掩码,掩码的二进制各个位都是1
2.get
ConcurrentHashMap的get操作和HashMap类似,只是第一次需要经过hash定位到segment的位置,然后再hash定位到HashEntry,遍历该HashEntry下的链表进行对比,成功就返回,不成功就返回null
可以看出,定位segment和定位HashEntry所使用的算法不一
public V get(Object key) {
Segment<K,V> s; // manually integrate access methods to reduce overhead
HashEntry<K,V>[] tab;
int h = hash(key);
long u = (((h >>> segmentShift) & segmentMask) << SSHIFT) + SBASE;
if ((s = (Segment<K,V>)UNSAFE.getObjectVolatile(segments, u)) != null &&
(tab = s.table) != null) {
for (HashEntry<K,V> e = (HashEntry<K,V>) UNSAFE.getObjectVolatile
(tab, ((long)(((tab.length - 1) & h)) << TSHIFT) + TBASE);
e != null; e = e.next) {
K k;
if ((k = e.key) == key || (e.hash == h && key.equals(k)))
return e.value;
}
}
return null;
}
3.put
put需要对共享变量进行写入操作,so,需要加锁。该方法首先定位到segment,在segment里进行插入操作。插入操作需要两步,第一判断是否需要对segment里的HashEntry数据进行扩容,第二定位添加元素的位置,然后将其放在HashEntry数组里
(1)是否需要扩容
在插入元素前会先判断segment里的HashEntry数组是否超过容量threshold,如果超过阈值则进行扩容,注意,segment扩容比HashMap更恰当,因为HashMap是在插入元素后判断元素是否已经达到容量的,如果达到了就进行扩容,但是很有可能扩容之后没有新元素插入,这时HashMap即进行了一次无效扩容
(2)如何扩容
首先创建一个容量是原来两倍的数组,然后将原数组里的元素进行再散列插入到新数组里,为了高效,ConcurrentHashMap不会对整个容器进行扩容,而只对某个segment进行扩容
static final class Segment<K,V> extends ReentrantLock implements Serializable {
public V put(K key, V value) {
Segment<K,V> s;
if (value == null)
throw new NullPointerException();
int hash = hash(key);
int j = (hash >>> segmentShift) & segmentMask;
if ((s = (Segment<K,V>)UNSAFE.getObject // nonvolatile; recheck
(segments, (j << SSHIFT) + SBASE)) == null) // in ensureSegment
s = ensureSegment(j);
return s.put(key, hash, value, false);
}
在将数据插入指定的HashEntry位置时,会通过继承的ReentrantLock的tryLock方法尝试去获取锁,如果获取成功就直接插入相应的位置,如果已经有线程获取该segment的锁,那么当前线程会以自旋的方式继续调用tryLock方法区获取锁,超过指定次数就挂起,等待唤醒
final V put(K key, int hash, V value, boolean onlyIfAbsent) {
HashEntry<K,V> node = tryLock() ? null :
scanAndLockForPut(key, hash, value);
V oldValue;
try {
HashEntry<K,V>[] tab = table;
int index = (tab.length - 1) & hash;
HashEntry<K,V> first = entryAt(tab, index);
for (HashEntry<K,V> e = first;;) {
if (e != null) {
K k;
if ((k = e.key) == key ||
(e.hash == hash && key.equals(k))) {
oldValue = e.value;
if (!onlyIfAbsent) {
e.value = value;
++modCount;
}
break;
}
e = e.next;
}
else {
if (node != null)
node.setNext(first);
else
node = new HashEntry<K,V>(hash, key, value, first);
int c = count + 1;
if (c > threshold && tab.length < MAXIMUM_CAPACITY)
rehash(node);
else
setEntryAt(tab, index, node);
++modCount;
count = c;
oldValue = null;
break;
}
}
} finally {
unlock();
}
return oldValue;
}
通过tryLock获得锁
private HashEntry<K,V> scanAndLockForPut(K key, int hash, V value) {
HashEntry<K,V> first = entryForHash(this, hash);
HashEntry<K,V> e = first;
HashEntry<K,V> node = null;
int retries = -1; // negative while locating node
while (!tryLock()) {
HashEntry<K,V> f; // to recheck first below
if (retries < 0) {
if (e == null) {
if (node == null) // speculatively create node
node = new HashEntry<K,V>(hash, key, value, null);
retries = 0;
}
else if (key.equals(e.key))
retries = 0;
else
e = e.next;
}
else if (++retries > MAX_SCAN_RETRIES) {
lock();
break;
}
else if ((retries & 1) == 0 &&
(f = entryForHash(this, hash)) != first) {
e = first = f; // re-traverse if entry changed
retries = -1;
}
}
return node;
}
4.size
ConcurrentHashMap先尝试2次通过不锁住segment的方式来统计各个segment大小,如果统计的过程中容器的count发生了变化,则再采用加锁的方式来统计所有segment的大小
如何判断在统计的时候容器发生变化?使用modCount变量,在put、remove和clean方法里操作元素前都会将该变量加1,在统计size前后比较modCount是否发生变化,从而得知容器的大小是否发生变化
public int size() {
// Try a few times to get accurate count. On failure due to
// continuous async changes in table, resort to locking.
final Segment<K,V>[] segments = this.segments;
int size;
boolean overflow; // true if size overflows 32 bits
long sum; // sum of modCounts
long last = 0L; // previous sum
int retries = -1; // first iteration isn't retry
try {
for (;;) {
if (retries++ == RETRIES_BEFORE_LOCK) {
for (int j = 0; j < segments.length; ++j)
ensureSegment(j).lock(); // force creation
}
sum = 0L;
size = 0;
overflow = false;
for (int j = 0; j < segments.length; ++j) {
Segment<K,V> seg = segmentAt(segments, j);
if (seg != null) {
sum += seg.modCount;
int c = seg.count;
if (c < 0 || (size += c) < 0)
overflow = true;
}
}
if (sum == last)
break;
last = sum;
}
} finally {
if (retries > RETRIES_BEFORE_LOCK) {
for (int j = 0; j < segments.length; ++j)
segmentAt(segments, j).unlock();
}
}
return overflow ? Integer.MAX_VALUE : size;
}
5.扩容rehash
和HashMap的resize原理类似,避免让所有的结点都进行复制:扩容是基于2的幂指来操作,假设扩容前某HashEntry对应的segment中数组的index为i,数组容量为 capacity,那么扩容后该HashEntry对应到新数组中的index只可能为i或者i+capacity,因此大多数HashEntry结点在扩容前后index可以保持不变,基于此,rehash方法中会定位第一个后续所有结点在扩容后index都保持不变的结点,然后将这个结点之前的所有节点重排即可
JDK1.8的实现
首先取消了segment分段锁的数据结构,取而代之的是数组+链表(红黑树)的结构,而对于锁的粒度,调整为对每个数组元素加锁;然后是定位结点的hash算法被简化,这样带来的弊端是hash冲突会加剧,因此在链表节点数量大于8时,会将链表转为红黑树进行存储,这样一来,查询的时间复杂度就会从原来的O(n)变为O(logN)
1.相关属性
//负载因子
private static final float LOAD_FACTOR = 0.75f;
//链表转为红黑树的阈值,大于8则转为红黑树结构
static final int TREEIFY_THRESHOLD = 8;
//红黑树转链表的阈值
static final int UNTREEIFY_THRESHOLD = 6;
//sizeCtl用于table的初始化和扩容操作,不同值代表状态如下:
//-1表示正在初始化;-N表示有N-1个线程正在进行扩容操作
//非负情况:如果table未初始化,则表示table需要初始化的大小;如果初始化完成,则表示table扩容的阈值,默认为table容量的0.75倍
private transient volatile int sizeCtl;
构造函数,在创建ConcurrentHashMap时,并没有初始化table数组,只对Map容量、并发级别等做了赋值操作
public ConcurrentHashMap(int initialCapacity,
float loadFactor, int concurrencyLevel) {
if (!(loadFactor > 0.0f) || initialCapacity < 0 || concurrencyLevel <= 0)
throw new IllegalArgumentException();
if (initialCapacity < concurrencyLevel) // Use at least as many bins
initialCapacity = concurrencyLevel; // as estimated threads
long size = (long)(1.0 + (long)initialCapacity / loadFactor);
int cap = (size >= (long)MAXIMUM_CAPACITY) ?
MAXIMUM_CAPACITY : tableSizeFor((int)size);
this.sizeCtl = cap;
}
2.Node
Node是ConcurrentHashMap存储结构的基本单元,只允许对数据进行查找,不允许修改
static class Node<K,V> implements Map.Entry<K,V> {
//链表的数据结构
final int hash;
final K key;
//val和next都会在扩容时发生变化,so加上volatile来保持可见性和禁止重排序
volatile V val;
volatile Node<K,V> next;
Node(int hash, K key, V val, Node<K,V> next) {
this.hash = hash;
this.key = key;
this.val = val;
this.next = next;
}
public final K getKey() { return key; }
public final V getValue() { return val; }
public final int hashCode() { return key.hashCode() ^ val.hashCode(); }
public final String toString(){ return key + "=" + val; }
//不允许更新value
public final V setValue(V value) {
throw new UnsupportedOperationException();
}
public final boolean equals(Object o) {
Object k, v, u; Map.Entry<?,?> e;
return ((o instanceof Map.Entry) &&
(k = (e = (Map.Entry<?,?>)o).getKey()) != null &&
(v = e.getValue()) != null &&
(k == key || k.equals(key)) &&
(v == (u = val) || v.equals(u)));
}
/**
* Virtualized support for map.get(); overridden in subclasses.
*/
Node<K,V> find(int h, Object k) {
Node<K,V> e = this;
if (k != null) {
do {
K ek;
if (e.hash == h &&
((ek = e.key) == k || (ek != null && k.equals(ek))))
return e;
} while ((e = e.next) != null);
}
return null;
}
}
3.TreeNode
继承Node,但是数据结构换成了二叉树结构,是红黑树的存储结构。ConcurrentHashMap链表转树时,并不会直接转,只是把这些节点包装成TreeNode放到TreeBin中,再由TreeBin转化红黑树
static final class TreeNode<K,V> extends Node<K,V> {
//树形结构的属性定义
TreeNode<K,V> parent; // red-black tree links
TreeNode<K,V> left;
TreeNode<K,V> right;
TreeNode<K,V> prev; // needed to unlink next upon deletion
boolean red;//标志红黑树的红结点
TreeNode(int hash, K key, V val, Node<K,V> next,
TreeNode<K,V> parent) {
super(hash, key, val, next);
this.parent = parent;
}
Node<K,V> find(int h, Object k) {
return findTreeNode(h, k, null);
}
/**
* 根据key查找,从根节点开始找出相应的TreeNode
*/
final TreeNode<K,V> findTreeNode(int h, Object k, Class<?> kc) {
if (k != null) {
TreeNode<K,V> p = this;
do {
int ph, dir; K pk; TreeNode<K,V> q;
TreeNode<K,V> pl = p.left, pr = p.right;
if ((ph = p.hash) > h)
p = pl;
else if (ph < h)
p = pr;
else if ((pk = p.key) == k || (pk != null && k.equals(pk)))
return p;
else if (pl == null)
p = pr;
else if (pr == null)
p = pl;
else if ((kc != null ||
(kc = comparableClassFor(k)) != null) &&
(dir = compareComparables(kc, k, pk)) != 0)
p = (dir < 0) ? pl : pr;
else if ((q = pr.findTreeNode(h, k, kc)) != null)
return q;
else
p = pl;
} while (p != null);
}
return null;
}
}
4.TreeBin
封装TreeNode的容器,提供转换红黑树的一些条件和锁的控制,ConcurrentHashMap底层存放的就是TreeBin对象,而不是TreeNode对象
static final class TreeBin<K,V> extends Node<K,V> {
//指向TreeNode列表和根节点
TreeNode<K,V> root;
volatile TreeNode<K,V> first;
volatile Thread waiter;
volatile int lockState;
// 读写锁的状态
static final int WRITER = 1; // 获取写锁时
static final int WAITER = 2; // 等待写锁时
static final int READER = 4; // 增加数据时读锁的状态
/**
* Tie-breaking utility for ordering insertions when equal
* hashCodes and non-comparable. We don't require a total
* order, just a consistent insertion rule to maintain
* equivalence across rebalancings. Tie-breaking further than
* necessary simplifies testing a bit.
*/
static int tieBreakOrder(Object a, Object b) {
int d;
if (a == null || b == null ||
(d = a.getClass().getName().
compareTo(b.getClass().getName())) == 0)
d = (System.identityHashCode(a) <= System.identityHashCode(b) ?
-1 : 1);
return d;
}
/**
* 初始化红黑树
*/
TreeBin(TreeNode<K,V> b) {
super(TREEBIN, null, null, null);
this.first = b;
TreeNode<K,V> r = null;
for (TreeNode<K,V> x = b, next; x != null; x = next) {
next = (TreeNode<K,V>)x.next;
x.left = x.right = null;
if (r == null) {
x.parent = null;
x.red = false;
r = x;
}
else {
K k = x.key;
int h = x.hash;
Class<?> kc = null;
for (TreeNode<K,V> p = r;;) {
int dir, ph;
K pk = p.key;
if ((ph = p.hash) > h)
dir = -1;
else if (ph < h)
dir = 1;
else if ((kc == null &&
(kc = comparableClassFor(k)) == null) ||
(dir = compareComparables(kc, k, pk)) == 0)
dir = tieBreakOrder(k, pk);
TreeNode<K,V> xp = p;
if ((p = (dir <= 0) ? p.left : p.right) == null) {
x.parent = xp;
if (dir <= 0)
xp.left = x;
else
xp.right = x;
r = balanceInsertion(r, x);
break;
}
}
}
}
this.root = r;
assert checkInvariants(root);
}
5.put操作
发现,代码中加锁片段用的是synchronized关键字,而不是像1.7中的ReentrantLock,说明synchronized在新版本的JDK中优化的程度和ReentrantLock差不多了
ConcurrentHashMap在对key求Hash值的时候,为了实现segment均匀分布,进行了两次hash
static final int spread(int h) {
return (h ^ (h >>> 16)) & HASH_BITS;
}
public V put(K key, V value) {
return putVal(key, value, false);
}
/** Implementation for put and putIfAbsent */
final V putVal(K key, V value, boolean onlyIfAbsent) {
if (key == null || value == null) throw new NullPointerException();
//两次hash,减少hash冲突,可以均匀分布
int hash = spread(key.hashCode());
int binCount = 0;
//对table进行迭代
for (Node<K,V>[] tab = table;;) {//类似于while(true),这道插入成功
Node<K,V> f; int n, i, fh;
//上面构造方法是否进行初始化在这里判断,为null就调用initTable进行初始化,属于懒汉模式初始化
if (tab == null || (n = tab.length) == 0)
tab = initTable();
else if ((f = tabAt(tab, i = (n - 1) & hash)) == null) {//如果i位置没有数据,就直接无锁插入
if (casTabAt(tab, i, null,
new Node<K,V>(hash, key, value, null)))
break; // no lock when adding to empty bin
}
else if ((fh = f.hash) == MOVED)//如果在进行扩容,则先进行扩容操作
tab = helpTransfer(tab, f);
else {
V oldVal = null;
//如果以上都不满足,则进行加锁操作,也就是存在hash冲突,锁住链表或者红黑树的头结点
synchronized (f) {//锁定,hash值相同的链表的头结点
if (tabAt(tab, i) == f) {//避免多线程,需要重新检查
if (fh >= 0) {//表示该结点是链表结构
binCount = 1;
//该for循环先查找链表中是否出现了此key如果出现则更新value并跳出循环,否则将结点插入到链表末尾并跳出循环
for (Node<K,V> e = f;; ++binCount) {
K ek;
//相同的key进行put就会覆盖原先的value
if (e.hash == hash &&
((ek = e.key) == key ||
(ek != null && key.equals(ek)))) {
oldVal = e.val;
if (!onlyIfAbsent)
e.val = value;
break;
}
Node<K,V> pred = e;
if ((e = e.next) == null) {//插入链表尾部
pred.next = new Node<K,V>(hash, key,
value, null);
break;
}
}
}
else if (f instanceof TreeBin) {//红黑树结构
Node<K,V> p;
binCount = 2;
//红黑树结构旋转插入
if ((p = ((TreeBin<K,V>)f).putTreeVal(hash, key,
value)) != null) {
oldVal = p.val;
if (!onlyIfAbsent)
p.val = value;
}
}
}
}
//插入成功后,如果插入的是链表结点,需要判断下该桶位是否要转化为树
//如果链表的长度大于8就进行红黑树的转换
if (binCount != 0) {
if (binCount >= TREEIFY_THRESHOLD)
treeifyBin(tab, i);
if (oldVal != null)
return oldVal;
break;
}
}
}
addCount(1L, binCount);
return null;
}
从代码可以看出,put的大致步骤如下:
- 如果没有初始化,就先调用initTable方法进行初始化过程
- 如果没有hash冲突则直接CAS插入
- 如果还在进行扩容操作就先进行扩容
- 如果存在hash冲突,就加锁来保证线程安全,有两种情况,一是链表形式就直接遍历到尾端插入,二是按照红黑树结构插入
- 如果该链表的数量大于阈值8,就先转换成红黑树结构,break再一次进入循环
- 如果添加成功就调用addCount方法统计size,并检查是否需要扩容
treeifyBin方法,检查下table长度是否大于64,如果不大于,则调用tryPresize方法将table两倍扩容即可,就不将其转为树了,如果大于,则就将table[i]的链表转为红黑树
private final void treeifyBin(Node<K,V>[] tab, int index) {
Node<K,V> b; int n, sc;
if (tab != null) {
if ((n = tab.length) < MIN_TREEIFY_CAPACITY)//容量<64,则两倍扩容
tryPresize(n << 1);
else if ((b = tabAt(tab, index)) != null && b.hash >= 0) {
synchronized (b) {//读写锁
if (tabAt(tab, index) == b) {
TreeNode<K,V> hd = null, tl = null;
for (Node<K,V> e = b; e != null; e = e.next) {
TreeNode<K,V> p =
new TreeNode<K,V>(e.hash, e.key, e.val,
null, null);
if ((p.prev = tl) == null)
hd = p;
else
tl.next = p;
tl = p;
}
setTabAt(tab, index, new TreeBin<K,V>(hd));
}
}
}
}
}
不过在结构转换之前,会对数组长度进行判断
private final void tryPresize(int size) {
//若给定的容量>= (MAXIMUM_CAPACITY的一半,直接扩容到最大值,否则调用tableSizeFor扩容
int c = (size >= (MAXIMUM_CAPACITY >>> 1)) ? MAXIMUM_CAPACITY :
//tableSizeFor(count)作用是找到大于等于count的最小值
tableSizeFor(size + (size >>> 1) + 1);
int sc;
while ((sc = sizeCtl) >= 0) {//只有大于等于0才表示该线程可以扩容
Node<K,V>[] tab = table; int n;
if (tab == null || (n = tab.length) == 0) {//表示没有被初始化
n = (sc > c) ? sc : c;
//期间没有其他线程对表操作,则CAS将SIZECTL设为-1,表示正在初始化
if (U.compareAndSwapInt(this, SIZECTL, sc, -1)) {
try {
if (table == tab) {//再一次检查
@SuppressWarnings("unchecked")
Node<K,V>[] nt = (Node<K,V>[])new Node<?,?>[n];
table = nt;
sc = n - (n >>> 2);
}
} finally {
sizeCtl = sc;//更新扩容阈值
}
}
}
else if (c <= sc || n >= MAXIMUM_CAPACITY)
break;
else if (tab == table) {
int rs = resizeStamp(n);
if (sc < 0) {
Node<K,V>[] nt;
if ((sc >>> RESIZE_STAMP_SHIFT) != rs || sc == rs + 1 ||
sc == rs + MAX_RESIZERS || (nt = nextTable) == null ||
transferIndex <= 0)
break;
if (U.compareAndSwapInt(this, SIZECTL, sc, sc + 1))
transfer(tab, nt);
}
else if (U.compareAndSwapInt(this, SIZECTL, sc,
(rs << RESIZE_STAMP_SHIFT) + 2))
transfer(tab, null);
}
}
}
新增结点之后,会调用addCount方法记录元素个数,并检查是否需要进行扩容,当数组元素个数达到阈值时,会触发transfer方法,重新调整结点的位置,详细,参考:https://blog.youkuaiyun.com/elricboa/article/details/70199409
6.get操作
ForwardingNode表示临时结点,在扩容时使用
static final class ForwardingNode<K,V> extends Node<K,V> {
final Node<K,V>[] nextTable;
ForwardingNode(Node<K,V>[] tab) {
super(MOVED, null, null, null);
this.nextTable = tab;
}
Node<K,V> find(int h, Object k) {
// loop to avoid arbitrarily deep recursion on forwarding nodes
outer: for (Node<K,V>[] tab = nextTable;;) {
Node<K,V> e; int n;
public V get(Object key) {
Node<K,V>[] tab; Node<K,V> e, p; int n, eh; K ek;
int h = spread(key.hashCode());//定位到table中的i
if ((tab = table) != null && (n = tab.length) > 0 &&
(e = tabAt(tab, (n - 1) & h)) != null) {//读取首节点的node
if ((eh = e.hash) == h) {//如果该结点是首节点就返回
if ((ek = e.key) == key || (ek != null && key.equals(ek)))
return e.val;
}
//hash值为负值表示正在扩容,这个时候查的是ForwardingNode方法来定位到nextTable来查找,找得到就返回
else if (eh < 0)
return (p = e.find(h, key)) != null ? p.val : null;
while ((e = e.next) != null) {
if (e.hash == h &&
((ek = e.key) == key || (ek != null && key.equals(ek))))
return e.val;
}
}
return null;
}
- 计算hash,定位到该table索引值,如果是首节点符合就返回
- 如果遇到扩容,会调用标志正在扩容结点ForwardingNode的find方法,查找该结点,匹配就返回
- 以上都不符合的话,就往下遍历,匹配就返回,否则最后返回null
7.size
//1.2时就加入的
public int size() {
long n = sumCount();
return ((n < 0L) ? 0 :
(n > (long)Integer.MAX_VALUE) ? Integer.MAX_VALUE :
(int)n);
}
//1.8加入的API
public long mappingCount() {
long n = sumCount();
return (n < 0L) ? 0L : n; // ignore transient negative values
}
final long sumCount() {
CounterCell[] as = counterCells; CounterCell a;
long sum = baseCount;
if (as != null) {
for (int i = 0; i < as.length; ++i) {
if ((a = as[i]) != null)
sum += a.value;
}
}
return sum;
}
JDK1.8中新增了mappingCount的API,这个API的不同之处就是返回值是long类型,这样就不受Integer.MAX_VALUE大小的限制了,两个方法都调用sumCount,且返回的都是一个估计值(JDK1.7使用加锁方式实现,而1.8牺牲了精度来换取更高的效率)
总结
从JDK1.7的ReentrantLock+Segment+HashEntry到JDK1.8的synchronized+CAS+HashEntry+红黑树:
1.JDK1.8的实现降低了锁的粒度,JDK1.7版本的锁的粒度是基于segment的,包含多个HashEntry,而JDK1.8锁的粒度就是HashEntry
2.JDK1.8的数据结构变得更简单,使得操作也更加清晰流畅,因为已经使用synchronized来进行同步,所以不需要分段锁的概念,但是由于粒度的降低,实现的复杂度也相对应提高
3.JDK1.8使用红黑树来优化链表,基于长度很长的链表的遍历是一个漫长的过程,而红黑树的遍历效率时很快的,代替一定阈值的链表,形成最佳拍档
参考
https://www.cnblogs.com/study-everyday/p/6430462.html#autoid-0-0-0
https://blog.youkuaiyun.com/fouy_yun/article/details/77816587
https://blog.youkuaiyun.com/bolang789/article/details/79855053
https://blog.youkuaiyun.com/elricboa/article/details/70199409
《java并发编程的艺术》