服务器之家:专注于服务器技术及软件下载分享
分类导航

PHP教程|ASP.NET教程|JAVA教程|ASP教程|编程技术|正则表达式|

服务器之家 - 编程语言 - JAVA教程 - Java HashMap源码及并发环境常见问题解决

Java HashMap源码及并发环境常见问题解决

2020-09-15 13:53yaominghui JAVA教程

这篇文章主要介绍了Java HashMap源码及并发环境常见问题解决,文中通过示例代码介绍的非常详细,对大家的学习或者工作具有一定的参考学习价值,需要的朋友可以参考下

HashMap源码简单分析:

1 一切需要从HashMap属性字段说起:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
/** The default initial capacity - MUST be a power of two. 初始容量 */
  static final int DEFAULT_INITIAL_CAPACITY = 1 << 4; // aka 16
 
  /**
   * The maximum capacity, used if a higher value is implicitly specified
   * by either of the constructors with arguments.
   * MUST be a power of two <= 1<<30. 最大容量
   */
  static final int MAXIMUM_CAPACITY = 1 << 30;
 
  /**
   * The load factor used when none specified in constructor.    * 默认的负载因子,当map的size>=负载因子*capacity时候并且插入元素时候的table[i]!=null进行扩容   * 扩容判断逻辑:java.util.HashMap#addEntry函数中   *
   */
  static final float DEFAULT_LOAD_FACTOR = 0.75f;
 
  /**
   * An empty table instance to share when the table is not inflated.
   */
  static final Entry<?,?>[] EMPTY_TABLE = {};
 
  /**
   * The table, resized as necessary. Length MUST Always be a power of two. 哈希表
   */
  transient Entry<K,V>[] table = (Entry<K,V>[]) EMPTY_TABLE;
 
  /**
   * The number of key-value mappings contained in this map. map的大小
   */
  transient int size;
 
  /**
   * The next size value at which to resize (capacity * load factor).
   * @serial
   */
  // If table == EMPTY_TABLE then this is the initial capacity at which the
  // table will be created when inflated. 扩容的阈值 = capacity * 负载因子
  int threshold;
 
  /**
   * The load factor for the hash table. 负载因子,默认是0.75,可以在创建HashMap时候通过构造函数指定
   *
   * @serial
   */
  final float loadFactor;
 
  /**
   * The number of times this HashMap has been structurally modified
   * Structural modifications are those that change the number of mappings in
   * the HashMap or otherwise modify its internal structure (e.g.,
   * rehash). This field is used to make iterators on Collection-views of
   * the HashMap fail-fast. (See ConcurrentModificationException).   * 修改次数:例如进行rehash或者返回hashMap视图时候如果发生修改可以fast-fail
   */
  transient int modCount;
 
  /**
   * The default threshold of map capacity above which alternative hashing is
   * used for String keys. Alternative hashing reduces the incidence of
   * collisions due to weak hash code calculation for String keys.
   * <p/>
   * This value may be overridden by defining the system property
   * {@code jdk.map.althashing.threshold}. A property value of {@code 1}
   * forces alternative hashing to be used at all times whereas
   * {@code -1} value ensures that alternative hashing is never used.   * rehash时候判断的一个阈值
   */
  static final int ALTERNATIVE_HASHING_THRESHOLD_DEFAULT = Integer.MAX_VALUE;

2: 接下来查看一下HashMap的put方法:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
/**
   * Associates the specified value with the specified key in this map.
   * If the map previously contained a mapping for the key, the old
   * value is replaced.
   *
   * @param key key with which the specified value is to be associated
   * @param value value to be associated with the specified key
   * @return the previous value associated with <tt>key</tt>, or
   *     <tt>null</tt> if there was no mapping for <tt>key</tt>.
   *     (A <tt>null</tt> return can also indicate that the map
   *     previously associated <tt>null</tt> with <tt>key</tt>.)
   */
  public V put(K key, V value) {
    if (table == EMPTY_TABLE) {//初始化哈希表
      inflateTable(threshold);
    }
    if (key == null) //如果key 为null 存储到table[0]位置
      return putForNullKey(value);
    int hash = hash(key); //计算hash值
    int i = indexFor(hash, table.length);//计算entry在table中的位置
    //for循环逻辑用于修改key对应的value的
    for (Entry<K,V> e = table[i]; e != null; e = e.next) {
      Object k;
      if (e.hash == hash && ((k = e.key) == key || key.equals(k))) {
        V oldValue = e.value;
        e.value = value;
        e.recordAccess(this);
        return oldValue;//如果是更新返回旧值
      }
    }
    //修改次数++
    modCount++;
    //添加元素到哈希表中
    addEntry(hash, key, value, i);
    // 如果是添加元素则返回null
    return null;
  }

3 put中调用的inflateTable方法:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
/**
   * Inflates the table.
   */
  private void inflateTable(int toSize) {
    // Find a power of 2 >= toSize
    //计算大于等于toSize的最小的2的整数次幂的值
    int capacity = roundUpToPowerOf2(toSize);
    //计算扩容阈值
    threshold = (int) Math.min(capacity * loadFactor, MAXIMUM_CAPACITY + 1);
    //初始化哈希表
    table = new Entry[capacity];
    //更新一下rehash的判断条件,便于以后判断是否rehash
    initHashSeedAsNeeded(capacity);
  }

4 put方法中调用的indexFor方法:

?
1
2
3
4
5
6
7
8
/**
   * Returns index for hash code h. 返回哈希值对应的哈希表索引
   */
  static int indexFor(int h, int length) {
    // assert Integer.bitCount(length) == 1 : "length must be a non-zero power of 2";
   //使用&操作,而不使用取余原因:均匀分布在哈希表中 。length-1目的是:由于table的长度都是2的整数次幂进行扩容,length-1的二进制全是1,计算效率高
    return h & (length-1);
  }

5 put方法中调用的addEntry方法:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/**
   * Adds a new entry with the specified key, value and hash code to
   * the specified bucket. It is the responsibility of this
   * method to resize the table if appropriate.
   *
   * Subclass overrides this to alter the behavior of put method.
   */
  void addEntry(int hash, K key, V value, int bucketIndex) {
   //判断是否扩容,只有size大于等于阈值而且当前插入table[i]!=null(就是able[i]已经被占用则扩容)
   if ((size >= threshold) && (null != table[bucketIndex])) {
      resize(2 * table.length);
      hash = (null != key) ? hash(key) : 0;
      //如果需要扩容的话则需要更新再次重新计算哈希表位置
      bucketIndex = indexFor(hash, table.length);
    }
    //将值插入到哈希表中
    createEntry(hash, key, value, bucketIndex);
  }

6 addEntry方法中调用的createEntry方法:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
/**
   * Like addEntry except that this version is used when creating entries
   * as part of Map construction or "pseudo-construction" (cloning,
   * deserialization). This version needn't worry about resizing the table.
   *
   * Subclass overrides this to alter the behavior of HashMap(Map),
   * clone, and readObject.
   */
  void createEntry(int hash, K key, V value, int bucketIndex) {
    // 获取到哈希表指定位置
    Entry<K,V> e = table[bucketIndex];
    // 链表的头插入方式进行插入,插入逻辑在Entry的构造器中。然后将新节点存储到 table[bucketIndex]中
    table[bucketIndex] = new Entry<>(hash, key, value, e);
    size++;//更新size即可
  }

Entry构造器:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
/**
   *
   * @param h hash值
   * @param k key
   * @param v value
   * @param n 原始链表
   */
  Entry(int h, K k, V v, Entry<K,V> n) {
    value = v;
    //将原始链表接该节点后面
    next = n;
    key = k;
    hash = h;
  }

7 接下来看一下java.util.HashMap#addEntry扩容机制:

当进行扩容时候需要重新计算哈希值和在哈希表中的位置。

?
1
2
3
4
5
6
7
8
9
10
11
12
13
void addEntry(int hash, K key, V value, int bucketIndex) {
    //满足扩容条件进行扩容
    if ((size >= threshold) && (null != table[bucketIndex])) {
      //扩容,2倍进行扩容
      resize(2 * table.length);
      //重新计算哈数值
      hash = (null != key) ? hash(key) : 0;
      //重新计算哈希表中的位置
      bucketIndex = indexFor(hash, table.length);
    }
 
    createEntry(hash, key, value, bucketIndex);
  }

接下来看一下java.util.HashMap#resize方法:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
/**
   * Rehashes the contents of this map into a new array with a
   * larger capacity. This method is called automatically when the
   * number of keys in this map reaches its threshold.
   *
   * If current capacity is MAXIMUM_CAPACITY, this method does not
   * resize the map, but sets threshold to Integer.MAX_VALUE.
   * This has the effect of preventing future calls.
   *
   * @param newCapacity the new capacity, MUST be a power of two;
   *    must be greater than current capacity unless current
   *    capacity is MAXIMUM_CAPACITY (in which case value
   *    is irrelevant).
   */
  void resize(int newCapacity) {
    Entry[] oldTable = table;
    int oldCapacity = oldTable.length;
    if (oldCapacity == MAXIMUM_CAPACITY) {//判断当前old容量是否最最大容量,是的话更新阈值
      threshold = Integer.MAX_VALUE;
      return;
    }
    //创建新的表
    Entry[] newTable = new Entry[newCapacity];
    //元素转移,根据initHashSeedAsNeeded结果判断是否进行rehash
    transfer(newTable, initHashSeedAsNeeded(newCapacity));
    // 新表赋给table
    table = newTable;
    //更新阈值
    threshold = (int)Math.min(newCapacity * loadFactor, MAXIMUM_CAPACITY + 1);
  }

关于HashMap在并发情况下的常见问题,其实在多线程环境下使用HashMap本来就是有风险错误的,但是一般面试却喜欢这么问,下面列举一下自己印象中的常见问题:

1:在进行扩容时候,其他线程是否可以进行进行插入操作(多线程环境下可能会导致HashMap进入死循环,此处暂不考虑)?

答:首先HashMap就不是一个线程安全的容器,所以在多线程环境下使用就是错误的。其次在扩容时候可以进行插入的,但是不安全。例如:

当主线程在调用transfer方法进行复制元素:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/**
   * Transfers all entries from current table to newTable.
   */
  void transfer(Entry[] newTable, boolean rehash) {
    int newCapacity = newTable.length;
    for (Entry<K,V> e : table) {
      while(null != e) {
        Entry<K,V> next = e.next;
        if (rehash) {
          e.hash = null == e.key ? 0 : hash(e.key);
        }
        int i = indexFor(e.hash, newCapacity);
        e.next = newTable[i];
        newTable[i] = e;
        e = next;
      }
    }
  }

此时另一个线程在添加新元素是可以的,新元素添加到table中。如果子线程需要扩容的话可以进行扩容,然后将新容器赋给table。而此时主线程转移元素的工作就是将table中元素转移到newTable中。注意main线程的transfer方法:

如果main线程刚进入transfer方法时候newTable大小是32的话,由于子线程的添加操作导致table此时元素如果有128的话。则128个元素就会存储到大小为32的newTable中(此处不会扩容)。这就会导致HashMap性能下降!!!

可以使用多线程环境进行debug查看即可确定(推荐Idea的debug,的确强大,尤其是Evaluate Expression功能)。

2:进行扩容时候元素是否需要重新Hash?

这个需要具体情况判断,调用initHashSeedAsNeeded方法判断(判断逻辑这里先不介绍)。

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
/**
   * Rehashes the contents of this map into a new array with a
   * larger capacity. This method is called automatically when the
   * number of keys in this map reaches its threshold.
   *
   * If current capacity is MAXIMUM_CAPACITY, this method does not
   * resize the map, but sets threshold to Integer.MAX_VALUE.
   * This has the effect of preventing future calls.
   *
   * @param newCapacity the new capacity, MUST be a power of two;
   *    must be greater than current capacity unless current
   *    capacity is MAXIMUM_CAPACITY (in which case value
   *    is irrelevant).
   */
  void resize(int newCapacity) {
    Entry[] oldTable = table;
    int oldCapacity = oldTable.length;
    if (oldCapacity == MAXIMUM_CAPACITY) {
      threshold = Integer.MAX_VALUE;
      return;
    }
 
    Entry[] newTable = new Entry[newCapacity];
    //initHashSeedAsNeeded 判断是否需要重新Hash
    transfer(newTable, initHashSeedAsNeeded(newCapacity));
    table = newTable;
    threshold = (int)Math.min(newCapacity * loadFactor, MAXIMUM_CAPACITY + 1);
  }

然后进行转移元素:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
/**
   * Transfers all entries from current table to newTable.
   */
  void transfer(Entry[] newTable, boolean rehash) {
    int newCapacity = newTable.length;
    //多线程环境下,如果其他线程导致table快速扩大。newTable在此处无法扩容会导致性能下降。但是如果后面有再次调用put方法的话可以再次触发resize。
    for (Entry<K,V> e : table) {
      while(null != e) {
        Entry<K,V> next = e.next;
        if (rehash) { //判断是否需要重新Hash
          e.hash = null == e.key ? 0 : hash(e.key);
        }
        int i = indexFor(e.hash, newCapacity);
        e.next = newTable[i];
        newTable[i] = e;
        e = next;
      }
    }
  }

3:如何判断是否需要重新Hash?

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/**
   * Initialize the hashing mask value. We defer initialization until we
   * really need it.
   */
  final boolean initHashSeedAsNeeded(int capacity) {
 
    // hashSeed降低hash碰撞的hash种子,初始值为0
    boolean currentAltHashing = hashSeed != 0;
    //ALTERNATIVE_HASHING_THRESHOLD: 当map的capacity容量大于这个值的时候并满足其他条件时候进行重新hash
    boolean useAltHashing = sun.misc.VM.isBooted() && (capacity >= Holder.ALTERNATIVE_HASHING_THRESHOLD);
    //TODO 异或操作,二者满足一个条件即可rehash
    boolean switching = currentAltHashing ^ useAltHashing;
    if (switching) {
      // 更新hashseed的值
      hashSeed = useAltHashing ? sun.misc.Hashing.randomHashSeed(this) : 0;
    }
    return switching;
  }

4:HashMap在多线程环境下进行put操作如何导致的死循环?

死循环产生时机:

当两个线程同时需要进行扩容,而且对哈希表同一个桶(table[i])进行扩容时候,一个线程刚好确定e和next元素之后,线程被挂起。此时另一个线程得到cpu并顺利对该桶完成转移(需要要求被转移之后的线程1中的e和next指的元素在新哈希表的同一个桶中,此时e和next被逆序了)。接着线程从挂起恢复回来时候就会陷入死循环中。参考:https://coolshell.cn/articles/9606.html

产生原因:主要由于并发操作,对用一个桶的两个节点构成了环,导致对环进行无法转移完毕元素陷入死循环。

Java HashMap源码及并发环境常见问题解决

以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持服务器之家。

原文链接:https://www.cnblogs.com/leodaxin/p/7708618.html

延伸 · 阅读

精彩推荐