fe2e6dc2f2
Generated comprehensive interview preparation materials covering: - Distributed systems (transactions, locks, ID generation, consistency) - Database (indexing, sharding, replication, transactions) - Caching (Redis, cache problems, distributed lock) - Message queues (RocketMQ, Kafka) - Concurrency (ThreadLocal, ConcurrentHashMap, thread pools) - JVM (GC, memory, tuning) - System design (seckill, short URL, IM, feed, LBS) - Algorithms (B+ tree, LRU, Red-Black tree, Skip list, Timing wheel) - Network (TCP/IP, HTTP/HTTPS) - Security (encryption, SQL injection, XSS) - Performance tuning - Design patterns - Microservices (Spring Boot, Gateway, Service Mesh) - Container orchestration (Kubernetes, Docker) - CI/CD, observability Each file includes: - Detailed questions - Comprehensive answers - Code examples - Real project experience - Alibaba P7 level requirements Generated with [Claude Code](https://claude.com/claude-code) via [Happy](https://happy.engineering) Co-Authored-By: Claude <noreply@anthropic.com> Co-Authored-By: Happy <yesreply@happy.engineering>
477 lines
12 KiB
Markdown
477 lines
12 KiB
Markdown
# ConcurrentHashMap 原理
|
||
|
||
## 问题
|
||
|
||
1. ConcurrentHashMap 在 JDK 1.7 和 1.8 中的实现有什么区别？
|
||
2. ConcurrentHashMap 如何保证线程安全？
|
||
3. ConcurrentHashMap 的 size() 方法是如何实现的？
|
||
4. ConcurrentHashMap 和 Hashtable、Collections.synchronizedMap 的区别？
|
||
5. 在实际项目中如何选择线程安全的 Map？
|
||
|
||
---
|
||
|
||
## 标准答案
|
||
|
||
### 1. JDK 1.7 vs 1.8
|
||
|
||
#### **JDK 1.7：分段锁**
|
||
|
||
**结构**：
|
||
```
|
||
ConcurrentHashMap
|
||
↓
|
||
Segment[]（分段数组，默认 16 个）
|
||
↓
|
||
HashEntry[]（每个 Segment 有自己的 HashEntry 数组）
|
||
↓
|
||
HashEntry（键值对）
|
||
```
|
||
|
||
**特点**：
|
||
- **分段锁**：每个 Segment 独立加锁（ReentrantLock）
|
||
- **并发度**：默认 16（Segment 数量）
|
||
- **粒度**：Segment 级别
|
||
|
||
**获取锁流程**：
|
||
```java
|
||
// 1. 计算 Segment 索引
|
||
int hash = hash(key.hashCode());
|
||
int segmentIndex = (hash >>> segmentShift) & segmentMask;
|
||
|
||
// 2. 获取 Segment
|
||
Segment segment = segments[segmentIndex];
|
||
|
||
// 3. 加锁
|
||
segment.lock();
|
||
try {
|
||
// 操作 HashEntry[]
|
||
} finally {
|
||
segment.unlock();
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
#### **JDK 1.8：CAS + synchronized**
|
||
|
||
**结构**：
|
||
```
|
||
ConcurrentHashMap
|
||
↓
|
||
Node[] + TreeNode[]（数组 + 链表 + 红黑树）
|
||
↓
|
||
Node / TreeNode
|
||
```
|
||
|
||
**特点**：
|
||
- **CAS + synchronized**：CAS 失败后使用 synchronized
|
||
- **粒度**：Node 级别（更细）
|
||
- **并发度**：理论上无限制（实际受限于数组大小）
|
||
|
||
**核心改进**：
|
||
| 特性 | JDK 1.7 | JDK 1.8 |
|
||
|------|---------|---------|
|
||
| **锁机制** | ReentrantLock（分段） | CAS + synchronized（Node 级别） |
|
||
| **锁粒度** | Segment 级别 | Node 级别（更细） |
|
||
| **并发度** | 默认 16 | 理论无限制 |
|
||
| **查询** | 需要加锁（不强一致） | 无锁（volatile） |
|
||
| **红黑树** | 不支持 | 支持（链表长度 ≥ 8） |
|
||
|
||
---
|
||
|
||
### 2. JDK 1.8 实现原理
|
||
|
||
#### **核心数据结构**
|
||
|
||
```java
|
||
public class ConcurrentHashMap<K, V> {
|
||
// 数组
|
||
transient volatile Node<K,V>[] table;
|
||
|
||
// 数组（扩容时使用）
|
||
private transient volatile Node<K,V>[] nextTable;
|
||
|
||
// 基础计数器值
|
||
private transient volatile long baseCount;
|
||
|
||
// 控制位（sizeCtl < 0：初始化或扩容；-1：正在初始化；<-1：扩容线程数）
|
||
private transient volatile int sizeCtl;
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
#### **Node 节点**
|
||
|
||
```java
|
||
static class Node<K,V> implements Map.Entry<K,V> {
|
||
final int hash;
|
||
final K key;
|
||
volatile V val;
|
||
volatile Node<K,V> next; // 链表下一个节点
|
||
}
|
||
```
|
||
|
||
**注意**：`val` 和 `next` 都是 `volatile`，保证可见性。
|
||
|
||
---
|
||
|
||
#### **TreeNode（红黑树节点）**
|
||
|
||
```java
|
||
static final class TreeNode<K,V> extends Node<K,V> {
|
||
TreeNode<K,V> parent; // 父节点
|
||
TreeNode<K,V> left; // 左子节点
|
||
TreeNode<K,V> right; // 右子节点
|
||
TreeNode<K,V> prev; // 前驱节点
|
||
boolean red; // 颜色（红黑树）
|
||
}
|
||
```
|
||
|
||
**转换条件**：
|
||
- 链表 → 红黑树：链表长度 ≥ 8 **且** 数组长度 ≥ 64
|
||
- 红黑树 → 链表：红黑树节点数 ≤ 6
|
||
|
||
---
|
||
|
||
### 3. 核心 API 源码解析
|
||
|
||
#### **put() 方法**
|
||
|
||
```java
|
||
public V put(K key, V value) {
|
||
return putVal(key, value, false);
|
||
}
|
||
|
||
final V putVal(K key, V value, boolean onlyIfAbsent) {
|
||
if (key == null || value == null) throw new NullPointerException();
|
||
|
||
// 1. 计算哈希值（扰动函数，减少哈希冲突）
|
||
int hash = spread(key.hashCode());
|
||
int binCount = 0;
|
||
|
||
// 2. 无限循环（CAS + 重试）
|
||
for (Node<K,V>[] tab = table;;) {
|
||
Node<K,V> f; int n, i, fh;
|
||
|
||
// 2.1 初始化数组（延迟初始化）
|
||
if (tab == null || (n = tab.length) == 0)
|
||
tab = initTable();
|
||
|
||
// 2.2 计算索引位置
|
||
else if ((f = tabAt(tab, i = (n - 1) & hash)) == null) {
|
||
// 位置为空，CAS 插入
|
||
if (casTabAt(tab, i, null, new Node<K,V>(hash, key, value, null)))
|
||
break; // CAS 成功，退出
|
||
}
|
||
|
||
// 2.3 扩容中（MOVED = -1）
|
||
else if ((fh = f.hash) == MOVED)
|
||
tab = helpTransfer(tab, f); // 帮助扩容
|
||
|
||
// 2.4 位置不为空（链表或红黑树）
|
||
else {
|
||
V oldVal = null;
|
||
synchronized (f) { // 加锁（Node 级别）
|
||
if (tabAt(tab, i) == f) {
|
||
// 链表
|
||
if (fh >= 0) {
|
||
binCount = 1;
|
||
for (Node<K,V> e = f;; ++binCount) {
|
||
K ek;
|
||
// 键已存在，更新值
|
||
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;
|
||
// 到达链表尾部，插入新节点
|
||
if ((e = e.next) == null) {
|
||
pred.next = new Node<K,V>(hash, key,
|
||
value, null);
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
// 红黑树
|
||
else if (f instanceof TreeBin) {
|
||
Node<K,V> p;
|
||
binCount = 2;
|
||
if ((p = ((TreeBin<K,V>)f).putTreeVal(hash, key,
|
||
value)) != null) {
|
||
oldVal = p.val;
|
||
if (!onlyIfAbsent)
|
||
p.val = value;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
// 2.5 链表转红黑树
|
||
if (binCount != 0) {
|
||
if (binCount >= TREEIFY_THRESHOLD)
|
||
treeifyBin(tab, i); // TREEIFY_THRESHOLD = 8
|
||
if (oldVal != null)
|
||
return oldVal;
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
|
||
// 3. 增加元素个数（LongAdder）
|
||
addCount(1L, binCount);
|
||
return null;
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
#### **get() 方法**
|
||
|
||
```java
|
||
public V get(Object key) {
|
||
Node<K,V>[] tab; Node<K,V> e, p; int n, eh; K ek;
|
||
int h = spread(key.hashCode());
|
||
|
||
// 1. 数组不为空且索引位置不为空
|
||
if ((tab = table) != null && (n = tab.length) > 0 &&
|
||
(e = tabAt(tab, (n - 1) & h)) != null) {
|
||
|
||
// 2. 第一个节点匹配
|
||
if ((eh = e.hash) == h) {
|
||
if ((ek = e.key) == key || (ek != null && key.equals(ek)))
|
||
return e.val;
|
||
}
|
||
|
||
// 3. 红黑树
|
||
else if (eh < 0)
|
||
return (p = e.find(h, key)) != null ? p.val : null;
|
||
|
||
// 4. 链表遍历
|
||
while ((e = e.next) != null) {
|
||
if (e.hash == h &&
|
||
((ek = e.key) == key || (ek != null && key.equals(ek))))
|
||
return e.val;
|
||
}
|
||
}
|
||
|
||
return null;
|
||
}
|
||
```
|
||
|
||
**注意**：
|
||
- **无锁读取**：整个 `get()` 方法无锁
|
||
- **volatile 可见性**：`Node.val` 是 `volatile`，保证读到最新值
|
||
|
||
---
|
||
|
||
#### **size() 方法**
|
||
|
||
**问题**：如何高效统计元素个数？
|
||
|
||
**方案**：**LongAdder**（分段计数）
|
||
|
||
```java
|
||
public int size() {
|
||
long n = sumCount();
|
||
return ((n < 0L) ? 0 :
|
||
(n > (long)Integer.MAX_VALUE) ? Integer.MAX_VALUE :
|
||
(int)n);
|
||
}
|
||
|
||
final long sumCount() {
|
||
// 计算所有 CounterCell 的和
|
||
CounterCell[] as = counterCells;
|
||
long sum = baseCount;
|
||
if (as != null) {
|
||
for (CounterCell a : as)
|
||
if (a != null)
|
||
sum += a.value;
|
||
}
|
||
return sum;
|
||
}
|
||
```
|
||
|
||
**LongAdder 原理**：
|
||
```
|
||
LongAdder
|
||
↓
|
||
baseCount（基础值）
|
||
↓
|
||
CounterCell[]（计数器数组，避免 CAS 竞争）
|
||
↓
|
||
多线程更新时，随机选择一个 CounterCell 更新
|
||
↓
|
||
统计时，baseCount + 所有 CounterCell 的值
|
||
```
|
||
|
||
**为什么高效？**
|
||
- 高并发时，多线程更新不同的 CounterCell（无竞争）
|
||
- 统计时才累加（牺牲实时性换取性能）
|
||
|
||
---
|
||
|
||
### 4. ConcurrentHashMap vs 其他 Map
|
||
|
||
#### **对比表**
|
||
|
||
| 特性 | HashMap | Hashtable | Collections.synchronizedMap | ConcurrentHashMap |
|
||
|------|---------|-----------|----------------------------|-------------------|
|
||
| **线程安全** | ❌ 否 | ✅ 是 | ✅ 是 | ✅ 是 |
|
||
| **锁机制** | 无 | synchronized | synchronized | CAS + synchronized |
|
||
| **锁粒度** | - | 整个表 | 整个表 | Node 级别 |
|
||
| **并发度** | 无 | 低 | 低 | 高 |
|
||
| **迭代器** | Fail-Fast | Fail-Safe | Fail-Safe | Fail-Safe |
|
||
| **null 键值** | 允许 | 不允许 | 不允许 | 不允许 |
|
||
|
||
---
|
||
|
||
#### **详细对比**
|
||
|
||
**1. Hashtable（不推荐）**
|
||
|
||
```java
|
||
// Hashtable 的 put 方法（整个表加锁）
|
||
public synchronized V put(K key, V value) {
|
||
// ...
|
||
}
|
||
```
|
||
|
||
**问题**：
|
||
- **锁粒度太大**：整个表加锁
|
||
- **并发度低**：同一时刻只有一个线程能操作
|
||
|
||
---
|
||
|
||
**2. Collections.synchronizedMap()**
|
||
|
||
```java
|
||
Map<String, String> map = Collections.synchronizedMap(new HashMap<>());
|
||
```
|
||
|
||
**原理**：
|
||
```java
|
||
// 内部使用 mutex（Object）加锁
|
||
public V put(K key, V value) {
|
||
synchronized (mutex) { // 整个 Map 加锁
|
||
return m.put(key, value);
|
||
}
|
||
}
|
||
```
|
||
|
||
**问题**：
|
||
- **锁粒度太大**：整个 Map 加锁
|
||
- **迭代器需要手动加锁**：
|
||
```java
|
||
Map<String, String> map = Collections.synchronizedMap(new HashMap<>());
|
||
|
||
// 遍历需要手动加锁
|
||
synchronized (map) {
|
||
for (Map.Entry<String, String> entry : map.entrySet()) {
|
||
// ...
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
**3. ConcurrentHashMap（推荐）**
|
||
|
||
```java
|
||
Map<String, String> map = new ConcurrentHashMap<>();
|
||
```
|
||
|
||
**优点**：
|
||
- **锁粒度小**：Node 级别
|
||
- **并发度高**：理论上无限制
|
||
- **迭代器无需加锁**：Fail-Safe（弱一致迭代器）
|
||
|
||
---
|
||
|
||
### 5. 实际项目应用
|
||
|
||
#### **场景 1：本地缓存**
|
||
|
||
```java
|
||
@Component
|
||
public class LocalCache {
|
||
private final ConcurrentHashMap<String, Object> cache = new ConcurrentHashMap<>();
|
||
|
||
public void put(String key, Object value) {
|
||
cache.put(key, value);
|
||
}
|
||
|
||
public Object get(String key) {
|
||
return cache.get(key);
|
||
}
|
||
|
||
public void remove(String key) {
|
||
cache.remove(key);
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
#### **场景 2：计数器**
|
||
|
||
```java
|
||
@Component
|
||
public class CounterService {
|
||
private final ConcurrentHashMap<String, LongAdder> counters = new ConcurrentHashMap<>();
|
||
|
||
public void increment(String key) {
|
||
counters.computeIfAbsent(key, k -> new LongAdder()).increment();
|
||
}
|
||
|
||
public long get(String key) {
|
||
LongAdder adder = counters.get(key);
|
||
return adder != null ? adder.sum() : 0;
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
#### **场景 3：去重表**
|
||
|
||
```java
|
||
@Component
|
||
public class DeduplicationService {
|
||
private final ConcurrentHashMap<String, Boolean> dedupTable = new ConcurrentHashMap<>();
|
||
|
||
public boolean isDuplicate(String id) {
|
||
return dedupTable.putIfAbsent(id, true) != null;
|
||
}
|
||
}
|
||
```
|
||
|
||
---
|
||
|
||
### 6. 阿里 P7 加分项
|
||
|
||
**深度理解**：
|
||
- 理解 `ConcurrentHashMap` 的扩容机制（多线程协同扩容）
|
||
- 理解 `LongAdder` 的实现原理（Cell 数组 + CAS）
|
||
- 理解 `ConcurrentHashMap` 的弱一致迭代器
|
||
|
||
**实战经验**：
|
||
- 有使用 `ConcurrentHashMap` 解决并发问题的经验
|
||
- 有 `ConcurrentHashMap` 性能调优的经验
|
||
- 有处理 `ConcurrentHashMap` 相关线上问题的经验
|
||
|
||
**架构能力**：
|
||
- 能根据业务特点选择合适的 Map 实现
|
||
- 能设计高性能的并发数据结构
|
||
- 有分布式缓存的设计经验
|
||
|
||
**技术选型**：
|
||
- 了解 `ConcurrentSkipListMap`（跳表实现）
|
||
- 了解 Guava 的 `LocalCache`（Caffeine）
|
||
- 能根据场景选择本地缓存或分布式缓存
|