java并发编程(3):ThreadLocal

转载:http://www.cnblogs.com/dolphin0520/p/3920407.html

一. 对ThreadLocal的理解

ThreadLocal,很多地方叫做线程本地变量,也有地方叫做线程本地存储,其实意思差不多。可能很多朋友都知道ThreadLocal为变量在每个线程中都创建了一个副本,那么每个线程可以访问自己内部的副本变量。

这句话从字面上看起来看容易理解,但真正理解并不是那么容易。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
/**
 * 学习ThreadLocal
 * Created by mrf on 2016/3/6.
 */
public class ConnectionManager {
    private static Connection connection = null;
    public static Connection openConnection() throws SQLException {
        if(connection == null){
            connection = DriverManager.getConnection("url","username","password");
        }
        return connection;
    }
 
    public static void closeConnection() throws SQLException {
        if (connection!=null)
            connection.close();
    }
}

假设有这样一个数据库连接管理类,这段代码在单线程中使用没有任何问题的,但是如果在多线程就会存在线程安全问题:

  • 这里面的2个方法没有同步,很可能在openConnection方法中多次创建connect
  • 由于connect是共享变量,那么必然在调用connect的地方需要使用到同步来保证线程安全,因为很可能一个线程在使用connect进行数据库操作,而另外一个线程closeConnection关闭链接

所以出于线程安全的考虑,必须将这段代码的两个方法进行同步处理,并且在调用connect的地方需要进行同步处理。这样会大大影响程序的执行效率,因为一个线程在使用connect进行数据库操作的时候,其他线程只有等待。

那么,大家来自习分析一下这个问题,这地方到底需不需要讲connect变量进行共享?事实上,是不需要的。假如每个线程中都有一个connect变量,各个线程之间对connect变量的访问实际上是没有依赖关系的,即一个线程不需要关心其他线程是否对这个connect进行了修改的。那么直接在使用数据库连接的方法中具体使用时才创建数据库,然后在方法调用完毕再释放这个链接,比如:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
class ConnectionManager2{
    private Connection connection = null;
    public Connection openConnection() throws SQLException {
        if (connection==null){
            connection = DriverManager.getConnection("url","username","password");
        }
        return connection;
    }
 
    public void closeConnection() throws SQLException {
        if (connection!=null)
            connection.close();
    }
}
class Dao{
    public void insert() throws SQLException {
        ConnectionManager2 connectionManager2 = new ConnectionManager2();
        Connection connection = connectionManager2.openConnection();
        //使用connection进行操作
 
        connectionManager2.closeConnection();
    }
}

这样处理确实也没有任何问题,由于每次都是在方法内部创建的连接,那么线程之间自然不存在线程安全问题。但是这样会有一个致命的影响:导致服务器压力非常大,并且严重影响程序执行性能。由于在方法中需要频繁地开启和关闭数据库连接,这样不仅严重影响程序执行效率,还可能导致服务器压力巨大。

那么这种情况下使用ThreadLocal是再合适不过了,因为ThreadLocal在每个线程中对该变量会创建一个副本,即每个线程内部都会有一个该变量,且在线程内部任何地方都可以使用,线程之间互不影响,这样一来就不存在线程安全问题,也不会严重影响程序执行性能。

但是要注意,虽然ThreadLocal能够解决上面说的问题,但是由于在每个线程中都创建了副本,所以要考虑它对资源的消耗,比如内存的占用会比不是用ThreadLocal要大。

二. 深入解析ThreadLocal类

在上面谈到了对ThreadLocal的一些理解,那我们下面来看一下具体ThreadLocal是如何实现的。

先了解下ThreadLocal类提供的几个方法:

1
2
3
4
public T get() { }
public void set(T value) { }
public void remove() { }
protected T initialValue() { }

get()方法是用来获取ThreadLocal在当前线程中保存的变量副本,set()用来设置当前线程中变量的副本,remove()用来移除当前线程中变量的副本,initialValue()是一个protected方法,一般用来在使用时进行重写的,它是一个延迟加载方法。

首先我们来看一下ThreadLocal类时如何为每个线程创建一个变量的副本的。

先看get方法的实现:

第一句是取得当前线程,然后通过getMap(t)方法获取一个map,map的类型为ThreadLocalMap.然后下面获取到key-value,注意这里获取键值对传进去的是this,而不是当前线程t

如果获取成功,则返回value值。如果map为空,则调用setInitialValue方法返回value。我们上的每一句来自习分析:

首先看一下getMap方法中做了什么:

可能大家没有想到的是,在getMap中,是调用当前线程t,返回当前线程t中一个成员变量threadLocals.那么继续Thread类中看看成员变量ThreadLocals是什么:

1
2
3
/* ThreadLocal values pertaining to this thread. This map is maintained
     * by the ThreadLocal class. */
    ThreadLocal.ThreadLocalMap threadLocals = null;

实际上就是一个ThreadLocalMap,这个类型是ThreadLocal类的一个内部类,我们继续看ThreadLocalMap的实现:

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
/**
 * ThreadLocalMap is a customized hash map suitable only for
 * maintaining thread local values. No operations are exported
 * outside of the ThreadLocal class. The class is package private to
 * allow declaration of fields in class Thread.  To help deal with
 * very large and long-lived usages, the hash table entries use
 * WeakReferences for keys. However, since reference queues are not
 * used, stale entries are guaranteed to be removed only when
 * the table starts running out of space.
 */
static class ThreadLocalMap {
 
    /**
     * The entries in this hash map extend WeakReference, using
     * its main ref field as the key (which is always a
     * ThreadLocal object).  Note that null keys (i.e. entry.get()
     * == null) mean that the key is no longer referenced, so the
     * entry can be expunged from table.  Such entries are referred to
     * as "stale entries" in the code that follows.
     */
    static class Entry extends WeakReference<ThreadLocal> {
        /** The value associated with this ThreadLocal. */
        Object value;
 
        Entry(ThreadLocal k, Object v) {
            super(k);
            value = v;
        }
    }

可以看到ThreadLocalMap的Entry继承了WeakReference作为键值。然后继续看setInitialValue方法的具体实现:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
/**
 * Variant of set() to establish initialValue. Used instead
 * of set() in case user has overridden the set() method.
 *
 * @return the initial value
 */
private T setInitialValue() {
    T value = initialValue();
    Thread t = Thread.currentThread();
    ThreadLocalMap map = getMap(t);
    if (map != null)
        map.set(this, value);
    else
        createMap(t, value);
    return value;
}

很容易了解,就是如果map不为空,就设置键值对,为空则创建map,看下createMap的实现:

1
2
3
4
5
6
7
8
9
10
11
/**
    * Create the map associated with a ThreadLocal. Overridden in
    * InheritableThreadLocal.
    *
    * @param t the current thread
    * @param firstValue value for the initial entry of the map
    * @param map the map to store.
    */
   void createMap(Thread t, T firstValue) {
       t.threadLocals = new ThreadLocalMap(this, firstValue);
   }

至此,可能大部分朋友已经明白了ThreadLocal是如何为每个线程创建变量副本的:

首先,在每个线程Thread内部又一个ThreadLocal.ThreadLocalMap类型的成员变量threadLocals,这个threaLocals就是用来存储实际的变量副本的,键值为当前ThreadLocal变量,value为变量副本(即T类型的变量)。

初始时,在Thread里面,ThreadLocals为空,当通过ThreadLocal变量调用get()方法或者set()方法,就会对Thread类中的threadLocals进行初始化,并且以当前ThreadLocal变量为键值,以ThreadLocal要保存的副本变量为value,存到threadLocals.

然后在当前线程里,如果要使用副本变量,就可以通过get方法在threadLocals里面查找。

下面通过一个例子来证明通过ThreadLocal能达到在每个线程中创建变量副本的效果:

 

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
/**
 * 测试ThreadLocal变量在不同线程中变量的值是不同的。
 * 因为每个线程都有个副本,然后可以针对副本进行修改而不影响其他线程中变量的副本值
 */
class TestThreadLocal2{
    ThreadLocal<Long> longLocal = new ThreadLocal<>();
    ThreadLocal<String> stringLocal = new ThreadLocal<>();
    public void set(){
        longLocal.set(Thread.currentThread().getId());
        stringLocal.set(Thread.currentThread().getName());
    }
 
    public long getLong(){
        return longLocal.get();
    }
 
    public String getString(){
        return stringLocal.get();
    }
 
    public static void main(String[] args) throws InterruptedException {
        final TestThreadLocal2 test = new TestThreadLocal2();
        test.set();
        System.out.println("===========main线程==============");
        System.out.println(test.getLong());
        System.out.println(test.getString());
 
        Thread t1 = new Thread(){
            public void run(){
                test.set();
                System.out.println("===========t线程==============");
                System.out.println(test.getLong());
                System.out.println(test.getString());
            }
        };
        t1.start();
        t1.join();
        System.out.println("===========main线程==============");
        System.out.println(test.getLong());
        System.out.println(test.getString());
    }
}

这段代码的输出结果为:

从这段代码的输出结果可以看出,在main线程中和thread0线程中,longLocal保存的副本值和stringLocal保存的副本值都不一样。最后一次在main线程再次打印副本值是为了证明main线程中和thread0线程中的副本值确实是不同的。

总结一下:

  1. 实际的通过ThreadLocal创建的副本是存储在每个线程自己的threadLocals中的;
  2. 为何threadLocals的类型为ThreadLocalMap的键值为ThreadLocal对象?因为每个线程中个可以有多个ThreadLocal变量,就像上面的long和string;
  3. 在进行get之前必须先set,否则空指针异常;如果不想set则必须重写initialValue()方法。因为在上面的代码分析过程中,我们发现如果没有先set的话,即在map中查找不到对应的存储,则会通过调用setInitialValue方法返回i,而在setInitialValue方法中个,有一个语句是T value = initailValue(),而默认情况下,initialValue方法返回的是null。

因此,如果不set则要设置initialValue方法:

三.ThreadLocal的应用场景

最常见的ThreadLocal使用场景为:数据库连接,Session管理等。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
/**
 * Created by mrf on 2016/3/4.
 */
public class TestThreadLocal {
    private static ThreadLocal<Connection> connectionHolder = new ThreadLocal<Connection>(){
        public Connection initialValue(){
            try {
                return DriverManager.getConnection("url","username","password");
            } catch (SQLException e) {
                e.printStackTrace();
            }
            return null;
        }
    };
 
    public static Connection getConnection(){
        return connectionHolder.get();
    }
}

  

下面这段代码摘自:

  http://www.iteye.com/topic/103804

1
2
3
4
5
6
7
8
9
10
11
12
13
14
private static final ThreadLocal threadSession = new ThreadLocal();
  
public static Session getSession() throws InfrastructureException {
    Session s = (Session) threadSession.get();
    try {
        if (s == null) {
            s = getSessionFactory().openSession();
            threadSession.set(s);
        }
    } catch (HibernateException ex) {
        throw new InfrastructureException(ex);
    }
    return s;
}

  

posted @   Ryan.Miao  阅读(359)  评论(0编辑  收藏  举报
编辑推荐:
· Linux系列:如何用heaptrack跟踪.NET程序的非托管内存泄露
· 开发者必知的日志记录最佳实践
· SQL Server 2025 AI相关能力初探
· Linux系列:如何用 C#调用 C方法造成内存泄露
· AI与.NET技术实操系列(二):开始使用ML.NET
阅读排行:
· 无需6万激活码!GitHub神秘组织3小时极速复刻Manus,手把手教你使用OpenManus搭建本
· C#/.NET/.NET Core优秀项目和框架2025年2月简报
· Manus爆火,是硬核还是营销?
· 终于写完轮子一部分:tcp代理 了,记录一下
· 【杭电多校比赛记录】2025“钉耙编程”中国大学生算法设计春季联赛(1)
点击右上角即可分享
微信分享提示