4种Java引用浅解

最新推荐文章于 2021-08-03 15:28:54 发布

时光荏苒心依旧

最新推荐文章于 2021-08-03 15:28:54 发布

阅读量253

点赞数

分类专栏： java虚拟机文章标签： java四种引用 jvm

java虚拟机专栏收录该内容

7 篇文章 0 订阅

订阅专栏

最近研究Java Cache实现，发现使用到了软引用(SoftReference)，不太理解，查阅了JDK文档、代码以及几篇文章。做个小结，如有错误，欢迎指正。

之所以想学习一下Java的几种引用类型，原因有两个:

理解Java Cache实现、学习Java引用与Java垃圾回收机制的关系
内存资源是有限的，需要合理的利用。Cache不是仅仅HashMap那么简单，Java引用与Java垃圾回收机制也有非常紧密的关系。
避免对Java引用的错误使用
某个同事把5000+交易数据放到一个HashMap里面，用一个Spring Singleton Bean的全局属性指向该HashMap。大量运用这种技术，很快就报out of memory。再大的内存也架不住对内存的错误使用。理解原理有助于我们尽量少犯或不犯低级错误。

Java引用与Java垃圾回收机制的关系

当Java虚拟机(JVM)觉得内存不够用的时候，会触发垃圾回收操作(GC)，清除无用的对象，释放内存。可是如何判断一个对象是否是垃圾呢？其中的一个方法是计算指向该对象的引用数量，如果引用数量为0，那么该对象就为垃圾(Thread对象是例外)，否则还有用处，不能被回收。但是如果把引用数为0的对象都回收了，还是不能满足内存需求怎么办？Java把引用分为4种类型，垃圾回收器会尝试回收只有弱引用的对象。

按照一个对象的引用可达(Reachable)强度，由强到弱分为5类，如下:

强可达(Strong Reachable)
在一个线程内，无需引用直接可达，新创建的对象是强可达的。
软可达(Soft Reachable)
不是强可达的，但是通过一个软引用(SoftReference)可达。
弱可达(Soft Reachable)
既不是强可达也不是软可达，但是通过一个弱引用(WeakReference)可达。
虚可达(Phantom Reachable)
既不是强可达，不是软可达，也不是弱可达，但是通过一个虚引用(PhantomReference)可达。
不可达(Unreachable)
没有任何引用指向对象。

比较好、容易理解的是Java垃圾回收器会优先清理可达强度低的对象。另外有两个重要的点:

强可达的一定不会被清理
JVM保证抛出out of memory之前，清理所有的软引用对象

4种Java引用

在实现一个缓存系统的时候，如果全部使用强引用，那么你需要自己去手动的把某些引用clear掉(引用置位null)，否则迟早会抛出out of memory错误。缓存系统引入弱引用或者软引用的唯一原因是，把引用clear的事情交由Java垃圾回收器来处理，cache程序自己置身事外。

几种弱引用的使用方式非常相近。下面分别介绍4种引用类型。

强引用(StrongReference)

我们平时申明变量使用的就是强引用，普通系统99%以上都是强引用。比如，String s = "Hello World"

弱引用(WeakReference)

垃圾回收器某个时刻决定回收软可达的对象的时候，会清理软引用，并可选的把引用存放到一个引用队列(ReferenceQueue)。

软引用(SoftReference)

类似弱引用，只不过Java虚拟机会尽量让软引用的存活时间长一些，迫不得已才清理。

虚引用(PhantomReference)

仅用来处理资源的清理问题，比Object里面的finalize机制更灵活。get方法返回的永远是null，Java虚拟机不负责清理虚引用，但是它会把虚引用放到引用队列里面。

两个测试例子

使用HashMap，会报out of memory错误。

[java] view plain copy

print ?

public static void main(String[] args) {
Map<String, String> list = new HashMap<String, String>();
long i = 1;
while (i < 100000000L) {
list.put(
String.valueOf(i),
"JDJJDJJJJJJJJJJ%%%%%%%%JJJJJJJJJJJJJJJKKKKKKKKKKKKKKKKKJJJJJJ"
+ "JJJKKKKKHDDDJDJDJDJDJDJDJDJJDJDJDJDJDJDJJDJDJDJDJJDJDJJJJJJJJJ"
+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ"
+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJ");
// 测试第一个是否依然存活
if (i % 100000 == 0) {
System.out.println(list.get(String.valueOf(1)));
}
i++;
}
}

public static void main(String[] args) {
		Map<String, String> list = new HashMap<String, String>();
		long i = 1;
		while (i < 100000000L) {
			list.put(
					String.valueOf(i),
					"JDJJDJJJJJJJJJJ%%%%%%%%JJJJJJJJJJJJJJJKKKKKKKKKKKKKKKKKJJJJJJ"
							+ "JJJKKKKKHDDDJDJDJDJDJDJDJDJJDJDJDJDJDJDJJDJDJDJDJJDJDJJJJJJJJJ"
							+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ"
							+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJ");

			// 测试第一个是否依然存活
			if (i % 100000 == 0) {
				System.out.println(list.get(String.valueOf(1)));
			}

			i++;
		}
	}

使用WeakHashMap，不会报out of memory错误。

[java] view plain copy

print ?

public static void main(String[] args) {
Map<String, String> list = new WeakHashMap<String, String>();
long i = 1;
while (i < 100000000L) {
list.put(
String.valueOf(i),
"JDJJDJJJJJJJJJJ%%%%%%%%JJJJJJJJJJJJJJJKKKKKKKKKKKKKKKKKJJJJJJ"
+ "JJJKKKKKHDDDJDJDJDJDJDJDJDJJDJDJDJDJDJDJJDJDJDJDJJDJDJJJJJJJJJ"
+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ"
+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJ");
// 测试第一个是否依然存活
if (i % 100000 == 0) {
System.out.println(list.get(String.valueOf(1)));
}
i++;
}
}

public static void main(String[] args) {
		Map<String, String> list = new WeakHashMap<String, String>();
		long i = 1;
		while (i < 100000000L) {
			list.put(
					String.valueOf(i),
					"JDJJDJJJJJJJJJJ%%%%%%%%JJJJJJJJJJJJJJJKKKKKKKKKKKKKKKKKJJJJJJ"
							+ "JJJKKKKKHDDDJDJDJDJDJDJDJDJJDJDJDJDJDJDJJDJDJDJDJJDJDJJJJJJJJJ"
							+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJ"
							+ "JJJJJJJJJJJJJJJJJJJJJJJJJJJJ");

			// 测试第一个是否依然存活
			if (i % 100000 == 0) {
				System.out.println(list.get(String.valueOf(1)));
			}

			i++;
		}
	}