JVM内存分析工具——HProf

HProf是JDK自带一个简单的性能分析工具。它是一个动态链接库文件，监控CPU的1使用率、内存堆栈分配情况等。使用命令行格式为：

java -Xrunhprof ToBeProfiledClass

HProf常被用于内存使用情况分析。有两种分析方法：内存分配历史的跟踪记录（dump）和将占用内存的对象进行排序（sites）。命令行如下

java -Xrunhprof:heap=all|dump|sites ToBeProfiledClass

构造如下代码：

import java.lang.ref.SoftReference;
import java.lang.ref.WeakReference;
import java.lang.ref.PhantomReference;
import java.lang.ref.ReferenceQueue;
import java.util.Set;
import java.util.HashSet;

public class TestReferences
{
public static void main(String[] args)
{
int length=10;

//创建length个MyObject对象的强引用
Set<MyObject> a = new HashSet<MyObject>();
for(int i = 0; i < length; i++)
{
MyObject ref=new MyObject("Hard_" + i);
System.out.println("创建强引用：" +ref);
a.add(ref);
}
//a=null;
System.gc();

//创建length个MyObject对象的软引用
Set<SoftReference<MyObject>> sa = new HashSet<SoftReference<MyObject>>();
for(int i = 0; i < length; i++)
{
SoftReference<MyObject> ref=new SoftReference<MyObject>(new MyObject("Soft_" + i));
System.out.println("创建软引用：" +ref.get());
sa.add(ref);
}
System.gc();

//创建length个MyObject对象的弱引用
Set<WeakReference<MyObject>> wa = new HashSet<WeakReference<MyObject>>();
for(int i = 0; i < length; i++)
{
WeakReference<MyObject> ref=new WeakReference<MyObject>(new MyObject("Weak_" + i));
System.out.println("创建弱引用：" +ref.get());
wa.add(ref);
}
System.gc();

//创建length个MyObject对象的虚引用
ReferenceQueue<MyObject> rq = new ReferenceQueue<MyObject>();
Set<PhantomReference<MyObject>> pa = new HashSet<PhantomReference<MyObject>>();
for(int i = 0; i < length; i++)
{
PhantomReference<MyObject> ref = new PhantomReference<MyObject>(new MyObject("Phantom_" + i), rq);
System.out.println("创建虚引用：" +ref.get());
pa.add(ref);
}
System.gc();
}
}

将应用程序中创建的对象按照内存大小进行排序

java -Xrunhprof:heap=sites TestReferences

运行的截图如下



执行之后将在当前目录下生成java.hprof.txt

摘录如下

SITES BEGIN (ordered by live bytes) Wed Dec 20 00:53:56 2017

percent live alloc’ed stack class

rank self accum bytes objs bytes objs trace name

1 1.25% 1.25% 3952 46 3952 46 300078 char[]

2 0.78% 2.03% 2464 77 2464 77 300080 java.util.HashMapNode30.6340.6250.6060.6070.57Entry

8 0.41% 5.46% 1296 3 1296 3 300000 java.lang.Thread

9 0.41% 5.86% 1280 32 1880 47 300133 java.lang.ref.Finalizer

10 0.35% 6.21% 1104 46 1104 46 300075 java.lang.String

11 0.35% 6.56% 1104 46 1104 46 300077 java.lang.String

12 0.33% 6.89% 1040 1 1040 1 300020 java.lang.invoke.MethodHandle[]

…………………………………

639 0.01% 99.52% 32 1 32 1 300281 java.lang.ref.WeakReference

640 0.01% 99.53% 32 1 32 1 300215 java.lang.ThreadLocalThreadLocalMapEntry

SITES END