Android内存泄露原因分析与处理方案

Myth丶恋晨 2022-07-28 01:15 336阅读 0赞

**内存泄露？**  
泄露并不是真正意思的泄露，而是因为内存不能被GC，从而导致占用内存过大，发生Out of Memory，而被系统Kill。  
adb shell getprop | grep dalvik：

\[dalvik.vm.heapsize\]: \[64m\]

内存泄露-人怕出名猪怕壮

LMK（LowMemoryKiller）会周期性的运行，工作时，首先根据下面的值确定当前的警戒级数，高于警戒级数的进程是待杀的范围。若找到多个，则把占用进程最大的进程，发送SIGKILL，杀掉该进程。

而Android也会在E:\\Src\\IceCreamSandwich\\frameworks\\base\\services\\java\\com\\android\\server\\am\\ActivityManagerService.java中去动态的调整每一个app的Level。

/system/core/rootdir/init.rc  
\# Define the oom\_adj values for the classes of processes that can be killed by the kernel.  
    setprop ro.FOREGROUND\_APP\_ADJ 0  
    setprop ro.VISIBLE\_APP\_ADJ 1  
    setprop ro.PERCEPTIBLE\_APP\_ADJ 2  
    setprop ro.HEAVY\_WEIGHT\_APP\_ADJ 3  
    setprop ro.SECONDARY\_SERVER\_ADJ 4  
    setprop ro.BACKUP\_APP\_ADJ 5  
    setprop ro.HOME\_APP\_ADJ 6  
    setprop ro.HIDDEN\_APP\_MIN\_ADJ 7  
    setprop ro.EMPTY\_APP\_ADJ 15  
\# Define the memory thresholds at which the above process classes will be killed. in pages (4k).  
    setprop ro.FOREGROUND\_APP\_MEM 2048  
    setprop ro.VISIBLE\_APP\_MEM 3072  
    setprop ro.PERCEPTIBLE\_APP\_MEM 4096  
    setprop ro.HEAVY\_WEIGHT\_APP\_MEM 4096  
    setprop ro.SECONDARY\_SERVER\_MEM 6144  
    setprop ro.BACKUP\_APP\_MEM 6144  
    setprop ro.HOME\_APP\_MEM 6144  
    setprop ro.HIDDEN\_APP\_MEM 7168  
    setprop ro.EMPTY\_APP\_MEM 8192  
=======================================================  
  
  
**Android中如何查看内存**

**（一）DDMS 的Heap Dump**

![0_1325909160ggLo.gif][]

1) Data Object：java object.

2) Class Object：object of type Class, e.g. what you'd get from java.lang.String.class or myObject.getClass( ).  
3) 1,2,4,8-byte array：Number of bytes per entry.   
   1-byte array: byte, boolean  
   2-byte array: char, short  
   4-byte array: float, int  
   8-byte array: double, long

4) non-Java object：A non-Java Object is a piece of memory that isn't actually accessible from code written in Java. Essentially it's a blob of stuff that got stuck on the virtual heap but has no meaning to interpreted code. Shouldn't be much of that.

**（二）如果你想查看所有进程的内存使用情况，可以使用"adb shell procrank"命令。**

![0_132590919785Zf.gif][]

*  **VSS** - Virtual Set Size 虚拟耗用内存（包含共享库占用的内存）
 *  **RSS** - Resident Set Size 实际使用物理内存（包含共享库占用的内存）
 *  **PSS** - Proportional Set Size 实际使用的物理内存（比例分配共享库占用的内存）
 *  **USS** - Unique Set Size 进程独自占用的物理内存（不包含共享库占用的内存）

**VSS** (reported as VSZ from ps) is the total accessible address space of a process. This size also includes

memory that may not be resident in RAM like mallocs that have been allocated but not written to.

**VSS is of very little use for determing real memory usage of a process.  
  
RSS **is the total memory actually held in RAM for a process. RSS can be misleading,

because it reports the total all of the shared libraries that the process uses,

even though a shared library is only loaded into memory once regardless of how many processes use it.

**RSS is not an accurate representation of the memory usage for a single process.  
  
PSS **differs from RSS in that it reports the proportional size of its shared libraries,

i.e. if three processes all use a shared library that has 30 pages,

that library will only contribute 10 pages to the PSS that is reported for each of the three processes.

PSS is a very useful number because when the PSS for all processes in the system are summed together,

that is a good representation for the total memory usage in the system.

When a process is killed, the shared libraries that contributed to its PSS will be proportionally distributed to

the PSS totals for the remaining processes still using that library. In this way PSS can be slightly misleading,

because when a process is killed, PSS does not accurately represent the memory returned to the overall system.  
  
**USS **is the total private memory for a process, i.e. that memory that is completely unique to that process.

USS is an extremely useful number because it indicates the true incremental cost of running a particular process.

When a process is killed, the USS is the total memory that is actually returned to the system.

USS is the best number to watch when initially suspicious of memory leaks in a process .

**（三）如果只想查看单个进程，可以使用“adb shell****dumpsys meminfo”，该命令后面要加上进程的名字，以确定是哪个进程。**

![0_13259092207fSP.gif][]

**dalvikPrivateDirty**：The private dirty pages used by dalvik。

**dalvikPss**  ： The proportional set size for dalvik.   
**dalvikSharedDirty ** ： The shared dirty pages used by dalvik.   
**nativePrivateDirty** ： The private dirty pages used by the  native heap .   
**nativePss** ： The proportional set size for the native heap.   
**nativeSharedDirty**  ： The shared dirty pages used by the  native heap.   
**otherPrivateDirty** ： The private dirty pages used by everything else.   
**otherPss**  ： The proportional set size for everything else.   
**otherSharedDirty**  ： The shared dirty pages used by everything else.

PrivateDirty, which is basically the amount of RAM inside the process that can not be paged to disk (it is not backed by the same data on disk),  
and is not shared with any other processes.  
Another way to look at PrivateDirty is the RAM that will become available to the system  
when that process goes away (and probably quickly subsumed into caches and other uses of it).  
In one word，Priv Dirty(Total) is Uss。

Interesting thing to note here: Pss and Uss are slightly (or more than slightly) different than what we saw in meminfo. Why is that? Well procrank uses a different kernel mechanism to collect its data than meminfo does, and they give slightly different results. Why is that? Honestly I haven't a clue.

**（四）使用"adb shell cat /proc/meminfo" 命令，该方式只能得出系统整个内存的大概使用情况。**

![0_1325909241mzHF.gif][]

**（五）MAT**

Eclipse->Help->Install New Softwares

http://download.eclipse.org/mat/1.1/update-site/

If your ADT is not gelivable, then you can get the .hprof file and use hprof-conv to convert it.

The hprof-conv tool converts the HPROF file that is generated by theAndroid SDK tools to a standard format.

hprof-conv is under android sdk tools folder.  
**   hprof-conv   src.hprof   dst.hprof**

**Shallow Size and Retained Size**

Shallow size就是对象本身占用内存的大小，不包含对其他对象的引用  
  
Retained size是该对象自己的shallow size，加上只能从该对象直接或间接访问到对象的shallow size之和。换句话说，retained size是该对象被GC之后所能回收到内存的总和。

**Shallow size** of an object is the amount of memory allocated to store the object itself, not taking into account the referenced objects. Shallow size of a regular (non-array) object depends on the number and types of its fields. Shallow size of an array depends on the array length and the type of its elements (objects, primitive types). Shallow size of a set of objects represents the sum of shallow sizes of all objects in the set.

**Retained size** of an object is its shallow size plus the shallow sizes of the objects that are accessible, directly or indirectly,**only** from this object. In other words, the retained size represents the amount of memory that will be freed by the garbage collector when this object is collected.

To better understand the notion of the **retained size**, let us look at the following examples:

In order to measure the retained sizes, all objects in memory are treated as nodes of a graph where its edges represent references from objects to objects. There are also special nodes -*GC root* objects, which will not be collected by*Garbage Collector* at the time of measuring (read more about[ GC roots][GC roots]).

The pictures below show the same set of objects, but with varying internal references.

<table style="color:rgb(51,51,51); font-family:Arial; font-size:13.913043975830078px; line-height:25.99184799194336px"> 
 <tbody> 
  <tr> 
   <td align="left"><strong>Figure 1:</strong></td> 
   <td align="left"><strong>Figure 2:</strong></td> 
  </tr> 
  <tr> 
   <td><img alt="" src="http://www.yourkit.com/docs/90/help/retained_objects.gif" style="border:none; max-width:100%"></td> 
   <td><img alt="" src="http://www.yourkit.com/docs/90/help/retained_objects_2.gif" style="border:none; max-width:100%"></td> 
  </tr> 
 </tbody> 
</table>

Let us consider *obj1*.  
As you can see, in both pictures we have highlighted all of the objects that are directly or indirectly accessed** only** by*obj1*. If you look at Figure 1, you will see that* obj3* is not highlighted, because it is also referenced by a* GC root* object. On Figure 2, however, it is already included into the retained set, unlike* obj5*, which is still referenced by* GC root*.

Thus, the retained size of *obj1* will represent the following respective values:

*  For Figure 1: the sum of shallow sizes of *obj1*, *obj2* and *obj4*
 *  For Figure 2: the sum of shallow sizes of *obj1*, *obj2*, *obj3* and*obj4*

Looking at *obj2*, however, we see that its retained size in the above cases will be:

*  For Figure 1: the sum of shallow sizes of *obj2* and *obj4*
 *  For Figure 2: the sum of shallow sizes of *obj2*, *obj3* and *obj4*

In general, retained size is an integral measure, which helps to understand the structure (clustering) of memory and the dependencies between object sub-graphs, as well as find potential roots of those sub-graphs.

**Dominator Tree**

An object A dominates on an object B if all the paths toobject B pass through object A.

Using the "dominates" relationship we can create a dominatortree out of the graphof objects in memory.

![0_13259092561igI.gif][]

**（六）使用“adb shell ps x”命令，该方式主要得到的是内存信息是VSIZE 和RSS。由于RSS的价值不是很大，所以一般不用。**

**Android内存泄露的原因**

**（一）释放对象的引用，误将一个本来生命周期短的对象存放到一个生命周期相对较长的对象中，也称“对象游离“。**隐蔽的内部类（Anonymous Inner Class）：

**\[html\]**  [view plain][]  [copy][view plain]

1.  mHandler = new Handler() \{  
2.      @Override  
3.      public void handleMessage(Message msg) \{  
4.      ....  
5.      \}  
6.  \};

相当于：

**\[html\]**  [view plain][]  [copy][view plain]

1.  class MyHandler implements Handler \{  
2.    ....  
3.  \}  
4.  mHandler = new MyHandler();

类似的还有Listener，Observer，Receiver等。  
  
由于内部类隐含拥有一个外部类对象的引用，所以千万要注意内部类instance的声明周期。  
最好不要将一个内部类的instance暴漏出去，除非使用者采用weak reference引用内部类的instance。例如，一些Handler，Listener在做为返回值暴漏出去时，千万注意使用weak reference。  
  
注意：一些系统提供的类偷偷地将自己暴漏出去了，如FileObserver，会将自己暴漏到自己内部开的线程去，不过还好它使用的是weak reference。但是过分的是BroadcastReceiver，一旦我们使用Context的registerReceiver 之后，系统（ ActivityManagerNative ）就会有一个它的强引用，除非我们显式调用unregisterReceiver。

结论：任何提供了开关函数的类，都必须保证显式地调用它们的开关函数。

**（二） 构造Adapter时，没有使用缓存的 convertView，从而造成系统创建了大量的view而没有来得及回收。**

public View getView(int position, View convertView, ViewGroup parent)

**（三）对于资源性对象在不使用的时候，应该调用它的close()函数，将其关闭掉，然后才置为null。**资源性对象，比如（Cursor等）往往都用了一些缓冲，我们在不使用的时候，应该及时关闭它们，以便它们的缓冲及时回收内存。它们的缓冲不仅存在于java虚拟机内，还存在于java虚拟机外。如果我们仅仅是把它的引用设置为null，而不关闭它们，往往会造成内存泄露。

**（四）都是线程惹的祸**

由于线程的执行时间是未知的，所以线程所持有的外部对象，一般要使用weak reference，除非你肯定它们的生命周期。

另外，向其它线程post Runnable或者send Message时也需要注意Runnalbe内部使用的变量和Message中的内容。

**（五）AsyncTask-隐藏的故事**

Asynctask使用了一个静态的ThreadPoolExecutor，有三个参数是系统固定的：

CORE\_POOL\_SIZE：5

MAXIMUM\_POOL\_SIZE：128

QUEUE\_CAPACITY: 10

当一个AsyncTask被execute之后：

1.如果线程池中线程数目小于CORE\_POOL\_SIZE ，就直接开线程；

2.如果队列不满，就被加入队列中；

3.如果队列已满，就直接开线程执行；

4.如果线程池中线程数目大于MAXIMUM\_POOL\_SIZE，就会Exception。

Tricky的地方在于AsncTask的Cancel操作。

1.不一定都能成功，例如while（true）\{\}就无法被cancel；

2.如果AsyncTask还在Queue中，那么cancel操作并不会将它从Queue中remove，而仅仅是设置了flag，最终当它被thread执行时，判断该flag，如果false，就不会执行doBackground的代码。

悲剧剧情：你在队列中，正在执行的几个哥们都block了，你就一直呆着吧。

**（六）Bitmap-不得不说的故事**

Bitmaps in Android are created innative memory, not on the VM heap, so the actual Bitmap object on the VM heapis very small as it doesn't contain any actual bitmap data.

一般情况下，我们都是按需索图，没必要把原始图片都搞进来，所以我们可以先查询原始图片大小，在做scale：

1. 查询原始图片大小，通过BitmapFactory.Options中的属性inJustDecodeBounds来做，If set to true, the decoder will return null (nobitmap), but the out... fields will still be set, allowing the caller to querythe bitmap without having to llocate the memory for its pixels.

2. 然后计算出一个scale的值，通过BitmapFactory.Options中的属性inSampleSize进行缩放。

**\[html\]**  [view plain][]  [copy][view plain]

1.  Bitmap bitmap;    
2.  float imagew = 300;  // 目标大小  
3.  float imageh = 300;  // 目标大小    
4.       
5.  BitmapFactory.Options bitmapFactoryOptions = new BitmapFactory.Options();    
6.  bitmapFactoryOptions.inJustDecodeBounds = true;    
7.  bitmap = BitmapFactory.decodeFile(imageFile, bitmapFactoryOptions);    
8.       
9.  int yRatio = (int)Math.ceil(bitmapFactoryOptions.outHeight/imageh);    
10. int xRatio = (int)Math.ceil(bitmapFactoryOptions.outWidth/imagew);    
11.      
12. if (yRatio **>** 1 || xRatio **>** 1)\{    
13.  if (yRatio **>** xRatio) \{    
14.   bitmapFactoryOptions.inSampleSize = yRatio;    
15.  \}    
16.  else \{    
17.   bitmapFactoryOptions.inSampleSize = xRatio;    
18.  \}    
19. \}    
20. else\{   
21.  // inSampleSize = 1  
22. \}    
23.      
24. bitmapFactoryOptions.inJustDecodeBounds = false;    
25. bitmap = BitmapFactory.decodeFile(imageFile, bitmapFactoryOptions);    
26. myImageView.setImageBitmap(bitmap);

Bitmap-recycle有用吗？ 以下是官方源代码的注释：

Free the native object associated with this bitmap, and clearthe reference to the pixel data.

This will not free the pixel data synchronously;

it simply allows it to be garbagecollected if there are no other references. The bitmap is marked as "dead", meaning it willthrow an exception ifgetPixels() or setPixels() is called, and will draw nothing. This operation cannot be reversed, so it should only becalled if you are sure there are no further uses for the bitmap.

This is an advanced call,and normally need not be called, since the normal GC processwill free up this memory when there are no more references to this bitmap.

**（七）substring的恶梦**

**\[html\]**  [view plain][]  [copy][view plain]

1.  public String substring(int beginIndex, int endIndex) \{  
2.       return new String(offset + beginIndex,   
3.                                   endIndex - beginIndex,   
4.                                   value);  
5.  \}  
6.  String(int offset, int count, char value\[\]) \{  
7.       this.value = value;  
8.       this.offset = offset;  
9.       this.count = count;  
10.   \}

也就是说即使你只想用其中的某几个字节，Java也自动把原始String的整个内容帮你保存下来了。

**\[html\]**  [view plain][]  [copy][view plain]

1.  ArrayList**<****String****>** oomStringList = new ArrayList**<****String****>**(1024);  
2.  for (int i=0; i**<****32**\*1024; i++) \{  
3.        String srcString = new String(new char\[1024\]);  
4.        String dstString = srcString.substring(0,1);  
5.        oomStringList .add(dstString);  
6.  \}

**Reference：**

http://kohlerm.blogspot.com/2009/07/eclipse-memory-analyzer-10-useful.html

http://java.dzone.com/news/how-fix-memory-leaks-java

http://kohlerm.blogspot.com/search/label/memory

转载地址：[http://blog.csdn.net/edisonlg/article/details/7082063][http_blog.csdn.net_edisonlg_article_details_7082063]

[0_1325909160ggLo.gif]: /images/20220727/c7c279f8a64f4c05bb48ab6d6accaca3.png
[0_132590919785Zf.gif]: /images/20220727/ed1403539d5442da8fbc2350041d5225.png
[0_13259092207fSP.gif]: /images/20220727/405f962b5e744c49aedea8d7f3a35cd1.png
[0_1325909241mzHF.gif]: /images/20220727/1989b02b9ade42b3a0830ca77b6ae5fc.png
[GC roots]: http://www.yourkit.com/docs/90/help/gc_roots.jsp
[0_13259092561igI.gif]: /images/20220727/652feeb539b841ba88d587b1cebc454b.png
[view plain]: http://blog.csdn.net/edisonlg/article/details/7082063#
[http_blog.csdn.net_edisonlg_article_details_7082063]: http://blog.csdn.net/edisonlg/article/details/7082063