Distinguishing between a memory leak and normal memory usage - android

I am creating an Android app, and I've only just recently begun to examine the memory usage.
My app consists of a main activity and two fragments within (Fragment A holds a recyclerview with images, Fragment B holds an image and some text boxes).
I've noticed that when I start the app, my memory allocation is at roughly 20MB. By scrolling up and down on the recyclerview in Fragment A (not loading any new data) I can make the memory usage increase nonstop (until garbage collection).
Also, if I go from fragment A to fragment B back and forth, I can make the memory allocation increase nonstop (until garbage collection). 20 -> 23 -> 29, .. etc. Until it gets to 60+ and gets garbage collected.
If I start the app at roughly 23MB allocated, and do some scrolling and moving back and forth between fragments, my memory usage might be somewhere between 40 and 60+ MB.
When I use Android Studio to garbage collect multiple times in a row, I can bring it back down but not all the way to 23 (usually to 30ish).
Is this a memory leak since I'm unable to garbage collect 100% of the memory I started out with?

No, strictly speaking you can't have that kind of memory leak in java since it is garbage collected.
You'll get "leaks" in java if you are keeping a reference to something you no longer use (I.E. a big list or something that you never remove from). But keep in mind that these can always be fixed by being sure to discard references to things you don't need to keep around (removing things from lists/sets/maps, closing files ... etc). The persistent memory that you are looking at is probably from some static variables since those are bound to the classes.

Related

Why does app allocated memory permanently increase when showing fragments or starting activities?

I'd like to ask the community what they think about this behavior.
On my main activity I have one FloatingActionButton that when clicked will show a DialogFragment.
I followed Google's guidelines (I can't link it due to being a new user here at Stack Overflow) on how to properly create a DialogFragment so I believe I'm OK there. I've tested this behavior on fragments that use either onCreateDialog and onCreateView, and in both cases the results are the same.
When I click the button and I look at the memory usage, this is what I see:
Memory when fragment is first showed
So Android is allocating additional resources to compensate for showing a new fragment. That makes sense to me.
However, when I dismiss or cancel the dialog, the allocated memory doesn't drop.
When I open the same dialog using the same button, the allocated memory jumps again. I repeat this over and over (red arrows) until my allocated memory reaches just under 10MB and then (I believe) the Garbage Collector (blue arrow) kicks in and cleans up the app:
Memory after multiple showings
To me, this means I'm relying on the Garbage Collector to do the work for me and from what I understand that is not good practice.
Can anyone tell me if this is normal Android behavior, or is there something I'm doing wrong? If It isn't normal behavior, I will try to resolve the issue myself first before asking the community for further assistance (this is why I did not provide any code).
Thank you in advance.
I think this excerpt from this page sums up this behavior nicely.
Android does not offer swap space for memory, but it does use paging and memory-mapping (mmapping) to manage memory. This means that any memory you modify—whether by allocating new objects or touching mmapped pages—remains resident in RAM and cannot be paged out. So the only way to completely release memory from your app is to release object references you may be holding, making the memory available to the garbage collector. That is with one exception: any files mmapped in without modification, such as code, can be paged out of RAM if the system wants to use that memory elsewhere.
Basically, all you can do as an app developer is to release references to objects you allocate. So assuming you aren't keeping a reference to the object anywhere, like placing the fragment on the back stack for later use, the reference to the object should be released when the fragment is closed. At this point, it is available to the garbage collector to clear the previously allocated memory.
After that is done, it is up to the garbage collector to decide when to come in to action, so to say.

Android clear the back stack if out of memory

When running on Huawei G300 with Gingerbread, my app crashes after 5 minutes or so of usage during setContentView() as it runs out of memory.
Each individual page doesn't use much memory, but from some research it seems the memory accumulates in the back stack.
Following advice here, I've replaced all my calls to startActivity with a utility function that also calls finish().
Android: Clear the back stack
This works; but there is no more back stack - the back button immediately quits the app, which isn't what I wanted.
Is there a way to only finish() the applications when I actually do run out of memory, and is that a reasonable approach to take?
You should search for memory leaks. A good tool for that is MAT if you use eclipse. MAT is not that hard to handle and you can get quickly some very valuable information.
One of the most common mistakes I have seen on Android is to keep a reference on a context that is dead. For instance, having a singleton holding a reference on one of the activities you created. There is no real reason for an app to crash the memory if it is well coded.
The Android Activity Manager was designed to manage this exact problem. The OS is designed to kill activities in the background and then restore them using onSaveInstanceState and onRestoreInstanceState.
The fact that your app is accumulating memory usage over time indicates to me that you may have a Context leak somewhere (a static reference to a view, adapter, etc. that has a reference to a Context), or that you have a caching mechanism that's not adjusting to your memory heap, or some other situation that's causing the out of memory.
I highly doubt that it's the Activities in the Back Stack causing the Out of Memory.
Here's a great guide on tracking down memory leaks on Android:
http://android-developers.blogspot.com/2011/03/memory-analysis-for-android.html
Memory is a very tricky subject in Android.
Every app gets a heap memory limit depending on the device. This heap memory is the dalvik memory plus the native memory, and you can see it as the total column in the dumpsys meminfo results. The dalvik memory deals with everything except with the bitmaps, that are allocated in the native memory (this is true for Android versions before Honeycomb).
Having said that I can only answer to some of your questions:
As far as I know, Android will always allocate memory for Bitmaps, even if they are the same. Therefore, in your case, every activity allocates memory for your background.
I don't know if its better to work with themes, you'll have to try that.
On one hand, activities are not reclaimed while the device has enough memory to deal with the next activity. Every activity is pushed to a pile from where it is recovered when you pressed the back button. In case Android needs more memory it removes one activity from the pile deallocating its memory (going back to question number one, maybe this is the reason for not sharing memory). On the other hand, you can set the activities launchMode to change this behaviour (have a look here).
I think MAT doesn't show native memory data. Use dumpsys meminfo's native column to see how much allocated memory for Bitmaps you have.
I have had hard times dealing with OutOfMemory problems myself. Now I have a much more clear idea of how it works and I am able to work with large files without running out of memory. I would highly recommend these two resources that helped me a lot:

android GoogleMaps memory leaks (?)

I'm working on a project dealing with map activities.
I was running a testing app consisting of a single activity, MaptestActivity, that extends MapActivity and does nothing else. I rotated the device more than ten times and I analyzed the heap dump with MAT. I noticed the existence of two instances of the main activity hanging around (a leak?). I re-tested the app with the same activity extending only Activity: the dump showed only one instance being kept by the system, the current one.
I'm attaching the screenshots of both cases. I'm not much into MAT, maybe I'm just misunderstanding the reults and everything works fine. Could you please shed some light on it?
With memory leak
Without memory leak
If you still have two activities in memory after rotating the device then you do have a memory leak. You can use MAT to see what is preventing that activity from being garbage collected by using the Histogram. Search on the class name to find it. Then right click on it and select list objects/with incoming references. That will list each occurrence of this object in memory. Right click on one of these objects and select Path to GC roots/exclude weak references. Now you should be able to see a hierarchy of objects, of which something is not getting cleaned up completely after the activity's onDestroy has finished. Things like hard references to a static context will cause this. To resolve some of our memory issues we've had to do things like use weakReferences where appropriate, and clean things up in onDestroy (like setting listeners to null).
Normally if your activities are not holding onto large objects like bitmaps you can "get away" with memory leaks since it would mean lots of app usage before you exceed the device's application heap size. But if say all of your activities hold onto a large bitmap, then leaking activities become much more of a big deal.
This is a good video: Memory Management for Android Apps
And a good article: Avoiding Memory Leaks
add the below line in for your activity in your manifeast file
android:configChanges="keyboardHidden|orientation"

Android Heap Fragmentation Strategy?

I have an OpenGL Android app that uses a considerable amount of memory to set up a complex scene and this clearly causes significant heap fragmentation. Even though there are no memory leaks it is impossible to destroy and create the app without it running out of memory due to fragmentation. (Fragmentation is definitely the problem, not leaks)
This causes a major problem since Android has a habit of destroying and creating activities on the same VM/heap which obviously causes the activity to crash. As a strategy to counter this I have used the following technique:
#Override
protected void onStop() {
super.onStop();
if(isFinishing()) {
System.runFinalizersOnExit(true);
System.exit(0);
}
}
This ensures that when the activity is finishing it causes a complete VM shutdown and therefore next time the activity is started it gets a fresh unfragmented heap.
Note: I realise that this is not the "Android way" but given that the garbage collector is non-compacting it is impossible to continuously re-use the heap.
This techinque does actually work in general, however it doesn't work when the activity is destroyed in a non-finishing mode and then re-created.
Has anyone got any good suggestions about how to handle the degredation of the heap?
Further note: Reducing memory consumption is not really an option either. The activity doesn't actually use that much memory, but the heap (and native heap) seem to get easily fragmented, probably due to some large'ish memory chunks
Fragmentation is almost always a consequence of an ill conditioned allocation pattern. Large objects are frequently created and destroyed. In conjunction with smaller objects may persisting (or a least having a different lifetime) - holes in the heap are created.
The only working fragmentation prevention in such scenarios is: prevent the specific allocation pattern. This can often be done by pooling the large objects. If successfull, the application will thankfully acknowledge this with a much better execution speed as well.
#edit: yet more specific to your question: if the heap after a restart of the application is yet not empty, so what is there to remain on the heap? You confirmed that its not a problem of a memory leak, but this is, what it seems. Since you are using OpenGL - could it possibly be, some native wrappers have survived, because the OpenGL ressources have not properly been disposed?

How to implement an in Memory Image Cache in Android?

Until now I'm using a SoftReference Cache for images in Android. This cache is used for images that are shown in ListViews and should help holding images of items in memory that are not shown on the screen, if there is enough memory left.
The problem with this is that SoftReferences are garbage collected nearly at the very moment the last hard reference is released. The result of this is that an image that is removed from the screen is garbage collected at that moment and if the user scrolls back to this entry in the ListView the image is reloaded from the internal phone memory resulting in a complex lazy loading process, resulting in frequent redraws of the list and general bad performance.
The bug request for the soft reference behavior states that this is the intended behavior and that you should use a LRU-Cache for caching this kind of stuff.
Know to my question. The LRU cache will only take as much memory as I allow him to. But if the app needs a lot of memory it will not free memory. How should I determine how much memory I can allow the Cache to use, and is there a way to reduce the size of the cache if the memory situation of the phone becomes tight?
At the moment the image cache is saved inside the application as a kind of global image storage for all activities. This would result in my app constantly using all the memory of the image cache even if my activities are in the background or destroyed.
Keeping background processes alive is an OS-level optimization that makes switching back to your process fast. Your process will stay alive only as long as the OS can afford the memory; when that memory is needed for another application your process will be killed and its resources will be released.
If you free your cache each time your process is backgrounded, switching back to your application would no longer be fast because it would see cache-misses. This defeats Android's keep-background-processes-alive optimization!
You should just use the LruCache, which is also included in the Android Support Package for releases prior to Android 3.0 (Honeycomb).

Categories

Resources