Android software NDK native c code profiling on actual Android phones

I am developing a rather large software on Android with a log native code, it's working now but having some performance issues.

I am hoping I can profile each module(function call) of the software for CPU cycles, memory usage, etc, on several real android phones. Is there a simple c library to do that?

I see people using oprofile, It seems to be a overkill for my case since that it is a system wild profiler, and it requires rebuild the kernel and system image.

As I have the full source code of my app, all I really need is a simple c library that I can embed in my code to do some profiling while the app r开发者_如何学Cuns several test cases.

BTW, what is the Linux way of doing this?

I've had pretty decent results with android-ndk-profiler.

http://code.google.com/p/android-ndk-profiler/

Outputs /mnt/sdcard/gmon.out