Java Bytecode Profiler

[Senthil Kumar]

Tom Harkins wrote:
> 
> I am trying to design a Java Bytecode Profiler using the kaffe source code.
>  I am trying to do some analysis to keep track of branches, jumps, and
> function calls done by the Java Bytecode.  I have downloaded the kaffe
> source code and compiled it on a Sun Ultra-Sparc machine.
> 
> Does anyone know where in the source code I might start to look for the
> code that does the branching and jumping.  I was looking at the
> /kaffe-1.0.b1/kaffe/kaffevm/code-analyse.c file which looks like it is
> parsing out the beginning and end of basic blocks.
> 
> Has anyone attempted to do such a project with or without success.
> 
> Tom

you are on the right track. I think, except for function
calls (INVOKE etc) you can get all such info. Getting info
on function calls shuold be trivial, though. (I think)

I did some work on this.

Just from the size of the byte code, some info can be gathered,
such as:

1. verification time (linear correlation, this is expected)
2. translation time (linear correlation)
3. upper bound to interpret it. This result is quite interesting
   and approximate. I use it to initially estimate the
   time it will take to interpret some given method. Which is
   used in a heuristic to determine if I want to JIT/interpret
   some given method. The interpret time is the time spent
   in some given method, i.e. it does not include time spent
   in subsequent functions calls inside the method.

Senthil