Re: Re: Re: Re: Fwd: Re: More precise values in Analyzer

On Fri, Jul 16, 2010 at 8:18 AM, Rémi Forax <[email protected]> wrote:

>  Le 16/07/2010 16:01, Joshua Warner a écrit :
>
> [...]
>
>
>
> Why not tracking store ? i.e each store create a new variable.
>
>> This is the usual algorithm to convert into SSA form.
>>
>
> This works fine in most cases, but this still wont work for parameters
> because there is no "load_param" bytecode instruction in Java - all of the
> parameters are implicitly loaded into the low local variable indices before
> the bytecode for a method begins executing.
>
>
> I don't see why it's a problem.
> You just start your algorithm by creating one SSA variable by local
> variable.
> I think this algorithm is described in Appel's book Modern Compiler
> Implementation in Java.
>
>
>
It is not a problem - if I want to rewrite all of the bytecode interpreting
logic in ASM.  The problem is working with the existing Analyzer,
Interpreter, Frame, etc. classes already in ASM.  They don't seem to have
support for detecting values coming from local variables as such.

Sincerely,

Joshua


>
>
>
>>
>>
>>
>> Even if this would work, I don't see a way to get the specific local
>> variables index that a certain value originated in (particularly in the case
>> of Values allocated with Interpreter.newValue(...).  Am I missing something?
>>
>> I've read significant portions of "ASM 3.0 A Java bytecode engineering
>> library <http://download.forge.objectweb.org/asm/asm-guide.pdf>", and I
>> didn't see anything that could help me with this particular problem.  Do you
>> have a certain section that you are thinking of?
>>
>> Thanks,
>>
>> Joshua
>>
>>
>>  There is also already a llvm project that implement a basic bytecode
>> frontend:
>> https://llvm.org/svn/llvm-project/java/trunk/docs/java-frontend.txt
>>
>> cheers,
>> Rémi
>>
>
> Thanks, that's definitely something to look into some more; however, as the
> generated code has to work closely with the internals of a specific JVM, I
> expect I would have to make numerous changes to how it lowers operations
> like virtual method calls and interface calls, as well as what format it
> expects its objects in.  This may or may not be easy.
>
> Sincerely,
>
> Joshua
>
>
> cheers,
> Rémi
>
>

-- 
You receive this message as a subscriber of the [email protected] mailing list.
To unsubscribe: mailto:[email protected]
For general help: mailto:[email protected]?subject=help
OW2 mailing lists service home page: http://www.ow2.org/wws