From: Mårten Cassel (mcasselkth.se)
Date: Thu Apr 27 2006 - 11:44:07 CDT

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi,

Thanks for your interesting cents.
 I think i've done some flawed design choices in my debugger and the
complexity of it has exploded, this was partly because i didn't
understand the problem clearly enough and the output from doing bytecode
tracing was so exciting but tracing execution was extremely slow...
 The design problems seems to me to be the most challenging ones, well the
"stuff" that happens before/inbetween? coding... :)

/ marten

> The internal architecture of both tools is quite different. Secondly
> OllyDBG
> is not really supporting anything besides x86, unless you already have a
> plugin that supports your proprietary bytecode. Also some capabilities of
> both tools are different (not to mention differences between IDA and
> OllyDBG
> code analysis options).
>
> Our approach to similar problem was to write a metaprocessor that works on
> simple abstract code and understands some simple actions like breakpoint
> management, memory read/write, stack unwinding etc. Actually the
> breakpoint
> and memory read/write operation for actual process are not implemented
> within metaprocessor code but are part of larger framework. This design
> was
> dictated by ability to use hardware breakpoints too and ability to deploy
> some debugging at kernel level on Win32 thanks to dedicated driver based
> on
> WMD.
> Just my 2 cents,
> Alex Czarnowski
>
>> -----Original Message-----
>> From: Marten Cassel [mailto:mcasselkth.se]
>> Sent: Wednesday, April 26, 2006 3:52 PM
>> To: Pedram Amini
>> Cc: Dave Aitel; dailydave
>> Subject: Re: [Dailydave] Request of the day
>> I'm writing a debugger plugin for IDA Pro that's supposed do
>> be a debugger for a proprietary bytecode interpreter /
>> virtual machine. What i'd like is a abstraction layer in som
>> form so i can easily port my debugger to use either Ida Pro
>> or OllyDbg and use the debugger services.
>> Do you have any clues of what kind of software design would
>> be the best way to accomplish this?
>>
>> - Marten
>>
>
>

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]