A crowdfunding campaign to build a free baseband

[Paul Sokolovsky]

Hello,

On Sat, 11 Apr 2015 18:15:03 GMT
falcon at ivan.Harhan.ORG (Spacefalcon the Outlaw) wrote:

> Paul Sokolovsky <pmiscml at gmail.com> wrote:
> 
> > 2+ MB is "small part"?
> 
> TI's full source for the GSM+GPRS protocol stack weighs 51 MiB - and
> that is just the C source (*.[ch] files) for the core protocol stack.
> The complete ZIP with the reference firmware deliverable for the
> LoCosto chipset weighs 120 MiB.

A spectacular example of vendor bloatware, thanks for sharing the
figures.

> > Thank you, god, that I don't have a fixation to
> > have my own GSM stack then.
> 
> So you have no problem with using proprietary software whenever you
> need to use a phone?

I don't ;-). As I mentioned right in the 2nd mail of our exchange (ref:
http://lists.osuosl.org/pipermail/replicant/Week-of-Mon-20150406/000677.html
), my biggest interests are: 1) RE methodology of your particular
project, and 2) RE methodology of all "RE-for-OpenSource" projects. And
the fact that most such projects, yours including, are done in the same
manner, repeating previous mistakes, like wasting time on
reimplementing tools which already exist.

The topic of your project is of course interesting too, but there're
many interesting OpenSource projects.

[]

> > ... I'd also think that L1, which needs to interface with hardware
> > is the most critical, upper layers can be taken from other
> > implementations, like OsmocomBB or FreeCalypso.
> 
> Then why don't *you* work on that project. 

Because I work on gazillion of other projects ;-). And it's just
coincidence that over last half year I hit 3rd interesting
reverse-engineering project, 2 of which I consider poking my nose into,
that I'm interested to discuss this topic at all.

> Would it be possible to do
> what you propose?  To the best of my knowledge and understanding,
> probably yes.  Is it something that I would be willing to spend my
> unpaid volunteer time on with absolutely no benefit to me or my family
> or any of my friends?  Hell no.

You're in denial. I also wrote about this in the mail linked above.
People start brave, waste their time reimplementing disassemblers,
object format parsers, binary signature matchers, etc. Of course, those
quick-n-dirty tools suck, so they crawl with actual RE targets, and
then swear not to do that RE stuff ever again. Judging by myself, that
oath doesn't work, as in few years some another interesting piece of
stuff needs liberation from vendor madness.


So, anyway, seeing your project was last straw to break a camel back,
and prompted me to do something which someone should have done long ago
- implement open-source easily hackable interactive disassembler with
direct-manipulation user interface:

https://github.com/pfalcon/ScratchABit  

That's certainly not the first attempt, tools like
http://bastard.sourceforge.net/ died ages ago in refactoring code
written in a language folks usually use to develop firmwares, not
applications, a lot of other tools didn't go from being a library into
something which has a usable UI, and turbopascal kiddies having 3rd
iteration of a tool imaginatively called "GNIDA":
https://github.com/Shooshpanchik/GNIDA3 .

That all certainly existed, but nothing was really usable, comparing to
de-facto standard proprietary tools. Well, a week later, I already set
to use my piece for my next RE task, and if it helps someone else in
the community, I'm only glad (I'm certainly write it to be usable by
other people, e.g. I in no way need to disassemble boring x86 code, but
adding default x86 plugin is what I spent last 2 days on, so folks
actually can start up ScratchABit and see something).

(I understand this discussion goes off-topic for Replicant mailing list,
so this is my last mail on this topic.)


-- 
Best regards,
 Paul                          mailto:pmiscml at gmail.com