Re: Pidgin IM Client Password Disclosure Vulnerability.

[John Bailey <rekkanoryo () rekkanoryo org>]

On Thu, Sep 18, 2008 at 03:16:18PM -0400, Memisyazici, Aras wrote:
> While I agree with your comments, I cannot help but suggest that maybe the method of choice could be 'security 
> through obscurity' whereby they take a hash of the password, with a non-std. hashing mechanism. The idea being that 
> in today's world where there are so many scr1pt-kiddi3 toolz out there allowing the avg. Joe Schmoe the capability of 
> analyzing one's memory processes i.e. Tsearch, memhack etc... It only makes it non-trivial for them to extract the 
> info needed. This way you are making it a tad more annoying and adding another buffer they need to bypass :)
>
> Just a thought,
> Aras 'Russ' Memisyazici
> Systems Administrator
>
> Virginia Tech

Wow, security through obscurity.  That's a good practice alright.  So
you propose that I and my fellow Pidgin developers implement security
through obscurity, thus giving our users a false sense of security?  No
chance.  Note also that we store passwords on-disk without any form of
encryption or obfuscation, which has been debated to death on numerous
occasions--so much so, in fact, that we've written an FAQ entry dealing
specifically with this.  Additionally, *any* form of encryption that we
were to use would have to be reversible, as storing protocol-specific
hashes is, as Siim pointed out, no better than storing the plain text.
Reversible encryption again makes it completely trivial to decrypt the
passwords (by using our own code against the user), to the point that
it really is no "improvement" at all.

I find it curious that the person disclosing this so-called
vulnerability made no effort to contact us before disclosing, let alone
do any research to find out where the affected area of code is (the code
being complained about here is in libpurple, not Pidgin).  We have
enough visible methods of contact that there is no excuse for not
attempting to contact us directly.

John

Attachment: signature.asc
Description: Digital signature