Re: CVE Request: Input validation failure affecting multiple KDE applications, as well as many other Qt-based applications

[Tim Brown <timb () nth-dimension org uk>]

On Monday 25 Jul 2011 19:51:47 Jeff Mitchell wrote:

> We've been made aware of an input validation failure affecting multiple
> KDE applications. (The details are not yet public as we're working on
> the fixes.) We'd like a CVE for this.
>
> The Arora and Rekonq web browsers are also vulnerable to the same attack
> vector, and other Qt-based programs may be as well. We're working with
> the Qt team to help enhance their documentation to warn developers to
> take care sanitizing their inputs, but it's not actually a Qt flaw. So
> we're a bit unsure how to proceed here. Do we get separate CVEs for
> Arora and Rekonq? Do we lump both of those into the same CVE as the KDE
> applications? I would think the former since other applications may be
> found to be vulnerable down the line, but wanted to check.
>
> (The Rekonq team has been made aware and are currently patching their
> code; I'm in the process of trying to notify the Arora team.)

Without disclosing too much, I'd go for a per module based CVE.  So one per 
affected KDE module + 1 per affected Qt module, + 1 per app that does its own 
thing (i.e. Arora, Rekonq and others we've already identified).  I think that's 
the most intelligent thing to do although I do think that one of the concerns 
is that more examples will likely come out of the woodwork (it's nothing that 
can easily be identified programatically) and each edge case is likely to be 
fairly application unique.  As an analogy, the best I can think of is XSS via 
PHP_SELF, do we give the CVE to PHP, frameworks that use PHP or individual 
apps that use it badly?  From what I can tell it seems to be per app/framework 
module.

Tim
-- 
Tim Brown
<mailto:timb () nth-dimension org uk>
<http://www.nth-dimension.org.uk/>

Attachment: signature.asc
Description: This is a digitally signed message part.