[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 17 Dec 2009 20:33:31 +0000
From: Joe Orton <jorton@...hat.com>
To: Raphael Geissert <geissert@...ian.org>
Cc: oss-security@...ts.openwall.com
Subject: Re: CVE request: php5: multiple issues
On Thu, Dec 17, 2009 at 01:23:33PM -0600, Raphael Geissert wrote:
> I think a cross-vendor security support and tracking effort for php5
> is needed. The number of issues silently fixed are a continuous risk,
> leaving users exposed. What does the others think?
The problem we face is the ambiguity around the threat model for the PHP
interpreter. If you assume that the PHP interpreter should be robust
against attack from a malicious script (or its author), then a vast
number of bugs can be considered a security vulnerability.
Even if you assume that the PHP interpreter - and scripts using it -
should be robust only against attack from a remote user, it is often
still difficult to draw the line between a script bug and an
interpreter/extension bug. Doing so requires interface documentation
which specifies API preconditions and guarantees with greater precision
than is usually available.
So whether or not security issues are being "silently fixed" depends a
lot on your frame of reference.
Ideally any effort to improve the lack of transparency around PHP
security would start by working with upstream to a) define a threat
model and b) improve strictness of API/quality of code in the context of
that model. I wouldn't underestimate the time and effort that would
require ;)
Using this list to track and share analysis of published issues is
certainly helpful, but I'm not sure what more we can/should do
independent of upstream to improve the situation - any specific ideas
you had?
Regards, Joe
Powered by blists - more mailing lists
Please check out the
Open Source Software Security Wiki, which is counterpart to this
mailing list.
Powered by Openwall GNU/*/Linux -
Powered by OpenVZ