[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 2 Oct 2014 05:21:40 +0400
From: Solar Designer <solar@...nwall.com>
To: oss-security@...ts.openwall.com
Cc: Chet Ramey <chet.ramey@...e.edu>
Subject: Re: More parser odities
On Wed, Oct 01, 2014 at 05:53:45PM -0700, Tavis Ormandy wrote:
> Eric - the prefix you're specifying _is_ the long-term solution, it
> may be a bug, but it's a non-security bug.
I fully agree with Tavis and Michal on this. The only reason I brought
this to Chet's attention is that Chet appears to want to take this
opportunity to make the parser more robust in general (great!), beyond
fixing the security problems.
Unfortunately, a lot of people are confused.
Some people want urgent fixes for each and every new parser bug. They
don't actually need those fixes much! These are not security issues
anymore, as long as a variation of the prefix/suffix patch is applied -
and this patch is now available upstream (for several days already).
It's the only patch people need (or alternatively a patch disabling
function imports altogether, for the same security effect). Even the
original CVE-2014-6271 fix becomes unimportant with the prefix/suffix
patch in place.
In that sense, it's a pity we have several CVEs assigned to issues the
patches for which are relatively unimportant to have, whereas we can't
get a CVE assigned that would somehow correspond to the only important
fix (removing the parser from being exposed via arbitrarily-named env
vars), because it's not strictly fixing a vulnerability - it "just"
hardens the code so greatly that multiple separately patched
vulnerabilities stop being such. I guess this could fall under some CWE
rather than a CVE... or maybe we can in fact get a CVE assigned to "bash
exposes its code parser, which is unprepared to handling untrusted code
fragments, to untrusted input via arbitrarily-named env vars"? Like I
just said, we already have a fix for this upstream. Maybe it's just the
right opportunity to do that, instead of assigning yet another CVE to
Michal's latest finding? Oh, I think Michal's latest already got a CVE
too? Does this once again render the (previously) exposed parser not
CVE-worthy? ;-(
Then, some people somehow think that bash sort-of trusting BASH_FUNC_*
env vars is somehow a security problem or that it somehow leaves much
room for security hardening, even though bash and dynamic linker and
libc also trust many other fixed-name env vars. I think this is just
another kind of confusion. Several of us said so before, but I felt I
needed to add my +1 here: not a security problem.
Alexander
Powered by blists - more mailing lists
Please check out the
Open Source Software Security Wiki, which is counterpart to this
mailing list.
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
