oss-sec mailing list archives

Re: Old cscope buffer overflow


From: Tomas Hoger <thoger () redhat com>
Date: Wed, 6 May 2009 18:49:22 +0200

On Wed, 6 May 2009 11:49:14 -0400 (EDT) "Steven M. Christey"
<coley () linus mitre org> wrote:

If you're preparing cscope updates for CVE-2009-0148 and you may
still be shipping packages based on 15.5, you may want to have a
look at:

  https://bugzilla.redhat.com/show_bug.cgi?id=499174

Steve, as the first public report for this is from 2006:

  https://bugzilla.redhat.com/show_bug.cgi?id=189666

I believe 2006 CVE id is needed here.

We recently updated CVE-2009-0148 for overflows in cscope before
15.7a. Is this the same issue, or do we need a different one?

This seems to be distinct from CVE-2006-4262 as well...

Different from both.  CVE-2009-0148 is more of a dupe / re-occurrence /
incomplete fix of even older CVE-2004-2541.  Some vendors originally
addressed that via sprintf -> snprintf across whole sources.  Upstream
instead preferred to use "%.*s" with bit of length math to avoid
overflow and not harm portability.  Though the math could int
underflow, still allowing buffer overflow.

BZ#189666 issue was fixed upstream in 15.6.

-- 
Tomas Hoger / Red Hat Security Response Team


Current thread: