[prev in list] [next in list] [prev in thread] [next in thread] 

List:       oss-security
Subject:    Re: [oss-security] libtiff buffer underflow in LZWDecodeCompat
From:       Vincent Danen <vdanen () redhat ! com>
Date:       2009-06-23 23:14:09
Message-ID: 20090623231409.GA8606 () redhat ! com
[Download RAW message or body]

* [2009-06-21 17:14:24 -0700] Kees Cook wrote:

>A crafted TIFF can crash libtiff in LZWDecodeCompat via underflow (different
>from CVE-2008-2327).
>
>Based on discussions[1] and a quick analysis[2], I don't think this is
>exploitable, but it does lead to crashes in any application using libtiff.
>I've reported it upstream[3], with the attached patch.
>
>Has anyone else looked this over?
>
>-Kees
>
>[1] http://www.lan.st/showthread.php?t=1856&page=3
>[2] https://bugs.launchpad.net/bugs/380149
>[3] http://bugzilla.maptools.org/show_bug.cgi?id=2065

You saw that a new comment was posted to [3] that points to an earlier
bug and a different patch, right?  Looks like it was just updated today,
to point to this bug report from january:

https://bugzilla.redhat.com/show_bug.cgi?id=1985

Also, that report seems to agree with your quick analysis:

"However, the previous patch does appear to prevent a payload of more than one distinct byte,
making this effectively useless as a code injection vector. Nonetheless, it
still is effective at crashing applications that use LibTIFF."

In fact, I think the reporter of that bug was one of the writers in the
lan.st forum notes you're showing, particularly based on this comment
where he indicates it isn't exploitable and that he filed a bug:

http://www.lan.st/showpost.php?p=13094&postcount=58

-- 
Vincent Danen / Red Hat Security Response Team 
[prev in list] [next in list] [prev in thread] [next in thread]