Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 16 Nov 2014 18:15:57 +0100
From: Robert Święcki <>
Subject: Re: Fuzzing findings (and maybe CVE requests) -
 Image/GraphicsMagick, elfutils, GIMP, gdk-pixbuf, file, ndisasm, less

2014-11-16 15:10 GMT+01:00 Hanno Böck <>:
> Hi,
> I wanted to share a couple of issues I recently found via zzuf and afl
> fuzzing. It's a telling story about the state of some of the free
> software projects involved and I can only encourage others to join the
> effort to find bugs via fuzzing. Some of them are really low hanging
> fruit.
> I'm cc-ing cve-assigners, I leave it up to you to decide which you
> assign CVEs. If you want / need more info on details please ask.
> Imagemagick:
> Multiple issues in PCX, DCM parser and generic issue in resize code
> These already got CVEs:
> GraphicsMagick:
> Fork of Imagemagick, so some of the above also affect it, tests with
> the same fuzzed sample set turned out one independent other issue:
> Heap Overflow / oob read
> One more issue with PNGs that turned out to be weird, it caused an
> error message to overflow:
> elfutils:
> Checks done with the set of files that crashed binutils turned out one
> issue:
> Invalid read
> american fuzzy lop found a couple more:
> and more:
> Invalid reads in import plugins for fli and tga.
> claws-mail / gdk-pixbuf
> Assert in gdk-pixbuf when trying to load a malformed file as an
> animation. This was an accidental discovery when I clicked on a
> malformed PNG I send while reporting another issue (in graphicsmagick)
> in my mail client (and it crashed with an assert).
> file/libmagic:
> out of bounds read when parsing JPG header
> ndisasm:
> Actually I found this by running ndisasm on /dev/urandom - no joke!
> Crash / oob read:
> less:
> Out of bounds read, upstream doesn't answer and doesn't have a public
> bug tracker. This wasn't really found by fuzzing but by running less on
> a likely malwared gif, I reduced it to a smaller testcase:

Really nice job!

Just to reiterate what others in previous "fuzzing" threads had already said.

A given bugs is a security where given set of code is by design (and
after a solid security review process) supposed to be exposed to
untrusted inputs.

I can easily see that certain libs you're fuzzing (imagemagick et al)
are widely used to parse untrusted inputs (image conversion pipelines
and such esp. in the web software) and by discovering bugs in them
you're doing a great job preventing some web servers and a couple of
users from being pwnd.

However, even if tools like file/ndisasm/gimp/readelf can be used by
many (w/o strong system isolation boundaries) to analyze untrusted
inputs (for reverse engineering, malware analysis and similar
purposes) - I'd simply put a blame on those users when if they get
pwnd - as they're depending on tools, which hadn't been properly
evaluated for the purpose (by efforts of those users, or by their
contractors or by the community at large) and the likelihood that
we'll start accepting those tools as good enough for said purposes in
the coming years is seriously low.

There are thousands of software packages available under a typical
Linux distro. And a serious share of it will fail when dealing with
untrusted inputs. So, maybe instead of fuzzing everything what is
available from the command-line - we should rather draw a line between
software which is designed and tested for the said purpose (web
browsers, kernel networking stacks, network services) and software
which is not. The alternative is that we'll spend a lot of time in the
next year(s) discussing seemingly non-security bugs on security lists.

To sum up: If somebody uses 'file' in an unconstrained OS environment
on untrusted inputs, and he gets pwnd in the result, then it's not a
security problem, it's an incompetence problem - and IMO it should be
discussed elsewhere.

I appreciate your enthusiasm to fuzz every OSS project out there - I
just wanted to ask you (and others who might join this effort) to
apply a well-thought-out set of criteria when categorizing your
findings (security bug, usability bugs). And good luck with your

Robert Święcki

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.