oss-sec mailing list archives

Re: CVE-Request taglib vulnerabilities


From: Ludwig Nussel <ludwig.nussel () suse de>
Date: Mon, 26 Mar 2012 10:06:33 +0200

Zubin Mithra wrote:
On Wed, Mar 21, 2012 at 10:49 PM, Kurt Seifried <kseifried () redhat com>wrote:

On 03/21/2012 09:42 AM, Ludwig Nussel wrote:
Zubin Mithra wrote:
[...]
The issues which are present in the latest "release" but not in the
current
development head were :-

[3] Lack of sanity checks of fields which were read, and were used for
allocating memory; crafted files would lead of application crash.

Not an issue according to upstream:
http://mail.kde.org/pipermail/taglib-devel/2012-March/002187.html

Shouldn't it simply say "file to large" or "unable to allocate blah"
something rather than crashing? I assume by "large" file the file
doesn't actually need to be large, just the header information needs to
claim it is large?


Yes, the file does not need to be large, it just needs to have a crafted
header.

On investigating the issue further, discussing with a developer Lukas
Laninsky and providing PoC's, we had confirmed that the root issue was an
Integer overflow - which would cause a large allocation and crash the
application.

The changeset that corrects it can be found here =>
https://github.com/taglib/taglib/commit/dcdf4fd954e3213c355746fa15b7480461972308


[4] A one bit change in a working ogg file would cause a thread to loop
infinitely.

http://mail.kde.org/pipermail/taglib-devel/2012-March/002191.html

https://github.com/taglib/taglib/commit/b3646a07348ffa276ea41a9dae03ddc63ea6c532

Has this been confirmed? Does the looping thread actually cause a DoS,
simply slow down the application a bit, or?


Yes, it just causes a thread to cause an infinite loop and does not cause
an application crash.

So both issues qualify as security issue and CVE assignment then, right?

cu
Ludwig

-- 
 (o_   Ludwig Nussel
 //\
 V_/_  http://www.suse.de/
SUSE LINUX Products GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg) 


Current thread: