Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 27 Sep 2008 10:20:38 +0800
From: Eugene Teo <>
Subject: Re: CVE-2008-4113 update: kernel: sctp: fix random
 memory dereference with SCTP_HMAC_IDENT option

Hi Steve,

Steven M. Christey wrote:
> On Thu, 25 Sep 2008, Eugene Teo wrote:
>> The first three references to CVE-2008-4113[1] are incorrect. Please
>> update the CVE with the following references:
>> [1]
> This was in reference to the TKADV2008-007 advisory.
> I guess the question becomes - TKADV2008-007 talks about separate issues,
> one involving crashes by calling the API functions when SCTP-AUTH is
> disabled (CVE-2008-3792), and another involving SCTP_HMAC_IDENT and a
> length value for sctp_getsockopt_hmac_ident.

I see what the confusion is now.

TKADV2008-007[1] mentioned two separate, but related issues. The second
issue that the advisory mentioned is an example of a function that may
have two possible consequences, and it all depends on whether SCTP
authentication is enabled or not.

The patch[2] that addressed these issues mentioned only one of them in
the changelog description, even though it appears to be fixing possibly
more than two issues.

Should this be assigned with just one CVE name instead of two?


The sctp_setsockopt_hmac_ident() bug I found was reported around the
same time as Tobias. More on this below.

> which seems different from this one:
> and this one:
> So Eugene, it sounds like the issues that you found were variants of the
> issue reported for the sctp_getsockopt_hmac_ident (a different function)
> by TKADV2008-007 for CVE-2008-4113.

The sctp_setsockopt_hmac_ident() bug I found has a user-controllable
hmacs->shmac_num_idents. There is no sanity check for id in
sctp_auth_ep_set_hmacs(), and the check for has_sha1 can be bypassed, so
it is possible that memcpy copies a little more than it should. It
depends on sctp_getsockopt_hmac_ident() to read the data structure back
to the user-space. When I reported this, sctp_getsockopt_hmac_ident()
problem as reported in TKADV2008-007 was not fixed yet. We probably need
a CVE name for this.

Thanks for clearing up the confusion.

Eugene Teo / Red Hat Security Response Team

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.