Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 22 Jul 2013 11:22:39 -0700
From: Forest Monsen <forest.monsen@...il.com>
To: Kurt Seifried <kseifried@...hat.com>, oss-security@...ts.openwall.com
Cc: "security@...pal.org" <security@...pal.org>
Subject: Re: CVE request for a Drupal contributed module

Hi Kurt, regarding CVE assignment and your request for clarification at
http://www.openwall.com/lists/oss-security/2013/05/16/2:

On Wed, May 15, 2013 at 6:41 PM, Kurt Seifried <kseifried@...hat.com> wrote:

> This sounds like two separate issues:
[...]
> can you send me the code patches fixing this so I can make sure it
> gets the correct SPLIT/MERGE treatment? Thanks.

Yep - Diffs for the commits that fixed both of these issues are at:

Drupal 6: http://drupalcode.org/project/ga_login.git/commitdiff/dd04ea3
 Drupal 7: http://drupalcode.org/project/ga_login.git/commitdiff/c365097

For the first issue,


> Accidental removal of account configuration.
>
> In certain scenarios, Google Authenticator login incorrectly
> determines the user's account name. The change in account name could
> cause the two-factor authentication for existing accounts to be lost,
> allowing users to log in using just username and password.
>
> This vulnerability is mitigated by the fact while Google Authenticator
> login's additional verification is by-passed, a username and password
> are still required to log in.
>

It looks like the maintainer now concatenates a "Realm" (site name) and
suffix with the Drupal username to form the GA username. Any inconsistency
there will invalidate earlier credentials.

For the second,

One Time Password (OTP) replay
>
> If an attacker can intercept a login request with a username, password
> and OTP, an attacker could use this same data again to login to the
> website.
>
> This vulnerability is mitigated by the fact that an attacker who can
> intercept a login request with this level of detail can usually also
> intercept the ongoing session identifying token.
>

It looks to me like the maintainer now implements a skew value to either
(in the case of a time-based one-time password token) review only a certain
range of timed tokens on either side, or (in the case of an HMAC-based
one-time password token) to again test a range of tokens.

I'll copy the Drupal Security Team, in case I haven't understood it
correctly or if further clarification is necessary. Thanks.

Best,
Forest

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.