Bugzilla Bug 218966

http://lists.digium.com/pipermail/asterisk-announce/2008-April/000139.html


[asterisk-announce] Asterisk 1.2.28, 1.4.19.1, and 1.6.0-beta8 Released
The Asterisk Development Team asteriskteam at digium.com
Tue Apr 22 18:05:07 CDT 2008

The Asterisk development team has released versions 1.2.28, 1.4.19.1, and
1.6.0-beta8.

All of these releases contain a security patch for the vulnerability described
in the AST-2008-006 security advisory.  1.6.0-beta8 is also a regular update to
the 1.6.0 series with a number of bug fixes over the previous beta release.

Early last year, we made some modifications to the IAX2 channel driver to
combat
potential usage of IAX2 in traffic amplification attacks.  Unfortunately, our
fix was not complete and we were not notified of this until the original
reporter of the issue decided to release information on how to exploit it to
the
public.

This issue affects all users of IAX2 that have allowed non-authenticated calls.
 For more information on the vulnerability, see the published security
advisory.

 * http://downloads.digium.com/pub/security/AST-2008-006.pdf

All releases are available for download from the following location:

 * http://downloads.digium.com/pub/telephony/asterisk/

Thank you for your continued support of Asterisk!





Javantea originally reported an issue in IAX2, whereby an attacker could send a
spoofed IAX2 NEW message, and Asterisk would start sending early audio to the
target address, without ever receiving an initial response. That original
vulnerability was addressed in June 2007, by requiring a response to the
initial NEW message before starting to send any audio.


Javantea subsequently found that we were doing insufficent verification of the
ACK response and that the ACK response could be spoofed, just like the initial
NEW message. We have addressed this failure with two changes. First, we have
started to require that the ACK response contains the unique source call number
that we send in our reply to the NEW message. Any ACK response that does not
contain the source call number that we have created will be silently thrown
away. Second, we have made the generation of our source call number a little
more difficult to predict, by randomly selecting a source call number, instead
of allocating them sequentially.

fixed in voip overlay for versions 1.4.19.1 and 1.6.0-beta8. 

CVE-2008-1923 was assigned to the original "NEW" issue in June 2007.

(In reply to comment #2)
> CVE-2008-1923 was assigned to the original "NEW" issue in June 2007.

This was released with 1.2.20.

voip, any news here?

This is only first security bug report from 7 others opened.
Somebody has either mask asterisk stable packages in the portage or fix them
all.

The way how it is now doesn't make sense for me.

+*asterisk-1.2.31.1 (11 Mar 2009)
+
+  11 Mar 2009; <chainsaw@gentoo.org>
+  +files/1.2.0/asterisk-1.2.31.1-bri-fixups.diff,
+  +files/1.2.0/asterisk-1.2.31.1-comma-is-not-pipe.diff,
+  +files/1.2.0/asterisk-1.2.31.1-svn89254.diff, +asterisk-1.2.31.1.ebuild:
+  Version bump, for security bugs #250748 and #254304. Took a 1.4 build fix
+  that is relevant to 1.2, Digium bug #11238. Wrote patch to fix up typo in
+  open call, a comma is not a pipe sign. Used EAPI 2 for USE-based
+  dependencies instead of calling die. Patch from Mounir Lamouri adding
+  -lspeexdsp closes bug #206463 filed by John Read.

Stabling via bug 250748

GLSA 200905-01