bugtraq - Re: Cisco ACL bug when using VPN crypto engine accelerator, PPPoE dialer or ip route-cache

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <Pine.GSO.4.53.0305151853370.1898@bru-cse-128.cisco.com>
Date: Thu, 15 May 2003 18:56:36 +0200 (CEST)
From: Ilker Temir <itemir@...co.com>
To: Olivier <itsce.networkservices@...ntl.ch>
Subject: Re: Cisco ACL bug when using VPN crypto engine accelerator, PPPoE 
   dialer or ip route-cache

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

This is in response to the e-mail sent by Olivier. The original e-mail is
available at
http://www.securityfocus.com/archive/1/321552/2003-05-12/2003-05-18/0

Hi Olivier,

We can confirm that the inbound ACL is processed twice for IPSec traffic
which is the root cause of the issue you have described.

This issue is being addressed by the Cisco Bug ID CSCdz54626. Cisco's
Product Security Incident Response Team (PSIRT) was not previously aware
of it. We are now addressing it with due priority.

While this issue does cause administrative overhead to configuration,
there is no significant security impact.

Permitting internal networks in the inbound ACL may be exploited to inject
spoofed packets into the network. However this has no practical impact
while using static crypto maps. With static crypto maps, the unencrypted
traffic will be dropped even if it passes the inbound ACL.

In the case of dynamic crypto maps, in order to bypass the inbound ACL to
inject spoofed packets, an attacker would also require control of the
neighboring routers that are connected to the interface where the inbound
ACL is applied, or the medium in between the neighbors.

Protecting our customers' networks is very important for us and we are
always open for vulnerability reports regarding any Cisco product.
Such reports should be directly sent to psirt@...co.com or
security-alert@...co.com.

Thank you again for your report,

Regards,

- --
Ilker Temir
Incident Manager, PSIRT
Cisco Systems, Inc.
+32 2 704-6031
http://www.cisco.com/go/psirt

On Wed, 14 May 2003, Olivier wrote:

>
>
> Platform Cisco 1760 dual Ethernet
>
> IOS 12.2.xT IP/ADSL/FW/IDS PLUS IPSEC 3DES
>
> Environment: Site to site VPN for small offices.
>
>
>
> ACL are not properly parsed as soon as you enable:
>
> crypto engine accelerator
> PPPoE dialer
> Ip route-cache
>
>
> Without the feature mentioned above, you can apply an ACL on the outside
> interface allowing only inbound ISAKMP and IPSEC traffic.
>
> I.E.
>
> ip access-list extended Block-Inbound-unwanted-Trafic
>
>  permit udp 100.100.100.0 0.0.0.255 host 102.168.1.2 eq isakmp
>
>  permit esp 100.100. 100.0 0.0.0.255 host 102.168.1.2
>
>  deny   ip any any log
>
>
>
> If you activate the crypto engine, the ACL is parsed as well on decrypted
> traffic which forces you to allow as well all traffic for the decrypted
> traffic.
> I.E. If you are using 10.x addressees internally and the subnet
> 10.200.0.0/24 for your Soho LAN. Can be worst if you have a huge network
> inside where you would prefer to add permit ip  any 10.200.0.0 0.0.0.255.
>
>
> ip access-list extended Block-Inbound-unwanted-Trafic
>  permit udp 100.100.100.0 0.0.0.255 host 102.168.1.2 eq isakmp
>  permit esp 100.100. 100.0 0.0.0.255 host 102.168.1.2
>  permit ip  10.0.0.0 0.255.255.255 10.200.0.0 0.0.0.255  <-----------@...%@
>  deny   ip any any log
>
>
> This looks pretty bad for a VPN box running a Firewall feature set IOS
> seen as the best candidate for VPN for small offices.
>
> The worst is the reply from Cisco:
> -------------------------------------------------------------------
> We will be addressing this in the next few months however
> the release time frame could be as late as the end
> of the year.
>
> We do have plans to address it but do
> not expect it in a released image until the
> last calendar quarter of the year. If its possible we
> can get it done and released sooner than what I've
> mentioned, we will do it, no guarantees however.
> -------------------------------------------------------------------
>
> We would have hope that they put more resources and concern in solving
> security issue.
>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (SunOS)

iD8DBQE+w8bR8/wE0ppYtwURAunzAJ4oUlepUBjdJzQ1jZfBQGNI3UNNkwCcCcRh
CJBxZPAMkMO9/PwFxVcibTs=
=4dJU
-----END PGP SIGNATURE-----