Re: [ GLSA 200512-04 ] Openswan, IPsec-Tools: Vulnerabilities in

[Paul Wouters]

On Mon, 12 Dec 2005, Thierry Carrez wrote:

> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> Gentoo Linux Security Advisory GLSA 200512-04
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> http://security.gentoo.org/
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>
> Severity: Normal
> Title: Openswan, IPsec-Tools: Vulnerabilities in ISAKMP Protocol
> implementation
> Date: December 12, 2005
> Bugs: #112568, #113201
> ID: 200512-04

> Openswan and IPsec-Tools suffer from an implementation flaw which may
> allow a Denial of Service attack.

That is correct (for openswan)

> Impact
> ======
>
> A remote attacker can create a specially crafted packet using 3DES with
> an invalid key length, resulting in a Denial of Service attack, format
> string vulnerabilities or buffer overflows.

That's a copy and paste from the IPsec proto testsuite.

1) It conflicts with the above comment that this is only a DOS
2) It's incorrect (for openswan)

> Workaround
> ==========
>
> Avoid using "aggressive mode" in ISAKMP Phase 1, which exchanges
> information between the sides before there is a secure channel.

In fact, you would to both have aggressive mode enabled AND know the PSK.
If you have those two enabled, you are vulnerable to a MITM anyway, since
any client knowing the PSK can pretend to be the IPsec security gateway.

Paul

[Thierry Carrez]

This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enigA204690C7A6963B7962AB27F
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

VANHULLEBUS Yvan wrote:

> On Tue, Dec 13, 2005 at 09:49:40PM +0100, Paul Wouters wrote:
>
>>On Mon, 12 Dec 2005, Thierry Carrez wrote:
>>
>> [...]
>>>Impact
>>>======
>>>
>>>A remote attacker can create a specially crafted packet using 3DES with
>>>an invalid key length, resulting in a Denial of Service attack, format
>>>string vulnerabilities or buffer overflows.
>>
>>That's a copy and paste from the IPsec proto testsuite.
>>
>>1) It conflicts with the above comment that this is only a DOS
>>2) It's incorrect (for openswan)
>
> Also incorrect for ipsec-tools AFAIK. The only problem we noticed with
> protos testsuite was a lack of verification for some payloads
> existency in aggressive mode.
>
>>>Workaround
>>>==========
>>>
>>>Avoid using "aggressive mode" in ISAKMP Phase 1, which exchanges
>>>information between the sides before there is a secure channel.
>>
>>In fact, you would to both have aggressive mode enabled AND know the PSK.
>>If you have those two enabled, you are vulnerable to a MITM anyway, since
>>any client knowing the PSK can pretend to be the IPsec security gateway.
>
>
> Knowing the PSK is not really needed, as AGGRESSIVE+PSK mode is known
> to be quite unsecure, and can be bruteforced offline.
>
> The "workaround" for ipsec-tools is to upgrade, and is only needed for
> some people which really have a week configuration and should care
> about lots of potential problems !

Paul, Yvan,

Many thanks for the precisions and sorry for getting it wrong. I'll fix
the GLSA text for Impact and Workaround on the Gentoo website asap.

--
Thierry Carrez (Koon)
Gentoo Linux Security

--------------enigA204690C7A6963B7962AB27F
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDoAwlvcL1obalX08RAtFeAJsFAKIHZYGvMatHsDdnlm QQ8dSpVwCfQDQA
Xobb2+IVec0SgSifjnlv7t4=
=ff+t
-----END PGP SIGNATURE-----

--------------enigA204690C7A6963B7962AB27F--