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
signature.asc
Description: OpenPGP digital signature
