Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 23 Feb 2010 15:49:42 +0300
From:      Denis Antrushin <DAntrushin@mail.ru>
To:        VANHULLEBUS Yvan <vanhu@freebsd.org>
Cc:        freebsd-net@freebsd.org
Subject:   Re: IPSec connection troubles
Message-ID:  <4B83CEE6.9040409@mail.ru>
In-Reply-To: <20100223122127.GA45649@zeninc.net>
References:  <4B73E902.6050301@mail.ru> <20100211124756.GA9528@zeninc.net> <20100211125420.G27327@maildrop.int.zabbadoz.net> <4B83B79F.102@mail.ru> <20100223122127.GA45649@zeninc.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On 02/23/10 15:21, VANHULLEBUS Yvan wrote:
> On Tue, Feb 23, 2010 at 02:10:23PM +0300, Denis Antrushin wrote:
> [...]
>> ipsec-tools understand NAT-OA payload in IKE exchange, but then simply
>> discard it and do not send this information to kernel.
>> In ipsec-tool mailing list archives I found mention that linux does not
>> need this OA info, because it simply recomputes/ignore TCP checksums.
>
> Userland part is the most simple to do, as PFKey extension for NAT-OA
> already exists, it haven't been done so far because it's useless until
> someone does the big part of the kob on a kernel...

Taking into account this quote:

On 02/11/10 15:55, Bjoern A. Zeeb wrote:
 > Him saying it works on linux - has ipsec-tools grown proper OA support
 > these days? If that would be the case the kernel would probably a
 > minor task.

this means that I have to come up with patches for both FreeBSD kernel
and racoon at the same time. :-)
May I contact you off-list with patches for both, when ready?
As far as I understand, you are the one who can review both.

>> Can we do the same or this is unacceptable for FreeBSD and we want
>> NAT-OA communicated to kernel by IKEd?
>> I made a simple patch to ipsec_common_input_cb() to ignore TCP/UDP
>> checksums of ESP-protected packets and I happily can connect to
>> Solaris VPN server from behind the NAT device (after working around
>> some security policy matching issues).
>
> Just adding some code to always ignore such checksums sounds like a
> bad idea for me.....
>
> But maybe we could have at least a sysctl (disabled by default) to
> ignore them.....
>
> Yvan.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4B83CEE6.9040409>