Subject: Re: possible kernel NAT-T bug
To: Emmanuel Dreyfus <manu@netbsd.org>
From: Martin Husemann <martin@duskware.de>
List: tech-net
Date: 12/07/2005 08:56:36
On Wed, Dec 07, 2005 at 07:52:14AM +0000, Emmanuel Dreyfus wrote:
> I guess they could, or at least they could pass a struct mbuf **,
> but doesn't that propagate the problem? I guess ip_input() does not
> expect udp_input() to change the mbuf either.

Actually the relay via the protocols input routine is the last
thing it does with the mbuf:

        (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);  
        return;
 

Martin