Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 01 Jan 1998 13:25:32 +0000
From:      Brian Somers <brian@awfulhak.org>
To:        John-Mark Gurney <gurney_j@efn.org>
Cc:        freebsd-bugs@hub.freebsd.org, freebsd-hackers@FreeBSD.ORG, joerg@FreeBSD.ORG
Subject:   Re: kern/5404: slXX slip (tun & ppp) interfaces always point to point 
Message-ID:  <199801011325.NAA17803@awfulhak.demon.co.uk>
In-Reply-To: Your message of "Wed, 31 Dec 1997 17:30:01 PST." <199801010130.RAA10049@hub.freebsd.org> 

next in thread | previous in thread | raw e-mail | index | archive | help
[cc'd to joerg and freebsd-hackers]

> The following reply was made to PR kern/5404; it has been noted by GNATS.
> 
> From: John-Mark Gurney <gurney_j@efn.org>
> To: Matthew Dillon <dillon@backplane.com>
> Cc: FreeBSD-gnats-submit@FreeBSD.ORG
> Subject: Re: kern/5404: slXX slip interfaces always point to point
> Date: Wed, 31 Dec 1997 17:23:58 +875200
> 
[.....]
>  or are you wanting to do something like:
>  
>  	+-------------+       +-------------+
>  	|  machine a  |-------|  machine b  |---- rest of 192.168.3.x
>  	| 192.168.3.2 |       | 192.168.3.1 |
>  	+-------------+       +-------------+
>  
>  and trying to get machine a on the same network as machine b?
>  
>  if your going for the second one, you need proxyarp on machine b...
>  I've actually done something similar to this before...  actually my
>  terminal server is setup like that... the ethernet is 192.168.0.x.. and
>  the ip's for the dialin's are 192.168.40-60 or so...
[.....]

This ties in with a recent discussion on -hackers:

> Brian Somers <brian@awfulhak.org> wrote:
> 
> > I agree, and I'll implement the change unless someone has a good 
> > reason not to..... any takers ?
> 
> I think it's really best to just not display the netmask in the output
> of ifconfig iff IFF_POINTOPOINT is set.
> 
> Routes to the remote end apart from the implied host route seem to be
> dangerous to me, and they break the current behaviour (i.e. could
> cause surprises for people who are used to how it's done now).  It's
> not always that the IP address of the remote end is indeed identical
> with the remote network address.
> 
> -- 
> cheers, J"org

I intended to remove the possibility of netmasks and broadcast 
addresses for pointopoint links, but retrospectively, this will break 
the ability to attach to a peer that proxy arps for you (well, it's 
already broken).

I would suggest (and I'm willing to do it) adding the ability to use 
SIOCSIFPOINTOPOINT and SIOCGIFPOINTOPOINT on sl* ppp* and tun*.  The 
default is that these interfaces are pointopoint, but you can change 
that by issuing the `S' ioctl.  Additionally, I'd change ifconfig so 
that it doesn't display a netmask & broadcast address when the 
POINTOPOINT flag is set.

I can then implement the ability to make these interfaces 
non-pointopoint in their respective programs - if configured that 
way.

None of the existing code would break as the defaults are the same.

Comments ?

-- 
Brian <brian@Awfulhak.org>, <brian@FreeBSD.org>, <brian@OpenBSD.org>
      <http://www.Awfulhak.org>;
Don't _EVER_ lose your sense of humour....







Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199801011325.NAA17803>