Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 2 Oct 2003 16:17:38 -0700
From:      Brooks Davis <brooks@one-eyed-alien.net>
To:        Julian Elischer <julian@elischer.org>
Cc:        net@freebsd.org
Subject:   Re: ng_ip_input use case
Message-ID:  <20031002231738.GB4770@Odin.AC.HMC.Edu>
In-Reply-To: <Pine.BSF.4.21.0310021608390.9220-100000@InterJet.elischer.org>
References:  <20031002230508.GA4770@Odin.AC.HMC.Edu> <Pine.BSF.4.21.0310021608390.9220-100000@InterJet.elischer.org>

next in thread | previous in thread | raw e-mail | index | archive | help

--jq0ap7NbKX2Kqbes
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Oct 02, 2003 at 04:09:50PM -0700, Julian Elischer wrote:
>=20
>=20
> On Thu, 2 Oct 2003, Brooks Davis wrote:
>=20
> > On Thu, Oct 02, 2003 at 03:45:47PM -0700, Julian Elischer wrote:
> > > On Thu, 2 Oct 2003, Vincent Jardin wrote:
> > >=20
> > > > I read the man of ng_ip_input and the source code. However I do not
> > > > understand what the use cases could be. Who is using it and what's
> > > > for ?
> > >
> > > I have no idea what it is used for.. similar effect can be made by
> > > using the ng_iface node, but I think that they didn't want a separate
> > > interface for each packet source..
> > >
> > > I suggest we ask brooks..(cc'd).  I could imagine it somehow connected
> > > with the 300 processor FreeBSD based cluster that he is working on at
> > > hos job (especially as it has their copyright).
> >=20
> > It was for a network emulator we were trying to build to allow us to
> > run real, unmodified programs in an environment where each program
> > was assumed to be an independent agent and the communication topology
> > between agents changed in realtime.  Each agent was to bind to the IP
> > address of an IP-over-IP tunnel (gif on the FreeBSD end).  We would then
> > take the packets, mangle them slightly and send them off to the emulator
> > (I think part of the goal was to let us use more or less off the shelf
> > emulation gear in the middle without needing a workstation for each
> > agent).  When we got them back, we'd remanged them and stuff them back
> > into the IP stack with ng_ip_input so it could decide which gif tunnel
> > to send it back down.  The emulation part of the project died when our
> > funding for it dried up so we've not actually using this module for
> > anything.
>=20
> was there a reason to not use the ng_iface node?
> (packets enterred on the INET hook if an iface node will be injected
> into the ip stack) (from memory)

Probably not.  I didn't know netgraph all that well at the time so it
wouldn't suprise me if I reinvented a wheel or two.

-- Brooks

--=20
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4

--jq0ap7NbKX2Kqbes
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE/fLIQXY6L6fI4GtQRAiVcAJ4l7QdprW2PEXcev9LFGUBk8+oZjgCfbK9z
cZD9bi3PFDvP+C7h3rslLXE=
=SaAZ
-----END PGP SIGNATURE-----

--jq0ap7NbKX2Kqbes--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20031002231738.GB4770>