Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 21 Jan 2005 11:50:50 -0800
From:      Brooks Davis <brooks@one-eyed-alien.net>
To:        Boris Kovalenko <boris@ntmk.ru>
Cc:        freebsd-net@freebsd.org
Subject:   Re: [PATCH] 802.1p priority (fixed)
Message-ID:  <20050121195050.GA2866@odin.ac.hmc.edu>
In-Reply-To: <41F0A457.5010304@ntmk.ru>
References:  <41EF9495.5080601@ntmk.ru> <20050120190516.GA12156@odin.ac.hmc.edu> <41F07622.5040102@ntmk.ru> <20050121054732.GA30766@odin.ac.hmc.edu> <41F0A457.5010304@ntmk.ru>

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

--J/dobhs11T7y2rNN
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, Jan 21, 2005 at 11:42:31AM +0500, Boris Kovalenko wrote:
> Brooks Davis wrote:
> Hello!
>=20
> >>Yes, the outgoing packets are tagged with specified priority. Then next=
=20
> >>device (Cisco Catalyst for example) will assign traffic to different=20
> >>queues according to 802.1p header information. The only thing (IMHO)=20
> >>that may be coded for FreeBSD is to allow PF & IPFW assign packets to=
=20
> >>ALTQ or DUMMYNET according 802.1p information.
>=20
> >
> >OK, that makes sense.  Hmm, do we actually want to be using seperate
> >interfaces for this?  I'm sure it's very useful in some applications,
> >but if the real point is to get packets on the wire with the priority
> >tags, won't IPFW, PF, or maybe even the application be the best place
> >for this tagging rather then effectivly using the source address to set
> >it?  Again, I'm not familiar with the way 802.1p is intended to work, so
> >this may be a dumb question.
> By the usual way, application does not have access to Layer 2 headers,=20
> so it can not set 802.1p priority itself. It may only set ToS value, but=
=20
> Layer 2 switches can not access Layer 3 information :) Indeed I'm not=20
> familar with BSD network structure interoperability. Andre Oppermann=20
> said that there is a way to mark this packets with m_tag from PF/IPFW.=20
> So, if this is really possible, the best way (IMHO) should be: if=20
> packet, that going out the vlan iterface has m_tag with 802.1p, we use=20
> this value, or value provided for vlan instead. This is just the way=20
> Cisco Catalyst does: trust the received 802.1p inforamtion, or override i=
t.

My letting the application handle it, I was thinking of adding a socket
option (possibly requiring privilege) to set the priority.

As to having PF/IPFW deal with it, I was thinking about two
modifications to IPFW.  First, the ability to filter based on .1p tags.
You might want to peal that information off in the Ethernet
code and tag the packets so you could still inspect it at a higher
level, but maybe not.  The second modification would be to give ipfw/pf
the ability to set .1p priorities on packets, ie:

ipfw add 802.1p 6 on any to any ssh

My concern is that 802.1p is like the TOS bits in that it differentiates
packets within a network rather then segregating them in to networks
like 802.1Q.  In a switch it makes sense to handle priorities as separate
networks, but I'm not sure it makes sense in a host.  If nothing else,
it seems to make sense to be able to set priorities on vlan encapsulated
frames.

I've done a little googling on 802.1p and that hardened my believe that
the application and packet filter are the places to deal with this.  I'm
downloading the standard now to take a look at it.

-- 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

--J/dobhs11T7y2rNN
Content-Type: application/pgp-signature
Content-Disposition: inline

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

iD8DBQFB8V0ZXY6L6fI4GtQRAl3FAJ9+vzn1FbfuEW6FdSxxNKsTNI62VACgm5+a
OuFFgVXwOr8YfWtWQdkuPoA=
=lz7Y
-----END PGP SIGNATURE-----

--J/dobhs11T7y2rNN--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050121195050.GA2866>