Re: 802.1Q vlan performance. [PATCH]

From: Robert Watson <rwatson_at_FreeBSD.org>
Date: Sat, 26 Aug 2006 05:52:17 +0100 (BST)
On Fri, 25 Aug 2006, Gleb Smirnoff wrote:

> On Fri, Aug 25, 2006 at 10:31:33AM -0400, John Baldwin wrote:
> J> > I> +++ /usr/src/sys/conf/NOTES       25 Aug 2006 10:03:18 -0000
> J> > I> _at__at_ -646,6 +646,7 _at__at_
> J> > I>  #
> J> > I>  device         ether                   #Generic Ethernet
> J> > I>  device         vlan                    #VLAN support (needs miibus)
> J> > I> +options                VLAN_ARRAY              #Lookup table for trunk interface
> J> > I>  device         wlan                    #802.11 support
> J> > I>  device         wlan_wep                #802.11 WEP support
> J> > I>  device         wlan_ccmp               #802.11 CCMP support
> J> >
> J> > I'd prefer not to put it into NOTES. When enabled, less code will
> J> > be covered by LINT.
> J>
> J> Bah, GENERIC will cover it then.  All options should be in NOTES.  If
> J> an option removes code and it's enabled in GENERIC, then you should put
> J> it in NOTES but comment it out.
>
> But "device vlan" isn't in GENERIC.

Only for historical reasons -- at the recent devsummit, I argued that if_vlan 
should, in fact, be part of our generic ethernet support code, as vlans have 
become an expected ethernet link layer feature.  I also argued that tigher 
integration of vlan support into ethernet would help avoid some of the 
trickier issues in keeping vlan interfaces synchronized to the set of ethernet 
interfaces, as well as avoiding the mbuf tag overhead, etc.  I have some hope 
of looking into this once if_startmbuf is off the ground.

Robert N M Watson
Computer Laboratory
University of Cambridge
Received on Sat Aug 26 2006 - 02:52:18 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:59 UTC