From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jon Masters Subject: Re: network interface *name* alias support? Date: Fri, 23 May 2008 10:50:43 -0400 Message-ID: <1211554243.11907.4.camel@perihelion> References: <1211410078.27149.70.camel@perihelion> <1211548070.21380.500.camel@pmac.infradead.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: netdev To: David Woodhouse Return-path: Received: from dallas.jonmasters.org ([72.29.103.172]:53253 "EHLO dallas.jonmasters.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751327AbYEWOuq (ORCPT ); Fri, 23 May 2008 10:50:46 -0400 In-Reply-To: <1211548070.21380.500.camel@pmac.infradead.org> Sender: netdev-owner@vger.kernel.org List-ID: On Fri, 2008-05-23 at 14:07 +0100, David Woodhouse wrote: > On Wed, 2008-05-21 at 18:47 -0400, Jon Masters wrote: > > > > I would like to add support for network interface name aliases to the > > kernel. For example, rather than simply "eth0", one might have: > > > > eth0 -> eth_slot_0_0 > > > > And use either name for device configuration, etc. This should be pretty > > easy, but I would like some comments - in particular, has this been done > > already and I'm missing something? > > > > Why is this useful? Say, for example that your a manufacturer building > > machines using the latest DMI extension to SMBIOS that allows you to > > specify what the physical ordering of interfaces is. Then, a > > distribution might choose to configure based on the physical slot > > ordering specified by the vendor - allowing for "rip and replace". > > Why do you need a given interface to have more than one name? Why not > just assign names according to whatever criteria you care about? > That's entirely a udev problem, surely? Well, for various reasons, we have folks who want to always retain the existing "legacy" naming to "avoid confusion". Yeah, personally I don't really think it matters...but apparently it does, so I'm happy to oblige and have udev set an alias according to physical slot position aswell. Jon.