From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tim Gardner Subject: Re: Net driver module dependencies on PHY driver modules Date: Sat, 31 Oct 2009 09:34:13 -0600 Message-ID: <4AEC58F5.2030905@canonical.com> References: <1256989326.3136.284.camel@localhost> Reply-To: tim.gardner@canonical.com Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: Matt Carlson , Michael Chan , netdev , 553024@bugs.debian.org To: Ben Hutchings Return-path: Received: from mail.tpi.com ([70.99.223.143]:1519 "EHLO mail.tpi.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752766AbZJaPdZ (ORCPT ); Sat, 31 Oct 2009 11:33:25 -0400 In-Reply-To: <1256989326.3136.284.camel@localhost> Sender: netdev-owner@vger.kernel.org List-ID: Ben Hutchings wrote: > I'm handling a bug report relating to > tg3 using phylib and missing the broadcom module. > > There seem to be two general problems here: > 1. PHY drivers are not auto-loaded > 2. An initramfs builder cannot statically determine that a net device > requires a PHY driver > > It seems to me that PHY drivers should have aliases based on MDIO PHY > ids, and phylib should request_module() if no suitable driver is > registered. This would solve (1) and go some way to solving (2). > > Ben. > I encountered the same issue with Karmic and worked around it by building in all of the phylib modules, e.g., CONFIG_BROADCOM_PHY=y. I agree that its not the best long term solution. rtg -- Tim Gardner tim.gardner@canonical.com