From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH 00/21] NET: use unique MDIO bus names Date: Tue, 10 Jan 2012 16:35:24 -0800 (PST) Message-ID: <20120110.163524.906859005643528607.davem@davemloft.net> References: <1326189567-21663-1-git-send-email-florian@openwrt.org> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org To: florian@openwrt.org Return-path: Received: from shards.monkeyblade.net ([198.137.202.13]:48145 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756667Ab2AKAf2 (ORCPT ); Tue, 10 Jan 2012 19:35:28 -0500 In-Reply-To: <1326189567-21663-1-git-send-email-florian@openwrt.org> Sender: netdev-owner@vger.kernel.org List-ID: From: Florian Fainelli Date: Tue, 10 Jan 2012 10:59:06 +0100 > While troubleshooting an issue with an user of the RDC r6040 driver > having compiled support for the fixed MDIO bus, I noticed that quite > some network drivers will have a MDIO bus named "0" most of the time. > > Some network drivers already try to provide an unique MDIO bus name > using their PCI slot and function, this is okay. > > Some other network drivers will format the name using their platform_device > id, which will usually result in a collision with the fixed MDIO bus > name ("0"), these drivers are fixed to use the following pattern instead: > > - which is unique > > Finally the MDIO bus drivers themselves are patched to use an unique name > as well. > > There is a potential to break user-space applications assuming there is a > specific MDIO bus named "0", "1" or other non-explanatory names like this. > > Note that I compile tested most of the drivers on x86 and MIPS, ARM and Blackfin > drivers have not been compile tested. All applied, thanks Florian.