From mboxrd@z Thu Jan 1 00:00:00 1970 From: Florian Fainelli Subject: Re: [RFC PATCH 4/4] net: mvmdio: add getter and setter for PHY addresses Date: Fri, 25 Jan 2013 20:58:51 +0100 Message-ID: <5102E3FB.7030808@openwrt.org> References: <1359108409-4378-1-git-send-email-florian@openwrt.org> <1359108409-4378-5-git-send-email-florian@openwrt.org> <20130125181629.GA7393@obsidianresearch.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: thomas.petazzoni@free-electrons.com, andrew@lunn.ch, jason@lakedaemon.net, arnd@arndb.de, netdev@vger.kernel.org, gregory.clement@free-electrons.com, ian.molton@codethink.co.uk, linux-arm-kernel@lists.infradead.org To: Jason Gunthorpe Return-path: Received: from mail-bk0-f49.google.com ([209.85.214.49]:38140 "EHLO mail-bk0-f49.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751074Ab3AYT66 (ORCPT ); Fri, 25 Jan 2013 14:58:58 -0500 Received: by mail-bk0-f49.google.com with SMTP id w11so482745bku.8 for ; Fri, 25 Jan 2013 11:58:56 -0800 (PST) In-Reply-To: <20130125181629.GA7393@obsidianresearch.com> Sender: netdev-owner@vger.kernel.org List-ID: Le 25/01/2013 19:16, Jason Gunthorpe a =E9crit : > On Fri, Jan 25, 2013 at 11:06:49AM +0100, Florian Fainelli wrote: >> This patch adds orion_mdio_{set,get}_phy_addr(.., port_number, phy_a= ddr) >> which is a feature available in this MDIO driver to monitor a partic= ular >> PHY address. This brings mvmdio one step closer to what is used in >> mv643x_eth. > > This seems really strange. > > Are you sure this should be part of the MDIO driver? Based on my docs= , > this register is part of the ethernet controller HW, which > autonomously reads from the phy via MDIO. My reading of the datasheets for 88F628x and Armada 370 says that it=20 actually belongs to the MDIO part of the controller (it is just below=20 the SMI registers). > > It seems cleaner to set this register from the ethernet controller > based on the phy it is using and keep the MDIO driver purely for MDIO > bus access. > > It is acceptable to overlap the device address ranges, start the > ethernet controller at 72000 and start the MDIO at 72004, the platfor= m > bus code automatically nests them. Ok I did not think this would work, but I kind of prefer that too. -- =46lorian