From mboxrd@z Thu Jan 1 00:00:00 1970 From: Vivien Didelot Subject: Re: [PATCH RFC v2 00/32] Make DSA switches linux devices. Date: Sun, 13 Mar 2016 03:26:07 -0400 Message-ID: <87a8m2ltyo.fsf@ketchup.mtl.sfl> References: <1456677700-23027-1-git-send-email-andrew@lunn.ch> <56D88757.7050803@gmail.com> <20160303202716.GQ15541@lunn.ch> <56E357C7.7090309@gmail.com> <20160312170827.GA2186@lunn.ch> Mime-Version: 1.0 Content-Type: text/plain Cc: netdev To: Andrew Lunn , Florian Fainelli Return-path: Received: from mail.savoirfairelinux.com ([208.88.110.44]:34982 "EHLO mail.savoirfairelinux.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750769AbcCMH0O (ORCPT ); Sun, 13 Mar 2016 03:26:14 -0400 In-Reply-To: <20160312170827.GA2186@lunn.ch> Sender: netdev-owner@vger.kernel.org List-ID: Hi Andrew, Florian, Andrew Lunn writes: > It probably means we need to turn slave.c and parts of dsa.c into a > library. Add a new dsa_v2.c file, containing the new binding > code. With things like having the switch devices instantiate there own > MDIO bus, fixed phys on a compatible MDIO bus referenced via phandles, > etc, I hope we can make parts of the dsa_v2 simpler. Nothing critical here, but as we are talking about better isolation of the DSA slave code, wouldn't it be the good moment to make it a dsa_switch driver, instead of a library? This code is actually pretty generic, and for example home routers could register an instance of such module. Thanks, -v