From mboxrd@z Thu Jan 1 00:00:00 1970 From: Grant Likely Subject: Re: [PATCH] Input: mpu3050: add of_match table for device-tree probing Date: Mon, 2 Jan 2012 01:42:43 -0700 Message-ID: <20120102084243.GH18381@ponder.secretlab.ca> References: <1324607992-2919-1-git-send-email-olof@lixom.net> <20111223092246.GA7120@core.coreip.homeip.net> <4EF9ED4E.6000408@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-input-owner@vger.kernel.org To: Olof Johansson Cc: Rob Herring , devicetree-discuss@lists.ozlabs.org, Dmitry Torokhov , linux-input@vger.kernel.org List-Id: devicetree@vger.kernel.org On Tue, Dec 27, 2011 at 09:14:31AM -0800, Olof Johansson wrote: > On Tue, Dec 27, 2011 at 8:07 AM, Rob Herring wrote: > > On 12/23/2011 09:58 AM, Olof Johansson wrote: > >> Hi, > >> > >> On Fri, Dec 23, 2011 at 1:22 AM, Dmitry Torokhov > >> wrote: > >>> Hi Olof, > >>> > >>> On Thu, Dec 22, 2011 at 06:39:52PM -0800, Olof Johansson wrote: > >>>> Adding invn,mpu3050 as the initial id. > >>>> > >>> > >>> I believe you also need to add this to > >>> Documentation/devicetree/bindings/input/ > >> > >> For simple i2c devices that only have a name, address and possibly an > >> interrupt, there's no need to document a binding (it hasn't been done > >> in the past for the vast majority of those devices). > > > > But then how do you find out if there is a existing binding for a device > > without searching in the kernel source tree? > > That's a silly question, since today that is exactly what you do to > find out, since the Documentation/devicetree isn't the canonical > location for _all_ bindings anyway -- some are in ePAPR, some are in > the old IEEE1275 docs, and some are based on whatever the vendor that > first introduced the device chose (Apple, and some other vendors, > mostly PowerPC ones). I would like to keep a list of new bindings though, but I agree with the argument that a lot of bindings don't need a separate file with a whole bunch of meaningless boilerplate. g.