From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rob Herring Subject: Re: [PATCH v2 0/2] iio: add unit converter Date: Tue, 10 Apr 2018 08:06:52 -0500 Message-ID: <20180410130652.d3kk3vo6p4524mid@rob-hp-laptop> References: <20180403153635.8228-1-peda@axentia.se> <93a6f7fc-4a53-c65c-422c-1649732fed50@ti.com> <5c18fa2a-0b10-5365-4868-ead6b61b45b2@axentia.se> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <5c18fa2a-0b10-5365-4868-ead6b61b45b2@axentia.se> Sender: linux-kernel-owner@vger.kernel.org To: Peter Rosin Cc: "Andrew F. Davis" , linux-kernel@vger.kernel.org, Jonathan Cameron , Hartmut Knaack , Lars-Peter Clausen , Peter Meerwald-Stadler , Mark Rutland , "David S. Miller" , Mauro Carvalho Chehab , Greg Kroah-Hartman , Linus Walleij , Andrew Morton , Randy Dunlap , Michael Hennerich , Phil Reid , linux-iio@vger.kernel.org, devicetree@vger.kernel.org List-Id: devicetree@vger.kernel.org On Tue, Apr 03, 2018 at 08:09:08PM +0200, Peter Rosin wrote: > On 2018-04-03 19:41, Andrew F. Davis wrote: > > On 04/03/2018 10:36 AM, Peter Rosin wrote: > >> Hi! > >> > >> This driver implements support for voltage dividers and current > >> sense circuits. It's pretty generic and should be easily adaptable > >> to other linear scaling purposes... > >> > > > > I really like this idea, defining channel scaling / channel type > > conversion in DT will be very useful. So much so that I would recommend > > this not be a use specific driver but instead moved into the IIO core. > > > > This would allow defining these channel conversions in the device node > > itself, so as to not need a separate node just for the converter (the > > conversion is not a device and probably should not have it's own node > > anyway). It would also help with enabling this support for buffered > > readers/writers and not just devices only supporting _raw reads/writes. > > In my case, the voltage dividers and the current sense circuit are very > much real, and I can point my finger at them on the board. Sure, they > are not ICs, but to not call them devices is wrong IMHO, and the ADC > which is involved have very little to do with the voltage dividers and > the fact that it is involved in sensing current. This is not an argument > for not moving the functionally to the core though, but that has some > problems AFAICT. Because you need some kind of clever and generic > binding to make the core do its thing, and it might not be easy to come > up with something that fits all devices? And if we do put this in the > core, that opens the door for more complex unit converters later on > (non-linear etc), further complicating the generic bindings. I agree. We've learned the hard way that even things like the LED diode itself need to be described and not just the controller. Rob