From mboxrd@z Thu Jan 1 00:00:00 1970 From: Guenter Roeck Subject: Re: Problem with multiple i2c multiplexers on the same bus Date: Fri, 31 Oct 2014 14:17:58 -0700 Message-ID: <5453FC86.1080408@roeck-us.net> References: <20141031210301.GA4169@katana> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20141031210301.GA4169@katana> Sender: linux-i2c-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Wolfram Sang , Martin Belanger Cc: linux-i2c-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Jean Delvare List-Id: linux-i2c@vger.kernel.org On 10/31/2014 02:03 PM, Wolfram Sang wrote: > Hi, > > On Mon, Oct 27, 2014 at 11:46:01AM -0700, Martin Belanger wrote: >> This is regarding a series of emails between Guenter Roeck and Jean >> Delvare titled "Problem with multiple i2c multiplexers on one bus, and >> mux bus naming" sent in November 2013. Ref: >> http://thread.gmane.org/gmane.linux.drivers.i2c/16980 > > Please CC those people then, too. That helps getting their attention. > I've done this now. > >> I'm having the same problem with multiple PCA954x multiplexers on the >> same bus and there is no way to tell them apart just by looking at the >> "name" file. >> >> There was a suggestion to change the name from "i2c-N-mux (chan_id M)" >> to "i2c-N-mux-XX (chan_id M)" or even "i2c-N-mux-i2c-XX (chan_id M)", >> where XX is the multiplexer's i2c address. That would solve my >> problem, but unfortunately it looks like Guenter never submitted the >> patch (or maybe it was rejected?). > > It just dropped off :( But you guys have my attention now, let's fix > this issue for 3.19! I am just reading through the old mails and will > think about it. Input is welcome. > I didn't follow up on the issue since it was not an immediate concern, and my proposed solution had some problems. If I remember correctly, one of the problems was that the multiplexer does not have to be an i2c chip. In that case XX would be unknown and/or have to be omitted. Guenter >> I would like to submit a similar change, but I was thinking of adding >> a module parameter so that the change is not the default behavior. >> The idea is to preserve backward compatibility for applications that >> don't require this fix. For example, modprobe i2c-dev >> explicit_mux_id=1 would use i2c-N-mux-i2c-XX (chan_id M), whereas >> modprobe i2c-dev would default to the current behavior: i.e. i2c-N-mux >> (chan_id M). > > I don't like the need to set a module parameter to fix a flaw. I do > consider changing the ABI to have better strings in "name". But as said, > I need to think about it a little more... > > Thanks, > > Wolfram >