From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail.free-electrons.com ([62.4.15.54]:42117 "EHLO mail.free-electrons.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751816AbeAaKyU (ORCPT ); Wed, 31 Jan 2018 05:54:20 -0500 Date: Wed, 31 Jan 2018 11:54:18 +0100 From: Alexandre Belloni To: Guenter Roeck Cc: Denis OSTERLAND , "linux-rtc@vger.kernel.org" , "a.zummo@towertech.it" , "kernel@pengutronix.de" , "mgr@pengutronix.de" , "jdelvare@suse.com" , "devicetree@vger.kernel.org" , "linux-kernel@vger.kernel.org" Subject: Re: [PATCH 4/4] rtc: isl1208: add support for isl1219 with hwmon for tamper detection Message-ID: <20180131105418.GP2809@piout.net> References: <20180123121801.4214-1-m.grzeschik@pengutronix.de> <20180123121801.4214-5-m.grzeschik@pengutronix.de> <20180123182254.GA27379@roeck-us.net> <20180124090333.r5o2mzpm4q536w4r@pengutronix.de> <20180129215919.GA17988@roeck-us.net> <20180130102740.GD2809@piout.net> <1517312409.5307.22.camel@diehl.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 In-Reply-To: Sender: linux-rtc-owner@vger.kernel.org List-ID: On 30/01/2018 at 06:15:18 -0800, Guenter Roeck wrote: > On 01/30/2018 03:40 AM, Denis OSTERLAND wrote: > > Am Dienstag, den 30.01.2018, 11:27 +0100 schrieb Alexandre Belloni: > > > On 29/01/2018 at 13:59:19 -0800, Guenter Roeck wrote: > > > > > > > > On Wed, Jan 24, 2018 at 10:03:33AM +0100, Michael Grzeschik wrote: > > > > [ ... ] > > > > > > > > > > > > > > > > > > > > > > > > > + > > > > > > > diff --git a/Documentation/hwmon/sysfs-interface b/Documentation/hwmon/sysfs-interface > > > > > > > index fc337c317c673..a12b3c2b2a18c 100644 > > > > > > > --- a/Documentation/hwmon/sysfs-interface > > > > > > > +++ b/Documentation/hwmon/sysfs-interface > > > > > > > @@ -702,6 +702,13 @@ intrusion[0-*]_alarm > > > > > > >   the user. This is done by writing 0 to the file. Writing > > > > > > >   other values is unsupported. > > > > > > > +intrusion[0-*]_timestamp > > > > > > > + Chassis intrusion detection > > > > > > > + YYYY-MM-DD HH:MM:SS UTC (ts.sec): intrusion detected > > > > > > > + RO > > > > > > > + The corresponding timestamp on which the intrustion > > > > > > > + was detected. > > > > > > > + > > > > > > Sneaky. Nack. You don't just add attributes to the ABI because you want it, > > > > > > without serious discussion, and much less so hidden in an RTC driver > > > > > > (and even less as unparseable attribute). > > > > > Right; but it was not meant to be sneaky. I should have stick to my first > > > > > thought and label this patch RFC. Sorry for that. > > > > > > > > > > > > > > > > > In addition to that, I consider the attribute unnecessary. The intrusion > > > > > > already generates an event which should be sufficient for all practical > > > > > > purposes. > > > > > Would it make sense in between the other sysfs attributes of this driver? > > > > > > > > > I don't understand what you mean with that, sorry. > > > > > > > > From an ABI perspective, the attibute doesn't add value since it is > > > > highly device specific (or at least it is the only chip I am aware of > > > > which reports such a time stamp). Feel free to add the attribute to the > > > > driver and document it, but not as part of the hwmon ABI. In that > > > > case I would be inclined to accept it. However, keep in mind that > > > > your version, reporting a human readable date/time, would effectively > > > > preclude it from ever making it into the ABI. > > > > > > > Actually, there are many RTCs that are able to register one or more > > > timestamps. My plan was to add support for that soon but I was not > > > planning to do so in the hwmon ABI as this may be used for something > > > that is not intrusion detection (interval timers for example). > > What would you suggest? > > I think about something like this: > > event[0-*]_timestamp: timestamp in seconds since epoch or empty if not triggered > > event[0-*]_alarm: 1 if event was triggered, else 0; write 0 to clear event > > Sure, that makes sense if the events are not specifically related > to intrusion detection. Question is if there would ever be more than one > or if event_timestamp and event_alarm would be sufficient. > My target is a PCF85363A which supports up to 3 timestamps. SO I'd go for timestamp[0-*]. This would be empty if it never triggered (or the timestamp is invalid) writing anything to that file resets the event. I don't think we need more than one file per timestamp. -- Alexandre Belloni, Free Electrons Embedded Linux and Kernel engineering http://free-electrons.com