Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Sun, 4 Jun 2023 13:52:28 +0100
From:   Jonathan Cameron <jic23@kernel.org>
To:     Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Cc:     George Stark <GNStark@sberdevices.ru>,
        "lars@metafoo.de" <lars@metafoo.de>,
        "neil.armstrong@linaro.org" <neil.armstrong@linaro.org>,
        "khilman@baylibre.com" <khilman@baylibre.com>,
        "jbrunet@baylibre.com" <jbrunet@baylibre.com>,
        "andriy.shevchenko@linux.intel.com" 
        <andriy.shevchenko@linux.intel.com>,
        "nuno.sa@analog.com" <nuno.sa@analog.com>,
        "linux-iio@vger.kernel.org" <linux-iio@vger.kernel.org>,
        "linux-arm-kernel@lists.infradead.org" 
        <linux-arm-kernel@lists.infradead.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-amlogic@lists.infradead.org" 
        <linux-amlogic@lists.infradead.org>, kernel <kernel@sberdevices.ru>
Subject: Re: [PATCH v2] meson saradc: add iio device attrib to switch
 channel 7 mux
Message-ID: <20230604135228.5a5d743d@jic23-huawei>
In-Reply-To: <CAFBinCA=earf119bN7ohmTQGDzSMeV9qcL8RU9DUYe0195H5=Q@mail.gmail.com>
References: <20230527214854.126517-1-gnstark@sberdevices.ru>
        <20230528161117.197f7e61@jic23-huawei>
        <4d5a7691-68ac-6626-5502-383fad3a9436@sberdevices.ru>
        <CAFBinCA=earf119bN7ohmTQGDzSMeV9qcL8RU9DUYe0195H5=Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Mon, 29 May 2023 22:04:18 +0200
Martin Blumenstingl <martin.blumenstingl@googlemail.com> wrote:

> Hi George,
>=20
> On Sun, May 28, 2023 at 11:52=E2=80=AFPM George Stark <GNStark@sberdevice=
s.ru> wrote:
> [...]
> > > Based on the limited description we have here, I'm not understanding =
why
> > > you don't just express this as a set of channels. One channel per mux
> > > setting, with the in_voltageX_label providing the information on what=
 the
> > > channel is connected to.
> > >
> > > This is an interesting facility, so good to enable for high precision=
 calibration
> > > but we still want to map it to standards signals.  Userspace doesn't
> > > care that these are all being measured via the same input 7 - which
> > > is itself probably an input to a MUX.
> > >
> > > Jonathan =20
> >
> > Hello Jonathan
> >
> > Thanks for the review.
> >
> > Your idea of exposing the mux setting as iio channels is very
> > interesting and at least worth trying.
> > The sysfs approach was chosen because of the code changes are simple and
> > neat (compare to channels approach).
> > Also calibration by using those mux inputs are already supported in the
> > driver (performed at probe stage) so I expect very special usecases for
> > those mux settings like debug or device production stage tests. In those
> > usescases hardware specific knowledge is required anyway. =20

Agreed that using channels for this is more complex code wise, but the
avoidance of custom ABI is usually worthwhile.  However I get your point
that this is weird debug / corner case paths anyway.
However...


> Another downside to the debugfs approach is user support:
> If someone reports odd values on ADC channel 7 then we need to make
> sure to double check if the mux has been altered from userspace (the
> calibration during initialization ensures to leave channel 7 in a
> consistent state, while a user may change the mux, forget about it and
> then complain that values are wrong).

This problem raises red flags for me.  If the channel wasn't otherwise
useful then treating it as a weird debug thing is fine, but if a normal
read can end up returning weird answers because of leftover configuration
that is custom ABI, so standard code isn't even aware of it, that's bad.

Hence I would not want to see any path under which that can happen.
A read of in_voltage7_raw should continue to return the normal value
whatever this patch adds.

Jonathan

>=20
>=20
> Best regards,
> Martin