linux-kernel - Re: [RFC PATCH v1 0/2] hwmon: (iio_hwmon) optionally force iio channel type

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sun, 16 May 2021 10:06:31 +0100
From:   Jonathan Cameron <jic23@...nel.org>
To:     Liam Beguin <liambeguin@...il.com>
Cc:     jdelvare@...e.com, linux@...ck-us.net, lars@...afoo.de,
        pmeerw@...erw.net, linux-hwmon@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-iio@...r.kernel.org,
        devicetree@...r.kernel.org, robh+dt@...nel.org,
        Peter Rosin <peda@...ntia.se>
Subject: Re: [RFC PATCH v1 0/2] hwmon: (iio_hwmon) optionally force iio
 channel type

On Sun, 16 May 2021 00:43:13 -0400
Liam Beguin <liambeguin@...il.com> wrote:

> Add a devicetree binding to optionally force a different IIO channel
> type.
> 
> This is useful in cases where ADC channels are connected to a circuit
> that represent another unit such as a temperature or a current.
> 
> `channel-types` was chosen instead of `io-channel-types` as this is not
> part of the iio consumer bindings.
> 
> In the current form, this patch does what it's intended to do:
> change the unit displayed by `sensors`, but feels like the wrong way to
> address the problem.
> 
> Would it be possible to force the type of different IIO channels for
> this kind of use case with a devicetree binding from the IIO subsystem?
> 
> It would be convenient to do it within the IIO subsystem to have the
> right unit there too.
> 
> Thanks for your time,
> Liam

Hi Liam,

+CC Peter for AFE part.

It's an interesting approach, but I would suggest we think about this
a different way.

Whenever a channel is being used to measure something 'different' from
what it actually measures (e.g. a voltage ADC measuring a current) that
reflects their being some analog component involved.
If you look at drivers/iio/afe/iio-rescale.c you can see the approach
we currently use to handle this.

Effectively what you add to devicetree is a consumer of the ADC channel
which in turn provides services to other devices. For this current case
it would be either a current-sense-amplifier or a current-sense-shunt 
depending on what the analog front end looks like.  We have to describe
the characteristics of that front end which isn't something that can
be done via a simple channel type.

That afe consumer device can then provide services to another consumer
(e.g. iio-hwmon) which work for your usecase.

The main limitation of this approach currently is you end up with
one device per channel.  That could be improved upon if you have a usecase
where it matters.

I don't think we currently have an equivalent for temperature sensing
but it would be easy enough to do something similar.

Jonathan

> 
> Liam Beguin (2):
>   hwmon: (iio_hwmon) optionally force iio channel type
>   dt-bindings: hwmon: add iio-hwmon bindings
> 
>  .../devicetree/bindings/hwmon/iio-hwmon.yaml  | 41 +++++++++++++++++++
>  drivers/hwmon/iio_hwmon.c                     |  2 +
>  2 files changed, 43 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/hwmon/iio-hwmon.yaml
> 
> 
> base-commit: 9f4ad9e425a1d3b6a34617b8ea226d56a119a717