[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <15f25d73-32d5-4809-8096-32c856559d66@app.fastmail.com>
Date: Mon, 23 Oct 2023 16:19:32 +0200
From: "Arnd Bergmann" <arnd@...db.de>
To: "Eliza Balas" <Eliza.Balas@...log.com>,
"Greg Kroah-Hartman" <gregkh@...uxfoundation.org>
Cc: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
"Rob Herring" <robh+dt@...nel.org>,
"Krzysztof Kozlowski" <krzysztof.kozlowski+dt@...aro.org>,
"Conor Dooley" <conor+dt@...nel.org>,
"derek.kiernan@....com" <derek.kiernan@....com>,
"dragan.cvetic@....com" <dragan.cvetic@....com>,
"Jonathan Cameron" <jic23@...nel.org>,
"Lars-Peter Clausen" <lars@...afoo.de>, linux-iio@...r.kernel.org
Subject: Re: [PATCH v3 2/2] drivers: misc: adi-axi-tdd: Add TDD engine
On Mon, Oct 23, 2023, at 15:30, Balas, Eliza wrote:
>> -----Original Message-----
>> Cvetic <dragan.cvetic@....com>; Arnd Bergmann <arnd@...db.de>
>> Subject: Re: [PATCH v3 2/2] drivers: misc: adi-axi-tdd: Add TDD engine
>> > > > Since the device is not an iio device, using an iio function would be confusing.
>> > >
>> > > Why isn't this an iio device?
>> >
>> > The device is not registered into the IIO device tree,
>> > and does not rely on IIO kernel APIs.
>> > Even though there are a few attributes that resemble the
>> > ones from iio, and the sysfs structure is similar,
>> > this is not an IIO device.
>> > In the previous patch versions 1 and 2 we concluded
>> > that this device fits better in the misc subsystem.
>>
>> Ok, can you point to that in the changelog where the IIO maintainer
>> agreed that this doesn't fit into that subsystem?
>>
> This was one of the discussions from previous v2 :
> https://lore.kernel.org/all/5b6318f16799e6e2575fe541e83e42e0afebe6cf.camel@gmail.com/
>
> I will add it to the changelog the next time I submit the patches.
It sounds like Jonathan wasn't quite sure either here, and I would
still argue (as I did in that thread), that drivers/iio is probably
a better option than drivers/misc.
In particular, you mention that you actually make this device
appear as an IIO device to user space using the "iio-fake" hack.
I can see that IIO is not a perfect fit if this is the only
device of its kind, but going that way anyway avoids a number
of problems by reusing infrastructure for the IIO ABI and
serialization with in-kernel users, as well as giving you
the option of adding other compatible drivers later.
Arnd
Powered by blists - more mailing lists