Re: [PATCH] iio: afe: iio-rescale: Support processed channels

From: Jonathan Cameron <Jonathan.Cameron@Huawei.com>
To: Linus Walleij <linus.walleij@linaro.org>
Cc: Peter Rosin <peda@axentia.se>,
	Jonathan Cameron <jic23@kernel.org>,
	linux-iio <linux-iio@vger.kernel.org>,
	Hartmut Knaack <knaack.h@gmx.de>,
	Lars-Peter Clausen <lars@metafoo.de>,
	Peter Meerwald-Stadler <pmeerw@pmeerw.net>
Subject: Re: [PATCH] iio: afe: iio-rescale: Support processed channels
Date: Mon, 4 Jan 2021 17:11:14 +0000	[thread overview]
Message-ID: <20210104171114.000042df@Huawei.com> (raw)
In-Reply-To: <CACRpkdbOx0AiFOejZ5gtjQwQNo-qzuRRNVDhjAbAk1pGLMj8Yw@mail.gmail.com>

On Mon, 4 Jan 2021 15:45:07 +0100
Linus Walleij <linus.walleij@linaro.org> wrote:

> On Mon, Dec 14, 2020 at 5:34 PM Jonathan Cameron
> <Jonathan.Cameron@huawei.com> wrote:
> > On Mon, 14 Dec 2020 16:30:22 +0100 Peter Rosin <peda@axentia.se> wrote:  
> 
> > > >> And that gets transformed by the
> > > >> rescaler into the processed values being presented as raw, with rescaling
> > > >> added on top, but with the read_avail info for this new raw channel being
> > > >> completely wrong.
> > > >>
> > > >> For the intended driver (ab8500-gpadc) this is not the case (it has no
> > > >> read_avail for its raw channel). But it does have a raw channel, so adding
> > > >> read_avail seems easy and I can easily see other drivers already doing it.
> > > >> Haven't checked that though...  
> > > >
> > > > Drat. I'd failed to register this is one of those corner cases.  
> > >
> > > I'm not sure, I just browsed the code. Maybe I misread it?  
> >
> > It's doing both - you were right.  I think there are only a small number of
> > drivers that have that history.
> >
> > Looks superficially like it's easy enough to catch this corner case and
> > block it - so lets do that.  
> 
> Sorry if I am a bit confused here. I don't understand what I am supposed
> to do to proceed with using this driver with the ab8500 GPADC...
> 
> Shall I fix something in the AB8500 GPADC as a prerequisite?
> In that case I think I need some more pointers...

I confess I'm a bit lost, but I 'think' the problem we had
left was around read_avail which doesn't play well if we
it defined for the _raw value in the provider, but not the _processed value.

So if we detect their is a _processed channel (which we are going to use) we
just need to make sure that we don't pass the read_avail for _raw through
to be exposed by the rescale driver as the consumer as it will be garbage.
Best plan is probably to just pretend the read_avail for the provider doesn't
exist in this case.

@Peter, does that cover it of are there other similar cases?

It definitely also wants a big fat comment saying why we are hiding this!

Jonathan

> 
> Yours,
> Linus Walleij