Re: Documentation of iio_read_channel_processed return value

[Jonathan Cameron <jic23@kernel.org>]

On Sat, 08 Nov 2025 18:55:08 +0100
"Luca Weiss" <luca.weiss@fairphone.com> wrote:

> Hi all,
> 
> I've noticed that the docstring in include/linux/iio/consumer.h for
> iio_read_channel_processed (and likely some other functions in there) is
> wrong because it's saying "Returns an error code or 0." while often it's
> returning e.g. IIO_VAL_INT. In some code paths it's actually returning 0
> but it seems in standard cases it's not.
Hmm. Looks like we have a bug in handling the path that ends up in iio_multiple_value()

I think we just want to change iio_read_channel_processed_scale() to check
that the return value of iio_mutliple_value() < 0 and return ret if so, otherwise
return 0

So to me smells like a bug and the documentation is correct.

Good catch, perhaps spin a patch like the above?  This is made more complex
because a few drivers seem to assume IIO_INT_VAL such as iio/afe/iio-rescale.c so
we will need to fix those up as well.

J

> 
> Could the docstring please be updated so this is clear to users of this
> function? I don't understand IIO well enough to confidently propose
> changes there myself.
> 
> Thanks and kind regards
> Luca