Re: [PATCH v7 5/5] iio: ad4080: Rework filter_type "none" logic

[Ben Collins <bcollins@kernel.org>]

On Tue, Aug 26, 2025 at 11:51:56AM -0500, David Lechner wrote:
> On 8/25/25 7:10 PM, Ben Collins wrote:
> > The filter_type logic for "none" needed to be reworked to be more
> > general.
> > 
> > As documented, return IIO_VAL_EMPTY for sampling rates in "none" type
> > and EINVAL when there's an attempt to write a rate for "none" type.
> 
> This patch breaks usespace, which is something we always must avoid.

I was under the impression there was a need to make the use of
filter_type "none" more consistent.

I don't disagree with not breaking userspace, but it does create
ambiguity for other implementations.

> > 
> > Signed-off-by: Ben Collins <bcollins@kernel.org>
> > ---
> >  drivers/iio/adc/ad4080.c | 23 ++++++++++-------------
> >  1 file changed, 10 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/iio/adc/ad4080.c b/drivers/iio/adc/ad4080.c
> > index 6e61787ed3213fe4332bd92b938a7a717dada99f..c7408b9703731ee5d4229a85ffa91ea64b233cd9 100644
> > --- a/drivers/iio/adc/ad4080.c
> > +++ b/drivers/iio/adc/ad4080.c
> > @@ -154,8 +154,6 @@ static const int ad4080_dec_rate_avail[] = {
> >  	2, 4, 8, 16, 32, 64, 128, 256, 512, 1024,
> >  };
> >  
> > -static const int ad4080_dec_rate_none[] = { 1 };
> > -
> >  static const char * const ad4080_power_supplies[] = {
> >  	"vdd33", "vdd11", "vddldo", "iovdd", "vrefin",
> >  };
> > @@ -268,13 +266,13 @@ static int ad4080_read_raw(struct iio_dev *indio_dev,
> >  			*val = st->clk_rate;
> >  		return IIO_VAL_INT;
> >  	case IIO_CHAN_INFO_OVERSAMPLING_RATIO:
> > -		if (st->filter_type == FILTER_NONE) {
> > -			*val = 1;
> > -		} else {
> > -			*val = ad4080_get_dec_rate(indio_dev, chan);
> > -			if (*val < 0)
> > -				return *val;
> > -		}
> > +		if (st->filter_type == FILTER_NONE)
> > +			return IIO_VAL_EMPTY;
> > +
> > +		*val = ad4080_get_dec_rate(indio_dev, chan);
> > +		if (*val < 0)
> > +			return *val;
> > +
> >  		return IIO_VAL_INT;
> >  	default:
> >  		return -EINVAL;
> > @@ -289,7 +287,7 @@ static int ad4080_write_raw(struct iio_dev *indio_dev,
> >  
> >  	switch (mask) {
> >  	case IIO_CHAN_INFO_OVERSAMPLING_RATIO:
> > -		if (st->filter_type == FILTER_NONE && val > 1)
> > +		if (st->filter_type == FILTER_NONE)
> >  			return -EINVAL;
> >  
> >  		return ad4080_set_dec_rate(indio_dev, chan, val);
> > @@ -376,17 +374,16 @@ static int ad4080_read_avail(struct iio_dev *indio_dev,
> >  	case IIO_CHAN_INFO_OVERSAMPLING_RATIO:
> >  		switch (st->filter_type) {
> >  		case FILTER_NONE:
> > -			*vals = ad4080_dec_rate_none;
> > -			*length = ARRAY_SIZE(ad4080_dec_rate_none);
> > +			*type = IIO_VAL_EMPTY;
> >  			break;
> >  		default:
> >  			*vals = ad4080_dec_rate_avail;
> >  			*length = st->filter_type >= SINC_5 ?
> >  				  (ARRAY_SIZE(ad4080_dec_rate_avail) - 2) :
> >  				  ARRAY_SIZE(ad4080_dec_rate_avail);
> > +			*type = IIO_VAL_INT;
> >  			break;
> >  		}
> > -		*type = IIO_VAL_INT;
> >  		return IIO_AVAIL_LIST;
> >  	default:
> >  		return -EINVAL;
> > 
> 
> Returning a value of 1 for the oversampling ratio when there is no
> oversampling going on is perfectly reasonable and mathematically correct.
> So I don't consider this change an improvement.

-- 
 Ben Collins
 https://libjwt.io
 https://github.com/benmcollins
 --
 3EC9 7598 1672 961A 1139  173A 5D5A 57C7 242B 22CF