Re: [PATCH] iio: pressure: bmp280: fix relative humidity unit

[Jonathan Cameron <jic23@kernel.org>]

On Wed, 30 May 2018 15:20:37 +0800
Matt Ranostay <matt.ranostay@konsulko.com> wrote:

> On Wed, May 30, 2018 at 1:05 PM, Matt Ranostay
> <matt.ranostay@konsulko.com> wrote:
> > On Wed, May 30, 2018 at 9:17 AM, Phil Reid <preid@electromag.com.au> wrote:  
> >> On 30/05/2018 08:44, Matt Ranostay wrote:  
> >>>
> >>> On Mon, May 28, 2018 at 11:38 PM, Tomasz Duszynski <tduszyns@gmail.com>
> >>> wrote:  
> >>>>
> >>>> According to IIO ABI relative humidity reading should be
> >>>> returned in milli percent.
> >>>>
> >>>> This patch addresses that by applying proper scaling and
> >>>> returning integer instead of fractional format type specifier.
> >>>>  
> >>>
> >>> *sigh* seems this is my mistake, but good catch.  Slight nitpick
> >>> below.. otherwise looks good
> >>>  
> >>>> Signed-off-by: Tomasz Duszynski <tduszyns@gmail.com>
> >>>> ---
> >>>>   drivers/iio/pressure/bmp280-core.c | 5 ++---
> >>>>   1 file changed, 2 insertions(+), 3 deletions(-)
> >>>>
> >>>> diff --git a/drivers/iio/pressure/bmp280-core.c
> >>>> b/drivers/iio/pressure/bmp280-core.c
> >>>> index 5ec3e41b65f2..fe87d27779d9 100644
> >>>> --- a/drivers/iio/pressure/bmp280-core.c
> >>>> +++ b/drivers/iio/pressure/bmp280-core.c
> >>>> @@ -415,10 +415,9 @@ static int bmp280_read_humid(struct bmp280_data
> >>>> *data, int *val, int *val2)
> >>>>          }
> >>>>          comp_humidity = bmp280_compensate_humidity(data, adc_humidity);
> >>>>
> >>>> -       *val = comp_humidity;
> >>>> -       *val2 = 1024;
> >>>> +       *val = comp_humidity * 1000 / 1024;  
> >>>
> >>>
> >>> Minor nitpick  that it would look cleaner as:   (comp_humidity / 1024) *
> >>> 1000
> >>>  
> >> be careful of integer division.
> >>  
> >
> > Ah yes good point. You will have to check if comp_humidity isn't zero
> > or it is possible to have a divide-by-zero.  
> 
> D'oh actually divide by zero would be an issue. But you'll want to be
> sure of possible overflows (doubt that would be an issue here).
> 
> However it may be better to just add the scaling factor of 1000 with
> IIO_CHAN_INFO_SCALE  and make the processed value now a raw one.
> 

Don't do that.  It would be an ABI change.  Whilst we can hope that
everyone is using a nice library that would cope with this, there
is no guarantee they are.

Jonathan

> >
> > - Matt
> >  
> >>  
> >>>>
> >>>> -       return IIO_VAL_FRACTIONAL;
> >>>> +       return IIO_VAL_INT;
> >>>>   }
> >>>>
> >>>>   static int bmp280_read_raw(struct iio_dev *indio_dev,
> >>>> --
> >>>> 2.17.0  
> >>
> >>  
> >>>  
> >>
> >>
> >> --
> >> Regards
> >> Phil Reid
> >>