On Thu, Jun 27, 2013 at 10:36:31AM +0200, Jean Delvare wrote:
> I'm once again late, but I could not find the time to test your ds1621
> patches before today.
>
> On Wed, 8 May 2013 22:45:53 -0700, Robert Coulson wrote:
> > --- a/drivers/hwmon/ds1621.c
> > +++ b/drivers/hwmon/ds1621.c
> > (...)
> > +/*
> > + * TEMP: 0.001C/bit (-55C to +125C)
> > + * REG:
> > + * - 1621, 1625: x = 0.5C
> > + * - 1721: x = 0.0625C
> > + * Assume highest resolution and let the bits fall where they may..
> > + */
> > +static inline u16 DS1621_TEMP_TO_REG(long temp)
> > +{
> > + int ntemp = clamp_val(temp, DS1621_TEMP_MIN, DS1621_TEMP_MAX);
> > + ntemp += (ntemp < 0 ? -31 : 31);
> > + ntemp = DIV_ROUND_CLOSEST(ntemp * 10, 625) << 4;
> > + return (u16)ntemp;
> > +}
>
> This breaks limit rounding on the DS1621. Limits have a resolution of
> 0.5°C on that chip, so limit values should rounded to the closest
> half-degree. It used to work but no longer: if I write 20300 to
> temp1_min, it is rounded down to 20000, while it should be rounded up
> to 20500.
>
My fault, since I proposed that approach. Didn't I have the same problem
with some other driver ? Makes it even worse. I'll have to look up my notes.
Guenter
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
