Quoting Philip Pokorny <ppokorny at penguincomputing.com>:
> Alexandre d'Alton wrote:
>
> >Quoting Jean Delvare <khali at linux-fr.org>:
> >
> >
> >
> >>>>Let's compare with the ADM030 which has:
> >>>>1* a temperature over which the fan starts spinning;
> >>>>2* a range of temperature after Tmin for which the fan has variable
> >>>>PWM (this is another way to define Tmax);
> >>>>4* a PWM value for 1*;
> >>>>
> >>>>
> >>>Actually, the (4*) item, when in automatic mode, define the minimum
> >>>speed at which the fan will rotate. Even if Temp < (1*)
> >>>
> >>>
> >>This doesn't match the graph page 12 of the ADM1030-A datasheet. If I
> >>read it correctly, PWM (4*) applies if (1*) - 5 < T < (1*).
> >>
> >>
> >
> >OK, just read back, it's clear. Sorry.
> >
> >
> >
> >>>I would just add a file to activate the automatic mode. (It is not
> >>>activated by default.)
> >>>
> >>>
> >>fan1_auto_enable?
> >>
> >>
> >>
> >>
> >>>and to select on which temp channel the automatic mode
> >>>reacts. (temp1, temp2, temp3 or max value of the three.)
> >>>
> >>>
> >>I don't know if IT87 or Winbond chips can select that. I guess that
> >>"fan1_auto_temp_channel" would be a correct name?
> >>
> >>Note that I don't have strong opinions on the names except that they
> >>must be standardized and self-expressive. The more important (and
> >>difficult) problem is to ensure that all known chips (and it possible
> >>future ones) will fit into the chose interface.
> >>
> >>
> >>
>
> In the LM85/ADM1027 driver it's called pwm1_zone.
>
> It has the values:
> -1: PWM disabled, fan at 100%
> -2: PWM under manual control ('pwm1' or 'pwm1_value' is writable)
> 1: PWM is controlled by temp1_input
> 2: PWM is controlled by temp2_input
> 123: PWM is controlled by max value of all three
> 0: PWM disabled, fan at 0%
>
> The other parameters have equivalents in the lm85 driver as well.
>
> zone#_limit (fan "turn on" temperature (1*))
> zone#_hyst (range below limit fan remains on. Fixed at 5 deg in
> this chip?)
> zone#_range (temperature range (2*))
> pwm#_min (PWM value at low limit. (4*))
>
> :v)
Seems to be a good approach, and it has the advantage of already beeing
implemented on some drivers.
Jean, what do you think about that ?
--
-------------------------------------------------
Alexandre d'ALTON
alex at alexdalton.org
