Re: [PATCH] ACPI: Remove possible recursion from thermal driver

[Len Brown <lenb@kernel.org>]

On Tuesday 15 May 2007 22:19, Alexey Starikovskiy wrote:
> Please read the spec 11.1.2 and 11.1.2.3. Trip points change _only_ to
> create a hysteresis loop, not by themselves. This means that we will
> get 0x81 type _only_ in response to us changing active state, the
> moment then we have temperature and just decided to move away from
> this trip point, not closer to it.
> If the temperature change, we get another event, 0x80, and _then_ we
> do thermal_check()...
> 

Hmm, this sounds promising.
Bob and I discussed this a few weeks ago and for some reason
we were convinced that we needed to get the temperature and
compare it to the new trip points.  Maybe we were wrong.

-Len

> 
> On 5/16/07, Len Brown <lenb@kernel.org> wrote:
> >
> > > --- a/drivers/acpi/thermal.c
> > > +++ b/drivers/acpi/thermal.c
> > > @@ -1106,7 +1106,6 @@ static void acpi_thermal_notify(acpi_handle handle, u32 event, void *data)
> > >               break;
> > >       case ACPI_THERMAL_NOTIFY_THRESHOLDS:
> > >               acpi_thermal_get_trip_points(tz);
> > >-              acpi_thermal_check(tz);
> > >               acpi_bus_generate_event(device, event, 0);
> > >               break;
> > >       case ACPI_THERMAL_NOTIFY_DEVICES:
> >
> > I don't think we can do this.
> > When the thresholds change, there must be a check to compare the temperature
> > with the thresholds -- otherwise what good was it to change the thresholds?
> >
> > -Len
> >
>