Re: 5.0.1 boot messages: a few questions

[Jukka Ruohonen <jruohonen%iki.fi@localhost>]

On Fri, Jan 08, 2010 at 05:58:33AM -0800, Paul Goyette wrote:
> On Fri, 8 Jan 2010, Jukka Ruohonen wrote:
> 
> >On Fri, Jan 08, 2010 at 05:09:07AM -0800, Paul Goyette wrote:
> >>Yes.  As far as I know, with ACPI FAN you can either turn the fan on, or
> >>turn it off.  Nothing in between.  Not very useful.
> >
> >Agreed. Perhaps it should go to the list of ignored HIDs.
> 
> IMHO, we should either ignore it, or provide an option to control it.
> 
> "Control" could be done through sysmon_envsys(8) but I suspect a single, 
> simple sysctl(8) variable might be more appropriate.

Well, I think it should be ignored. Adding it to the list of ignored HIDs
would bypass it in autoconfiguration and stop the spam during boot.

If ACPI is functioning, there should be no reason to mangle with the fans.
That is after all the whole idea of "advanced configuration and power
interface".

As for the drivers, I had a laptop with four PNP0C0Bs but only a single fan.
Each instance corresponded with the four acpitz(4) instances that the laptop
had. I could control the fan (and its speed) by hooking the PNP0C0Bs to the
ACPI power resource code. I did it by sysctl-knobs and the result was
something like:

        hw.acpi.fan1=0
        hw.acpi.fan2=0
        hw.acpi.fan3=1
        hw.acpi.fan4=0

Now how this should be interpreted in the context of a generic driver? In my
case the fan was running "almost in full speed" (hw.acpi.fan3=1), but it
could well be that in some other machine there would actually be four fans
and the third one would be running.

There is also additional concern: the so-called ACPI power resources can
power multiple devices. So if we force PNP0C0B off manually, it could well
be that this turns off power resources required by some other device!
Ideally the power resource code should strictly forbid such request, of
course. (I haven't checked if the current code does that, but it should.)

- Jukka.