On Wednesday, December 15, 2004, at 06:03 AM, Michael wrote:

> Hello,

<...>
> That has a simple reason - the fan controller works autonomously. 
> That's why the fan spins up and down at all While NetBSD is running. 
> But OF programs rather high temperature values for the fan(s) to spin 
> up, MacOS X apparently changes them to lower ones, so the iBook will 
> run pretty hot in OF and NetBSD but remain cool in OSX. As soon as I 
> figured out how to (ab)use sysctl to talk to my driver NetBSD will be 
> able to change these things too, at least for those models using the 
> same controller ( an Analog Devices ADT 7467, see the 'compatibility' 
> property of /uni-n/i2c/fan in OF )
>
>> Netbsd-wise, I'm a little concerned if OSX feels the machine is hot
>> enough for some fan, unless its something else, like OSX sees 
>> different
>> values in fan related registers/varibles than it uses, and just
>> defaults to running the fan for a few secs..
> Hmm, the controller has 'calibration registers' to compensate linear 
> errors in temperature measurement, but I left them alone and the 
> readings looked rather high.
<...>
So the temp the fan kicks in is OF > NetBSD > OSX,  and since my fan 
only runs a little bit after booting into OSX (its off by end of bootup 
) temperature most be close to what OSX wants, and I can just avoid 
extended OF sessions on my ibook.

I tried your kernel, didn't crash, but didn't see anything fan related, 
but it's a G3
(hw.model = 750fx rev 2.2) and, dev /uni-n/i2c/fan .properties has 
"adm1030" in compatibility and type fields, so I guess I shouldn't be 
surprised.