Management of Thermal

Ian Smith smithi at nimnet.asn.au
Tue Oct 23 23:19:40 PDT 2007


On Tue, 16 Oct 2007, Richard Arends wrote (in freebsd-mobile):
 > On Wed, Oct 17, 2007 at 02:27:48AM +1000, Ian Smith wrote:
 > 
 > Ian,
 > 
 > > Just browsing:
 > > http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/dev/acpi_support/acpi_ibm.c
 > > 
 > > What revision of that have you?  It's not clear to me tha this fan level
 > > stuff, documented with rev 1.10, was ever MFC'd to 6-STABLE.  If I'm
 > > diffing the right versions it appears that it wasn't, and the comment on
 > > 1.7.2.3 (6.X) says only 'MFC: 1.11, 1.12', but that seems not so, if I'm
 > > reading the diffs against any of those versions right (incl 1.14 @ 7.0)
 > 
 > You are absolutely right. On my stable tree I MFC'ed the acpi_ibm a few times
 > myself. Totaly forget about it :) Now i'm running the normal acpi_ibm in stable
 > again and indeed the fan_level hook isn't there anymore.
 > 
 > Adapting acpi_ibm to stable was not that hard but you can also try to run
 > RELENG_7, it is fairly stable at the moment :)

Following up cc'd to acpi@ and the presumed maintainer:

Over on -mobile we've been exploring ways to help Norberto stop his
(6-STABLE) Thinkpad z60m from melting - well, regularly hitting critical
thermal shutdown on larger builds - and these fan speed control sysctls
came up, along with a neat perl daemon using them to control fan speed
vs temperature setpoints (plus powerd-like cpu freq adjustment when it
gets hotter than maximum fan speed control can manage). 

Running 7 sounds good, but is there any particular (API?) reason that

 - Add support for setting the fan control mode to manual or automatic
 - Add support for adjusting the fan speed if the fan control mode is manual

from acpi_ibm.c rev 1.10, was not MFC'd to 6-STABLE?

If not, any chance before 6.3?

Cheers, Ian



More information about the freebsd-acpi mailing list