[Pm-utils] sometimes on resume, fan very loud until power off

Holger Macht hmacht at suse.de
Tue Nov 28 12:05:59 PST 2006


On Tue 28. Nov - 15:01:20, Peter Jones wrote:
> On Tue, 2006-11-28 at 13:43 -0500, Peter Jones wrote:
> > On Mon, 2006-11-27 at 18:01 +0100, Holger Macht wrote:
> > > On Mon 27. Nov - 11:51:15, Peter Jones wrote:
> > > > On Mon, 2006-11-27 at 11:42 -0500, John H. wrote:
> > > > > fc6
> > > > 
> > > > If you do:
> > > > 
> > > > cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
> > > > 
> > > > what does it say?  If it's "performance", do this (as root):
> > > > 
> > > > echo ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
> > > > 
> > > > which will bring the system back to normal within a few seconds,
> > > > hopefully.  I'm not quite sure what's doing it, but sometimes it seems
> > > > something is setting it to performance occasionally, particularly right
> > > > after the system resumes.  I'm not sure what just yet.
> > > 
> > > I just wonder why this happens with suspend to ram. For hibernate, I know
> > > the cause.
> 
> > > So not entirely related to this problem, on hibernation, all CPUs except
> > > the first one is offlined. When coming back from suspend, they are onlined
> > > again, but the default governor is performance. And 94cpufreq only resets
> > > the governor for the suspend case. Peter, can you extend it to also reset
> > > it on hibernate?
> > 
> > Oh, good point.  Will fix.
> 
> OK, we're both wrong.  Hibernate is the *only* place we do this, and we
> do it for all CPUs.
> 
> So maybe John's problem indicates that we should be doing this on
> suspend/resume as well.

Oh, yes, I mixed it up ;-) What I wanted to say and what I've already in
my local pm-utils is the following:

--- pm/hooks/94cpufreq	2006-10-23 17:19:54.000000000 +0200
+++ /etc/pm/hooks/94cpufreq	2006-11-28 21:04:19.000000000 +0100
@@ -31,10 +31,10 @@
 }
 
 case "$1" in
-	hibernate)
+	suspend|hibernate)
 		hibernate_cpufreq
 		;;
-	thaw)
+	resume|thaw)
 		thaw_cpufreq
 		;;
 	*)


More information about the Pm-utils mailing list