Hi Rouven,
Please always reply to the list, not to me only. Only the DSDT file I
wanted you to send in private.
On Tue, 08 May 2007 10:58:49 +0000, Rouven Sacha wrote:
> Am Dienstag, den 08.05.2007, 11:55 +0200 schrieb Jean Delvare:
>
> > > May 2 18:41:43 blinkenmail kernel: ACPI: Unable to turn cooling device
> > > [c1b17dec] 'on'
> >
> > You may want to report this to the ACPI folks and/or your hardware
> > vendor, as this looks like a bug either in the Linux acpi code, or in
> > your system's BIOS.
>
> thanks for your kind reply. What makes you think it is a bug, i.e. how
> do I know that "Unable to turn cooling device on" isn't a proper error
> message from the acpi subsystem due to a broken fan?
Because lm_sensors shows two spinning fans.
> Unfortunately I forgot to mention that the system had a 3.0 load for
> about half an hour or so, which it never had before, so this incident
> doesn't look too random to me ...
Of course. I never meant to imply that it was random. The message you
got from ACPI makes it pretty clear that your system was really getting
hot.
> > I guess that you didn't power-off the system, and that it isn't really
> > an option?
>
> I can only guess. If the support guy really changed the fan, he would of
> course have been forced to do a cold boot after that. If he only
> switched of the sensors in the servers bios, it might as well have been
> a warm reboot. If you think a cold reboot would help, i can try to make
> the support initiate one tonight.
I don't really think it will help. But I just can't tell what happened
to your temperature reading, so I was suggesting a cold boot just in
case.
> > Does the ACPI temperature look reasonable ("acpi -t" or look directly
> > in /proc/acpi/thermal_zone/*/temperature)? It could be a conflict
> > between ACPI and the w83627hf driver. Can you please send me (in
> > private) a copy of /proc/acpi/dsdt?
>
> I unloaded w83627hf, modprobed thermal, with the following output:
>
> > 12:43:41 blinkenmail:/ # acpi -t
> > No support for device type: battery
> > Thermal 1: ok, 0.0 degrees C
Not exactly correct.
> sensors gives constant values of -93,5 ?C, regardless if thermal.ko is
> loaded alongside with w83627hf or not.
>
> You'll find the dsdt attached.
It seems to be corrupted, possibly because you named it "dsdt.txt"
while it is actually a binary file.
--
Jean Delvare
