fans stopped working when upgraded from 2.6.8-2-686 to 2.6.8-3-686

[Henrik Hellerstedt]

Didnt find a Component for kernel-image-2.6.8-3, so i used this, hopefully
someone will add and move it....

The fans in my HP Compaq NC8000 stopped working when i upgraded the kernel from
2.6.8-2-686 to 2.6.8-3-686.

If the box is hot when booted the bios will fire up the fans, on 2.6.8-2 the
fans continue and are managed by the kernel (afaik), but when 2.6.8-3 the fans
are halted right after the kernel has uncompressed itself

Following urls hopefully has (/me is guessimating) relevant info:

http://anka.org/henrik/ubuntu/bug/fan.NOT.working.2.6.8.1-3-686/
http://anka.org/henrik/ubuntu/bug/fan.working.2.6.8.1-2-686/

[Henrik Hellerstedt (henrik8888)]

I noticed one thing, when i boot the 2.6.8.1-3-686 kernel as opposed to then
booting the 2.6.8.1-2-686 a thing shows up
just after the kernel decompressed itself:

PCI: Cannot allocate resource region 4 of device 0000:00:1f.3

hope that helps the wizard dealing with this....

[Daniel Stone (daniels)]

When you run 'for i in /proc/acpi/fan/*; do echo 0 > $i/state; sleep 5; echo 3 >
$i/state; done' in a root shell, do the fans turn on for a couple of seconds,
then off, sequentially?

[Daniel Stone (daniels)]

To mine eyes, this looks the most significant:

Working:
--
ACPI: Thermal Zone [TZ1] (59 C)
ACPI: Thermal Zone [TZ2] (55 C)
ACPI: Thermal Zone [TZ3] (33 C)
--

Not working:
--
ACPI: Thermal Zone [TZ1] (16 C)
    ACPI-1133: *** Error: Method execution failed [\_TZ_.C204] (Node dfb44f00),
AE_AML_PACKAGE_LIMIT
    ACPI-1133: *** Error: Method execution failed [\_TZ_.C203] (Node dfb44f20),
AE_AML_PACKAGE_LIMIT
    ACPI-1133: *** Error: Method execution failed [\_TZ_.TZ2_._TMP] (Node
dfb437a0), AE_AML_PACKAGE_LIMIT
ACPI: Thermal Zone [TZ3] (33 C)
--

It looks as if TZ1's temperature is wrong, unless it's a cold day, and TZ2 can't
be read at all -- this would mean that the fans cannot be triggered
automatically as the correct temperature can't be picked up.

[Henrik Hellerstedt (henrik8888)]

The test with
for i in /proc/acpi/fan/*; do echo 0 > $i/state; sleep 5; echo 3 > $i/state;
done will turn on the fans for 5 sec each

Testing with the 2.6.8.1-3 kernel that fail to "autostart" the fans when needed...

[Daniel Stone (daniels)]

OK, so it looks like the ACPI Thermal Zones are not working correctly, as per
comment #3.

[Herbert Xu (herbert-gondor)]

I've forwarded this bug upstream at http://bugzilla.kernel.org/show_bug.cgi?id=3555.

[Scott James Remnant (Canonical) (canonical-scott)]

Is kacpid spinning and using 99% of the CPU ... This looks like a dup of 1922,
or at least another problem caused by the same change.

The nc6000 and nc8000 are roughly identical; so's my nc4010 which has both sets
of symptoms.

[Scott James Remnant (Canonical) (canonical-scott)]

Random thing to try (this fixes things for me):

boot with init=/bin/sh
rmmod thermal
modprobe thermal

(It looks like the order the modules are loaded in the initrd is wrong, thermal
has to come after fan to be able to control them)

[Daniel Stone (daniels)]

Please try with an updated initrd-tools:
During the install, near the very end of the first stage, hit alt-F3 to get a shell.
At the prompt:
  # chroot /target /bin/sh
  # cd /tmp
  # wget
http://people.ubuntu.com/~daniels/thermal-fans/all/initrd-tools_0.1.70ubuntu10_all.deb
  # dpkg -i initrd-tools_0.1.70ubuntu10_all.deb
  # apt-get install --reinstall linux-image-2.6.8.1-3-686

Hit alt-F1 to go back to the installer, press OK, and reboot -- your fans should
now work. Please confirm whether this fix works or not.

[Henrik Hellerstedt (henrik8888)]

# chroot /target /bin/sh
# cd /tmp
# wget
http://people.ubuntu.com/~daniels/thermal-fans/all/initrd-tools_0.1.70ubuntu10_all.deb
# dpkg -i initrd-tools_0.1.70ubuntu10_all.deb
# apt-get install --reinstall <email address hidden>

Did this on an already installed system after I asked <email address hidden>
how to do it (just skip the chroot part),
Sadly there no change, fans still not working.

I also tried what <email address hidden> suggested to remove the thermal and
then modprobe it again.

With the original initrd i was able to remove the module and insert it again...
Box rebooted itself after yelling
the cpu temp was 256C (which i highly doubt).

With the initrd from Daniel i wasnt able to rmmod thermal.ko.

[Daniel Stone (daniels)]

Sounds like you're actually experiencing Bug#8659 (the 256C thing).

This bug has been marked as a duplicate of bug 8659.