kernel bug jaunty fan control gist of below post: fan control problem with jaunty kernel on Thinkpads. Note: My laptop never shuts down on me but the heat generated is definitely a regression from olders ubuntu. Well I thought so until I tested a lot of them. Generated thinkpad lshw for my thinkpad T61 (file thinkpad-t61-lshw). Running i386 arch. Attached dmidecode (file thinkpad-t61-dmidecode) and acpidump (file thinkpad-t61-acpidump). on kernel 2.6.28-13-generic ----------------------------------- grabbed sensors for normal workload (email, firefox, music, a term or two) check file sensors $KERNEL normal usage ~ 78C stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 90C (see sensors $KERNEL stress file) temperature is too hot to have laptop on lap. It is CLEAR that the fans are not spinning at their max speed. Since I had already enabled the manual fan control workaround mentionned in this post ("options thinkpad_acpi fan_control=1" in /etc/modprobe.d/options) I decided to disengage them to see if temperature would lower. It did. olivier@boreale:~$ sudo -s root@boreale:~# echo "level disengaged" > /proc/acpi/ibm/fan Doing so, the fan speed gradually raise from 3000 to around 5000 in a few seconds (see proc-acpi-ibm-fan linux-2.6.28-13 file). 5 minutes later still under stress, the temp was back at 78C (see sensors $KERNEL stress with manual fan control) and dropped even to 74C later on kernel 2.6.27-14-generic ----------------------------------- olivier@boreale:~$ uname -a Linux boreale 2.6.27-14-generic #1 SMP Fri Jun 5 10:14:59 UTC 2009 i686 GNU/Linux normal workload: ~76C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 89C (see sensors $KERNEL stress file) olivier@boreale:~$ sudo -s root@boreale:~# echo "level disengaged" > /proc/acpi/ibm/fan Doing so, the fan speed gradually raise from 3300 to around 5000 in a few seconds 5 minutes later still under stress, the temp was at 78C (see sensors $KERNEL stress with manual fan control) on mainline kernel 2.6.28-x ----------------------------------- olivier@boreale:~$ uname -a Linux boreale 2.6.28-02062810-generic #02062810 SMP Tue May 5 19:28:54 UTC 2009 i686 GNU/Linux normal workload: ~73C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 90C (see sensors $KERNEL stress file) olivier@boreale:~$ sudo -s root@boreale:~# echo "level disengaged" > /proc/acpi/ibm/fan Doing so, the fan speed gradually raise from 3300 to around 5100 in a few seconds 5 minutes later still under stress, the temp was at 77C (see sensors $KERNEL stress with manual fan control) Warning: below tests performed on a cooler day on mainline kernel 2.6.30-x ----------------------------------- olivier@boreale:~$ uname -a Linux boreale 2.6.30-020630-generic #020630 SMP Wed Jun 10 09:45:40 UTC 2009 i686 GNU/Linux normal workload: ~62C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 86C (see sensors $KERNEL stress file) olivier@boreale:~$ sudo -s root@boreale:~# echo "level disengaged" > /proc/acpi/ibm/fan Doing so, the fan speed gradually raise from 3300 to around 5100 in a few seconds 5 minutes later still under stress, the temp was at 76C (see sensors $KERNEL stress with manual fan control) on mainline kernel 2.6.27-x --------------------------- olivier@boreale:~$ uname -a Linux boreale 2.6.27-020627-generic #020627 SMP Sat Jan 31 17:24:52 UTC 2009 i686 GNU/Linux normal workload: ~78C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 89C (see sensors $KERNEL stress file) olivier@boreale:~$ sudo -s root@boreale:~# echo "level disengaged" > /proc/acpi/ibm/fan Doing so, the fan speed gradually raise from 3300 to around 5100 in a few seconds 5 minutes later still under stress, the temp was at 76C (see sensors $KERNEL stress with manual fan control) 8.10 LiveCD on kernel 2.6.27-7-generic -------------------------- # uname -a Linux ubuntu 2.6.27-7-generic #1 SMP Fri Oct 24 06:42:44 UTC 2008 i686 GNU/Linux normal workload: ~50C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 91C (see sensors $KERNEL stress file) Fan still only at 3000 rpms. I haven't tried with the fan disengaged because I haven't added the module param. 8.04.2 LiveCD on kernel 2.6.27-7-generic -------------------------- ubuntu@ubuntu:~$ uname -a Linux ubuntu 2.6.24-23-generic #1 SMP Thu Nov 27 18:44:42 UTC 2008 i686 GNU/Linux The output of sensors is different in this case. I took the highest temp. normal workload: ~51C (see sensors $KERNEL normal usage) stressed system (with stress --cpu 16 --vm 2 --vm-bytes 128M) after 10 minutes: ~ 87C (see sensors $KERNEL stress file) Fan still only at 3000 rpms. I haven't tried with the fan disengaged because I haven't added the module param. Conclusion: I am very confused. Results are mostly the same with a lot of different kernels and I was _sure_ that my fan kicked in on heavy workloads when using earlier ubuntu releases. Where to go from here? Should I: - Try to mess with my BIOS settings? - Try another distro livecd and check temp? - Mess with thinkpad_acpi params? I am willing to run any other thing you need for further investigation.