Comment 127 for bug 22336

I first experienced that bug on kernel 2.6.20.14 - where frequency scaling begin to work on my IBM T43p.
Now when having the cpu scaled to 100%, my laptop will halt due to reaching cpu temperature limit.

the problem seems to be the scaling-steps in /proc/acpi/ibm/fan

I can manually speed the fan up by "echo level 7 > /proc/acpi/ibm/fan"
which will bring the speed up to "4263"
on the lowest level (level 1) the speed is quoted "3268"

Comparing the noise with the fan-noise while running windows, level 7 does not seem to be the maximum at all.

When having fan at auto-level at heavy cpu-load, fan-speed is increasing slowly, but will never be over speed "4263".

When having the fan in disengaged mode (echo 0x2F 0x40 > /proc/acpi/ibm/ecdump), it will speed up to it's full speed "6859".

Why is fan speed controlled by ibm-acpi and not by the bios?

Running feisty (2.6.20-15-386) on a T43p 1,86 GHz Pentium M.