Another acpi_thermal nit

[Kevin Oberman]

I've recently run into another issue with acpi_thermal. I don't know if
it can be fixed or is inherent in the design of acpi_thermal.

If I build openoffice.org (even better than buildworld for stress
testing a system), acpi_thermal will kick in and slow my processor,
eventually by two steps from 1.8G to 1.35G. This is fine and dandy, but
the build takes over 10 hours (I've not timed it recently) and I wanted
to go home. For the 50-80 minute trip I usually just slow the CPU to
about 600M and my battery will easily last for the trip. But once
acpi_thermal has kicked in, I can no longer reduce the CPU speed. I was
stuck at 1.35G.

I worked around the problem by entering ^S into the terminal window where
the build was running and waiting for the build to stop when it blocked
on the prints and the CPU cooled to where acpi_thermal dropped out and I
could again control the system speed. Then ^Q and head for home.

It would really be nice to be able to slow the system below where
acpi_thermal has lowered it. I just don't know if this is a matter of
code or a BIOS issue with no way out.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634