The current init.d script doens't ensure that the proper CPU governors are present. The attached fix will try to load them, then ensure that they are in fact present, and bail out if not. This will only kick in if the script has determined that we're doing a 2.5/2.6-style CPUfreq.
Created attachment 25728 [details, diff] cpudyn.init-governor-fix.diff It may be that we should do something similar for ACPI, but I've not tested for this. The modprobe -q cpufreq_powersave stuff should be harmless. No output will be made if the module is already loaded or it's absent altogether (ie, the user has these compiled-in).
in cvs now.