On Sun, Jun 10, 2007 at 03:47:01AM -0400, Joey Hess wrote: > Package: cpufreqd > Version: 2.1.1-1 > Severity: wishlish > > I'd kinda like to include this in the desktop task, but in my tests, it > doesn't behave well on some desktop machines. On my celeron (dodo), > it defaulted to powersave low mode, since it decided the machine had > both no AC adaptor (acpi doesn't show info for one), and no battery, > which I think it took to mean low battery.
Hummm... actually cpufreqd should be able to detect this situation correctly and simply unload the battery and ac modules (ok, these are not separate modules but are managed as if they were) and ignore all the acpi_battery and acpi_ac related directives. After that cpufreqd discards rules that are empty (most of them in your case with the default config). Can you try to collect some verbose log (cpufreqd -D -V7) of the initialization/parsing phase to better understand what's going on? Ah, even better: enable the remote control (enable_remote=1 in th econfig) and start it in manual mode (-m) so it will stop just after parsing the file. Then you can play with cpufreqd-set to switch it to dynamic mode and let it apply the rule. > I wonder if it might be best to have two config files (desktop, laptop) > and somehow choose which one to install based on laptop-detect. Hummmm... I'd like to see if it's possible to have a single config that works for both. What kind of use-case do you have in mind for desktop usage? thanks -- mattia :wq!
signature.asc
Description: Digital signature
