Li, Aubrey wrote: > Bill Holler wrote: > > >> Here is a list of things left to do for c-state support on systems >> with c-state invariant TSCs. The list is in no particular order: >> >> 1. Develop code to use the HPET to wakeup a deep c-state idle cpu. >> 2. Develop code to detect deep c-state support in intel_cpupm_init() >> similar to what speedstep_support() does for p-state support. >> 3. lAPIC timer fixup after deep c-state. >> >> Did I miss anything??? >> > > The c-state policy, now there is a simple one and ugly. > I'm not sure how it works with PAD, but I'd like to implement a new one > in the c-state driver. >
Go for it. :-) I am not sure what the time schedule is for PAD. We should implement something now for testing before PAD is available. The current c-state policy looks like it will work for now? >> I have #3 working using the HPET. It should be ready enough >> for review Monday. I am not sure if we will re-work it using the >> ACPI PM-timer? >> >> > This should be cheap and easy, we can handle this later. > We may find out the HPET is overloaded. Bill > Thanks, > -Aubrey >
