On Tuesday 01 Aug 2023 at 09:45:22 (+0000), Quentin Perret wrote:
> Hi David,
>
> On Monday 31 Jul 2023 at 10:46:09 (-0700), David Dai wrote:
> > +static unsigned int virt_cpufreq_set_perf(struct cpufreq_policy *policy)
> > +{
> > + struct virt_cpufreq_drv_data *data = policy->driver_data;
> > + /*
> > + * Use cached frequency to avoid rounding to freq table entries
> > + * and undo 25% frequency boost applied by schedutil.
> > + */
>
> The VMM would be a better place for this scaling I think, the driver
> can't/shouldn't make assumptions about the governor it is running with
> given that this is a guest userspace decision essentially.
>
> IIRC the fast_switch() path is only used by schedutil, so one could
> probably make a case to scale things there, but it'd be inconsistent
> with the "slow" switch case, and would create a fragile dependency, so
> it's probably not worth pursuing.
Alternatively we could make the schedutil margin configurable via the
cmdline or something along those lines, so we can set it to 0 in the
guest and avoid the issue entirely.
Some partners have been asking for this IIRC , so I suspect there would
be interest from other parties.
Thanks,
Quentin
