Re: [Update PATCH 1/1] Cpufreq: Make governor data on nonboot cpus across system suspend/resume

["Rafael J. Wysocki" <rjw@rjwysocki.net>]

On Saturday, November 16, 2013 08:27:07 PM Viresh Kumar wrote:
> On 16 November 2013 20:11, Rafael J. Wysocki <rjw@rjwysocki.net> wrote:
> > On Saturday, November 16, 2013 11:59:59 AM Lan Tianyu wrote:
> 
> >> Defaultly, all cpus use ondemand governor after bootup. Change one
> >> non-boot cpu's governor to conservative,
> >
> > Well, why would anyone want to do that?  Just out of curiosity ...
> 
> People may want to use different group/cluster/socket of CPUs differently,
> with different kind of policies. Maybe performance governor for boot cpu
> and ondemand for others.
> 
> This bug would also be there for big LITTLE where we want to have
> separate set of tunables for big and LITTLE clusters for the same type
> of governor.
> 
> > So this is acpi-cpufreq, right?
> 
> Probably yes, I saw something similar somewhere.. But this is driver
> independent..
> 
> > The patch looks basically OK to me, but ->
> 
> We wouldn't need this patch if my other patch (where I am disabling
> governors in suspend/resume goes in, in any form)..

Yes, I know that, but I don't think this is the right approach.

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.