Re: [PATCH v4 0/5] Add required-opps support to devfreq passive gov
From: Chanwoo Choi
Date: Thu Dec 19 2019 - 11:18:23 EST
Hi Saravana,
2019ë 11ì 14ì (ë) ìí 5:36, Chanwoo Choi <cw00.choi@xxxxxxxxxxx>ëì ìì:
>
> Hi Saravana,
>
> On 11/14/19 5:23 PM, Saravana Kannan wrote:
> > On Wed, Nov 13, 2019 at 11:48 PM Chanwoo Choi <cw00.choi@xxxxxxxxxxx> wrote:
> >>
> >> Dear Saravana,
> >>
> >> Any other progress of this series?
> >
> > Hi Chanwoo,
> >
> > Thanks for checking. I haven't abandoned this patch series. This patch
> > series depends on "lazy linking" of required-opps to avoid a cyclic
> > dependency between devfreq and OPP table population. But Viresh wasn't
> > happy with my implementation of the lazy liking for reasonable
> > reasons.
> >
> > I had a chat with Viresh during one of the several conferences that I
> > met him at. To fix the lazy linking in the way he wanted it meant we
> > had to fix other issues in the OPP framework that arise when OPP
> > tables are shared in DT but not in memory. So he was kind enough to
> > sign up to add lazy linking support to OPPs so that I won't have to do
> > it. So, I'm waiting on that. So once that's added, I should be able to
> > drop a few patches in this series, do some minor updates and then this
> > will be good to go.
>
> Thanks for the detailed explanation. I'll expect the your next version.
As I know, the lazy linking issue was fixed by Viresh.
If possible, I want to know your plan about 'required-opp' with
passive governor.
Because I think that the patch[1] is good for devfreq device.
As you mentioned, you have the other idea to implement the 'cpu based
scaling support'
to passive governor without cpu notifier. Actually, if there are no
any other best solution,
I prefer to use cpu notifier for 'cpu based scaling support to
passive_governor'.
[1] https://patchwork.kernel.org/patch/11046147/
- [RFC,v2] PM / devfreq: Add cpu based scaling support to passive_governor
--
Best Regards,
Chanwoo Choi