Re: [PATCH 0/7] Minor cleanup for thermal gov power allocator

[Lukasz Luba]

Hi Rafael,

Gentle ping

On 10/26/23 13:22, Lukasz Luba wrote:
> On 10/26/23 09:54, Rafael J. Wysocki wrote:
> > On Wed, Oct 25, 2023 at 9:21 PM Lukasz Luba <lukasz.luba@xxxxxxx> wrote:
> > > Hi all,
> > >
> > > The patch set does some small clean up for Intelligent Power Allocator.
> > > Those changes are not expected to alter the general functionality. 
> > > They just
> > > improve the code reading. Only patch 3/7 might improve the use case for
> > > binding the governor to thermal zone (very unlikely in real products, 
> > > but
> > > it's needed for correctness).
> > >
> > > The changes are based on top of current PM thermal branch, so with the
> > > new trip points.
> > >
> > > Regards,
> > > Lukasz
> > >
> > > Lukasz Luba (7):
> > >    thermal: gov_power_allocator: Rename trip_max_desired_temperature
> > >    thermal: gov_power_allocator: Setup trip points earlier
> > >    thermal: gov_power_allocator: Check the cooling devices only for
> > >      trip_max
> > >    thermal: gov_power_allocator: Rearrange the order of variables
> > >    thermal: gov_power_allocator: Use shorter variable when possible
> > >    thermal: gov_power_allocator: Remove unneeded local variables
> > >    thermal: gov_power_allocator: Clean needed variables at the beginning
> > >
> > >   drivers/thermal/gov_power_allocator.c | 123 ++++++++++++++------------
> > >   1 file changed, 64 insertions(+), 59 deletions(-)
> > >
> > > --
> >
> > The series looks good to me overall, but I'd prefer to make these
> > changes in the 6.8 cycle, because the 6.7 merge window is around the
> > corner and there is quite a bit of thermal material in this cycle
> > already.
>
> Thanks for having a look! Yes, I agree, we can wait after the
> merge window. It just have to be cleaned one day a bit and I postponed
> this a few times, so no rush ;)

I've seen you've created the new pm/thermal. Could you consider to take
those in, please?

I would send some RFC on top showing the issue with reading back the CPU
max frequency from the PM_QoS chain.

Regards,
Lukasz