On Fri, Mar 08, 2019 at 02:07:51PM +0100, Ulf Hansson wrote:
> On Fri, 8 Mar 2019 at 12:49, Lorenzo Pieralisi
> <lorenzo.pieralisi@xxxxxxx> wrote:
> >
> > On Fri, Mar 08, 2019 at 11:36:49AM +0100, Ulf Hansson wrote:
> >
> > [...]
> >
> > > Instead, my suggestion is according to what I propose in patch 4 and
> > > $subject patch, which means minor adjustments to be able to pass the
> > > struct cpuidle_driver * to the init functions. This, I need it for
> > > next steps, but already at this point it improves things as it avoids
> > > some of the OF parsing, and that's good, isn't it?
> >
> > I will take the patches Mark ACKed and send them for v5.2 as
> > early as it gets in v5.1-rc* cycle.
>
> Actually, may I suggest we funnel these through Rafael's tree, unless
> you are expecting other PSCI changes for v.5.2, which could cause
> conflicts?
>
> The reason is, other PM core changes, to genpd for example, needs to
> go via Rafael's tree. Those would then potentially block us for
> applying any other changes to your tree (arm-soc?) for PSCI (as there
> is dependency) until v5.3.
>
> How about if you provides your explicit acks for those PSCI changes
> your are happy with, then Rafael can pick them?
It is fine we can do that, I would have not sent the patches Mark
has ACKed to arm-soc till -{rc2/rc3} anyway.
Thanks,
Lorenzo
> > For this one maybe you can post the changes on top and see what's
> > the best way forward ?
> >
> > I agree that duplicating idle state parsing code across back-ends
> > is silly - we just want to keep PSCI and kernel data structure
> > decoupled.
>
> Right. Let's continue this discussion later on and move forward here.
> Make sense.
>
> >
> > Post the code on top and we will find a way forward, OK ?
>
> Sure, let me do that.
>
> Thanks and kind regards
> Uffe
