Re: [PATCH v3 0/7] Fix RK3588 GPU domain

[Sebastian Reichel]

Hi,

On Fri, Nov 01, 2024 at 10:41:14PM +0800, Chen-Yu Tsai wrote:
> On Fri, Nov 1, 2024 at 10:36 PM Mark Brown <broonie@xxxxxxxxxx> wrote:
> > On Fri, Nov 01, 2024 at 12:56:16PM +0100, Ulf Hansson wrote:
> > > On Wed, 23 Oct 2024 at 12:05, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> >
> > > > The merge strategy seems reasonable to me. But I am fine with that
> > > > whatever works for Mark.
> >
> > > Mark, any update on this?
> >
> > > If easier, you could also just ack the regulator patch (patch1), and
> > > can just take it all via my tree.
> >
> > I'm still deciding what I think about the regulator patch, I can see why
> > it's wanted in this situation but it's also an invitation to misuse by
> > drivers just blindly requesting all supplies and not caring if things
> > work.
> 
> I suppose an alternative is to flag which power domains actually need
> a regulator supply. The MediaTek power domain driver does this.

If you look at patch 6/7, which actually makes use of devm_of_regulator_get()
you will notice that I did actually flag which power domains have/need a
regulator.

> There's still the issue of backwards compatibility with older device
> trees that are missing said supply though.

Exactly :)

As far as I can see the same misuse potential also exists for the
plain devm_regulator_get() version.

Greetings,

-- Sebastian

Attachment: signature.asc
Description: PGP signature