[PATCH 2/7] dt-bindings: gpu: mali-utgard: add optional supply regulator

Heiko Stuebner heiko at sntech.de
Fri Sep 15 02:01:45 PDT 2017


Am Mittwoch, 13. September 2017, 09:34:15 CEST schrieb Rob Herring:
> On Wed, Sep 13, 2017 at 12:12:24AM +0200, Heiko Stübner wrote:
> > Hi Rob,
> > 
> > Am Dienstag, 12. September 2017, 17:09:20 CEST schrieb Rob Herring:
> > > On Tue, Sep 05, 2017 at 12:57:34PM +0200, Heiko Stuebner wrote:
> > > > Mali GPUs have a separate supplying regulator in a lot of socs,
> > > > so describe a mali-supply property. The already described
> > > > operating points will likely also need access to this regulator.
> > > > 
> > > > Signed-off-by: Heiko Stuebner <heiko at sntech.de>
> > > > ---
> > > > 
> > > >  Documentation/devicetree/bindings/gpu/arm,mali-utgard.txt | 3 +++
> > > >  1 file changed, 3 insertions(+)
> > > > 
> > > > diff --git a/Documentation/devicetree/bindings/gpu/arm,mali-utgard.txt
> > > > b/Documentation/devicetree/bindings/gpu/arm,mali-utgard.txt index
> > > > 3b7f6f72f032..bcaa640c883f 100644
> > > > --- a/Documentation/devicetree/bindings/gpu/arm,mali-utgard.txt
> > > > +++ b/Documentation/devicetree/bindings/gpu/arm,mali-utgard.txt
> > > > 
> > > > @@ -39,6 +39,9 @@ Optional properties:
> > > >      Memory region to allocate from, as defined in
> > > >      Documentation/devicetree/bindi/reserved-memory/reserved-memory.txt
> > > > 
> > > > +  - mali-supply : Phandle to regulator for the Mali device. Refer to
> > > > +    Documentation/devicetree/bindings/regulator/regulator.txt for
> > > > details.
> > > 
> > > Wouldn't a power domain be more appropriate?
> > 
> > At least on Rockchip socs there is a power-domain, but also the separate 
> > additional regulator. See the similar mali-midgard binding.
> 
> And that regulator's state is independent of the power domain's state? 
>
> But I guess OPPs need a regulator. Really we should allow OPPs to be 
> tied to the power domain. Maybe we do, I can't keep up with the ever 
> evolving PM stuff.

Yep, the external gpu regulator state is completely independent of the
power-domain inside the soc and is most of the time one of the regulators
coming from the external pmic or sometimes even a completely separate
i2c-connected ic.

It's the same for the cpu cores on Rockchip socs, they have power-domains
for controlling the actual on/off state and need separate supplying
i2c-connected regulators that still have to be enabled on their own.


> So, given we already have it for midgard,
> 
> Acked-by: Rob Herring <robh at kernel.org>

Thanks, added for v2.




More information about the Linux-rockchip mailing list