Re: [PATCH 10/11] dt-bindings: misc: add bindings for throttler

[Matthias Kaehlcke]

On Thu, May 31, 2018 at 03:04:18PM -0500, Rob Herring wrote:
> On Thu, May 31, 2018 at 1:34 PM, Matthias Kaehlcke <mka@xxxxxxxxxxxx> wrote:
> > Hi Rob,
> >
> > On Thu, May 31, 2018 at 11:31:59AM -0500, Rob Herring wrote:
> >> On Fri, May 25, 2018 at 01:30:42PM -0700, Matthias Kaehlcke wrote:
> >>
> >> Commit msg?
> >
> > Will add some more info in the next revision.
> >
> >> > Signed-off-by: Matthias Kaehlcke <mka@xxxxxxxxxxxx>
> >> > ---
> >> >  .../devicetree/bindings/misc/throttler.txt    | 41 +++++++++++++++++++
> >> >  1 file changed, 41 insertions(+)
> >> >  create mode 100644 Documentation/devicetree/bindings/misc/throttler.txt
> >> >
> >> > diff --git a/Documentation/devicetree/bindings/misc/throttler.txt b/Documentation/devicetree/bindings/misc/throttler.txt
> >> > new file mode 100644
> >> > index 000000000000..92f13e94451a
> >> > --- /dev/null
> >> > +++ b/Documentation/devicetree/bindings/misc/throttler.txt
> >> > @@ -0,0 +1,41 @@
> >> > +Throttler driver
> >> > +
> >> > +The Throttler is used for non-thermal throttling of system components like
> >> > +CPUs or devfreq devices.
> >>
> >> This all looks very Linux specific and not a h/w device. Perhaps you can
> >> add hint properties to OPP tables as to what entries can be used for
> >> throttling, but otherwise this doesn't belong in DT.
> >
> > My idea is to allow multiple throttlers, which might operate on
> > different throttling devices or use different OPPs for the same
> > device. To support this a simple boolean hint that the OPP can be used
> > for throttling would not be sufficient.
> >
> > What should work is a property with an array of phandles of the
> > throttlers that use a given OPP.
> >
> > AFAIK it is currently not possible to enumerate the devfreq devices in
> > the system, so besides the info in the OPPs the throttler itself would
> > still need a phandle to the devfreq device(s) it uses.
> 
> Why don't you fix that OS problem instead of working around it in DT?

I can try, though it's not exclusively in my hands, depends on what
the devfreq maintainers think about it.

> > I envision something like this:
> >
> > gpu_opp_table: opp-table2 {
> >   compatible = "operating-points-v2";
> >
> >   opp00 {
> >     opp-hz = /bits/ 64 <200000000>;
> >     opp-microvolt = <800000>;
> >   };
> >   opp01 {
> >     opp-hz = /bits/ 64 <297000000>;
> >     opp-microvolt = <800000>;
> >     opp-throttlers = <&cros_ec_throttler>;
> >   };
> >   ...
> > };
> >
> > cros_ec_throttler: cros-ec-throttler {
> >   compatible = "google,cros-ec-throttler";
> 
> Is this an actual h/w device?

The Chrome OS Embedded Controller is a MCU that communicates with
Linux over SPI or I2C. The low-level communication with the EC is
handled by drivers/mfd/cros_ec* and then there are multiple drivers
representing 'remote devices' on top of this (rtc-cros-ec.c,
extcon-usbc-cros-ec.c, pwm-cros-ec.c, ...). cros_ec_throttler is a
similar 'device' that responds to signals from the EC with frequency
adjustments.