Re: [PATCH] dt-bindings: cpufreq: cpufreq-qcom-hw: Document SM8350 CPUfreq compatible

[Vinod Koul <vkoul@kernel.org>]

On 05-03-21, 15:57, Rob Herring wrote:
> On Thu, Feb 18, 2021 at 09:18:20PM +0530, Viresh Kumar wrote:
> > On 18-02-21, 18:14, Vinod Koul wrote:
> > > On 17-02-21, 10:19, Viresh Kumar wrote:
> > > > On 16-02-21, 16:42, Vinod Koul wrote:
> > > > > Add the CPUfreq compatible for SM8350 SoC along with note for using the
> > > > > specific compatible for SoCs
> > > > > 
> > > > > Signed-off-by: Vinod Koul <vkoul@kernel.org>
> > > > > ---
> > > > >  Documentation/devicetree/bindings/cpufreq/cpufreq-qcom-hw.txt | 4 +++-
> > > > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > > > > 
> > > > > diff --git a/Documentation/devicetree/bindings/cpufreq/cpufreq-qcom-hw.txt b/Documentation/devicetree/bindings/cpufreq/cpufreq-qcom-hw.txt
> > > > > index 9299028ee712..3eb3cee59d79 100644
> > > > > --- a/Documentation/devicetree/bindings/cpufreq/cpufreq-qcom-hw.txt
> > > > > +++ b/Documentation/devicetree/bindings/cpufreq/cpufreq-qcom-hw.txt
> > > > > @@ -8,7 +8,9 @@ Properties:
> > > > >  - compatible
> > > > >  	Usage:		required
> > > > >  	Value type:	<string>
> > > > > -	Definition:	must be "qcom,cpufreq-hw" or "qcom,cpufreq-epss".
> > > > > +	Definition:	must be "qcom,cpufreq-hw" or "qcom,cpufreq-epss"
> > > > > +			along with SoC specific compatible:
> > > > > +			  "qcom,sm8350-cpufreq-epss", "qcom,cpufreq-epss"
> > > > 
> > > > And why is SoC specific compatible required here ? Is the implementation on
> > > > sm8350 any different than the ones using "qcom,cpufreq-epss" compatible ?
> > > > 
> > > > FWIW, the same compatible string must be reused until the time there is
> > > > difference in the hardware. The compatible string must be considered as a marker
> > > > for a particular version of the hardware.
> > > 
> > > Rob has indicated that we should use a SoC specific compatible and I
> > > agree with that. We are using both soc and generic one here and driver
> > > will be loaded for generic one.
> > 
> > I am not sure of the context, lets see what Rob has to say on this. I
> > believe we only need 1 compatible string here (whatever it is), as
> > this is just one version of the hardware we are talking about. We
> > already have 2 somehow and you are trying to add one more and I don't
> > fell good about it. :(
> 
> The h/w block is the same features and bugs in every single 
> implementation? If not sure, better be safe.
> 
> I don't know that I'd go back and add SoC ones for everything though.

I would prefer we have SoC ones to be future proof..

-- 
~Vinod