On Thu, Aug 24, 2017 at 12:10:05AM +0800, Dong Aisheng wrote:
> It's used for platforms where different OPPs may use different clocks.
> With this extended binding, user could specify the correct clock for each
> OPP node.
> 
> Cc: Viresh Kumar <[email protected]>
> Cc: Nishanth Menon <[email protected]>
> Cc: Stephen Boyd <[email protected]>
> Cc: "Rafael J. Wysocki" <[email protected]>
> Cc: Rob Herring <[email protected]>
> Cc: Frank Rowand <[email protected]>
> Cc: [email protected]
> Signed-off-by: Dong Aisheng <[email protected]>
> ---
>  Documentation/devicetree/bindings/opp/opp.txt | 52 
> +++++++++++++++++++++++++++
>  1 file changed, 52 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/opp/opp.txt 
> b/Documentation/devicetree/bindings/opp/opp.txt
> index e36d261..40a4340 100644
> --- a/Documentation/devicetree/bindings/opp/opp.txt
> +++ b/Documentation/devicetree/bindings/opp/opp.txt
> @@ -152,6 +152,11 @@ Optional properties:
>    hierarchy can be contained in multiple 32 bit values. i.e. <X Y Z1 Z2> in 
> the
>    above example, Z1 & Z2 refer to the version hierarchy Z.
>  
> +- clocks: Clock phandle and specifier used for this opp.
> +
> +- clock-names: clock names for this opp. The valid clock names are platform
> +            specific.

You don't need -names if there's only 1 clock. But then how long until 
we have a list of any random clocks some how associated with an OPP.

I think this should really be solved within the clock framework. What 
you really need is "set my parent clock to the source that can provide X 
Hz". Could the assigned-clocks property work here (in the OPP nodes 
rather than the cpu nodes)?

> +
>  - status: Marks the node enabled/disabled.
>  
>  Example 1: Single cluster Dual-core ARM cortex A9, switch DVFS states 
> together.
> @@ -528,3 +533,50 @@ Example 6: opp-microvolt-<name>, opp-microamp-<name>:
>               };
>       };
>  };
> +
> +Example 7: Single core ARM cortex A7, switch separate clocks for each OPP:

Can't you add this to an existing example? We don't need to enumerate 
every possible option.

Rob

Reply via email to