Re: [PATCH 1/3] dt-bindings: power: Introduce 'assigned-performance-states' property

From: Stephen Boyd
Date: Wed Aug 05 2020 - 19:57:52 EST

Quoting Rajendra Nayak (2020-08-05 01:13:06)
> On 8/5/2020 12:09 PM, Stephen Boyd wrote:
> > Quoting Rajendra Nayak (2020-08-04 04:46:54)
> >
> >> + device's performance, also known as DVFS techniques. The list of performance
> >> + state values should correspond to the list of power domains specified as part
> >> + of the power-domains property.
> >
> > This is different than assigned-clock-rates. I guess that's OK because
> > we don't need to assign parents with more specifiers. Maybe it should be
> > worded more strongly to clearly state that each cell corresponds to one
> > power domain? And that it should match the opp-level inside any OPP
> > table for the power domain?
> Sure, I'll reword it to make it clear that we need the same number of cells
> as power-domains, and as you pointed out below that 0 corresponds to not setting
> anything.
> For the matching of opp-level inside the OPP table of the power-domain, I don't
> think from the power-domain bindings we limit providers with only OPP tables to
> support performance states? It could be just a range that the provider manages
> internally?

Ok. The example made it match so maybe that can be clarified as well
that it doesn't need to match any OPP table performance state.