On Fri, Oct 30, 2020 at 01:55:14PM -0500, Rob Herring wrote:

> I'm a bit worried that now we're changing CPUs (at least?) from clocks 
> to 'performance domains' while at the same time here we're adding 
> low level, virtual regulators. Are we going to end up wanting something 
> more abstract here too?

My understanding is that this is aimed at systems which have done the
more abstract thing where regulators just aren't visible at all to the
kernel but then find that they actually need to control some of the
regulators explicitly for things like MMC so need a mechanism for the
firmware to expose select regulators.