On Tue, Oct 08, 2019 at 08:05:03PM +0300, Dmitry Osipenko wrote:
> 08.10.2019 19:15, Mark Brown пишет:

> > That sounds like it might just postpone the inevitable - if you set the
> > wrong voltage first it might decide to drop down some voltage that
> > wasn't expected.  There's a bit of a bootstrapping issue.  I think it
> > would be safer to just say that anything that is within spec won't get
> > changed any time we balance, we'd only change things if needed to bring
> > them back into spec.

> Yes, the case of changing voltage before regulator is enabled seems
> won't work as expected.

> Maybe it won't hurt to disallow a non always-on regulators to be coupled
> until there will be a real user for that case.

I thought that coupling with the CPU core and main SoC power regulators
was one of the main use cases for this feature?