On Tue, Dec 03, 2019 at 07:29:45PM +0100, Geert Uytterhoeven wrote:
> On Tue, Dec 3, 2019 at 3:19 PM Mark Brown <broonie@kernel.org> wrote:

> > > +     pm_runtime_put(&pdev->dev);
> > > +     pm_runtime_disable(&pdev->dev);

> > There seems to be no purpose in the runtime PM code in this
> > driver, there's no PM operations of any kind and the driver holds
> > a runtime PM reference for the entire lifetime of the device.

> It matters for the clock domain (assumed the module clock is not always
> marked as a critical clock).

That seems like a problem with what the clock domains are doing
surely?  The default is supposed to be that if runtime PM isn't
enabled we get the behaviour the driver is manually implementing
here.  Besides, if this is actually impacting power management
I'd expect us to be letting the IP idle rather than holding it
powered up all the time.