On Thu, Nov 30, 2017 at 09:56:08AM -0600, Andrew F. Davis wrote:
> On 11/30/2017 06:20 AM, Mark Brown wrote:

> > disable them.  Based on what I think the intention is I'd suggest
> > removing the bias level handling and then having probe enable runtime
> > PM with the device flagged as active, letting runtime PM do any
> > disabling if the device is idle.

> I beleive this was meant to be be SIMPLE_DEV_PM_OPS and not
> SET_RUNTIME_PM_OPS. I'll fix this all up for v3.

I was wondering that.

> Just thinking, the sound core sets SND_SOC_BIAS_OFF before suspend
> anyway, right? So the results would be similar just having all the PM
> stuff in the bias level handling for consistency, but I'm open to
> whatever is the preferred way.

It doesn't matter that much, if you do it only in set_bias_level() then
unless you set idle_bias_off there will be no runtime PM which may or
may not be what you want and you'll also not give the user the ability
to control if runtime PM happens via the sysfs files but I'm not
convinced that anyone ever actually does that.  Either approach is fine
really.