On Mon, Feb 11, 2019 at 5:32 PM Fabrice Gasnier <fabrice.gasnier@xxxxxx> wrote:
>
> On 1/28/19 2:20 PM, Fabrice Gasnier wrote:
> > On 1/16/19 4:11 PM, Arnd Bergmann wrote:
> >> On Wed, Jan 16, 2019 at 3:10 PM Fabrice Gasnier <fabrice.gasnier@xxxxxx> wrote:
> >>
> >> What I mean is that nodes that listed as 'compatible="syscon"' get
> >> probed by the syscon driver even when no other driver references
> >> them, and that in turn would acquire the clock, right?
> >
> > When no other driver references them, nothing happens at probe time on
> > the clock: no calls to get/prepare... the clock.
> >
> > => The clock will remain unrequested & unused until another driver calls
> > one of "of_syscon_register()" variants:
> > - syscon_node_to_regmap
> > - syscon_regmap_lookup_by_compatible
> > - syscon_regmap_lookup_by_phandle
> >
> > When another driver references them (e.g. one of the above calls), then
> > it will acquire the optional clock and use it, e.g.:
> > - clk_prepare() upon of_syscon_register() variants
> > - clk_enable & clk_disable when accessing the registers
> >
> > I hope this clarifies.
>
> I would appreciate to have your feedback.
Yes, I think that's all we need here, thanks for the clarification,
and sorry for dropping the ball on this again.
Acked-by: Arnd Bergmann <arnd@xxxxxxxx>
Arnd
