Hi Andy,
On Fri, Apr 12, 2024 at 04:43:43PM +0300, Andy Shevchenko wrote:
> On Fri, Apr 12, 2024 at 12:39 PM Sakari Ailus
> <sakari.ailus@xxxxxxxxxxxxxxx> wrote:
> > On Wed, Apr 10, 2024 at 07:00:02PM +0300, Andy Shevchenko wrote:
> > > > +static int gt97xx_open(struct v4l2_subdev *sd, struct v4l2_subdev_fh *fh)
> > > > +{
> > > > + return pm_runtime_resume_and_get(sd->dev);
> > > > +}
> > > > +
> > > > +static int gt97xx_close(struct v4l2_subdev *sd, struct v4l2_subdev_fh *fh)
> > > > +{
> > > > + return pm_runtime_put(sd->dev);
> > > > +}
> > >
> > > Hmm... Shouldn't v4l2 take care about these (PM calls)?
> >
> > Ideally yes. We don't have a good mechanism for this at the moment as the
> > lens isn't part of the image pipeline. Non-data links may be used for this
> > in the future but that's not implemented yet.
>
> Aren't you using devlinks? It was designed exactly to make sure that
> the PM chain of calls goes in the correct order.
Device links are already used by the IPU bridge, but in the other
direction: the VCM requires the sensor to be powered up in this case.
--
Regards,
Sakari Ailus
