On Tue, Sep 15, 2015 at 07:57:19PM +0300, Ville Syrjälä wrote:
> On Tue, Sep 15, 2015 at 09:27:27AM -0700, Rafael Antognolli wrote:
> > On Tue, Sep 15, 2015 at 10:46:43AM +0300, Ville Syrjälä wrote:
> > > On Mon, Sep 14, 2015 at 04:12:30PM -0700, Rafael Antognolli wrote:
> > > > This list will be used to get the aux channels registered through the
> > > > helpers. Two functions are provided to register/unregister notifier
> > > > listeners on the list, and another functiont to iterate over the list of
> > > > aux channels.
> > > >
> > > > Signed-off-by: Rafael Antognolli <rafael.antognolli@xxxxxxxxx>
> > > > ---
> > > > drivers/gpu/drm/drm_dp_helper.c | 71 +++++++++++++++++++++++++++++++++++++++++
> > > > include/drm/drm_dp_helper.h | 6 ++++
> > > > 2 files changed, 77 insertions(+)
> > > >
> > > > diff --git a/drivers/gpu/drm/drm_dp_helper.c b/drivers/gpu/drm/drm_dp_helper.c
> > > > index 291734e..01a1489 100644
> > > > --- a/drivers/gpu/drm/drm_dp_helper.c
> > > > +++ b/drivers/gpu/drm/drm_dp_helper.c
> > > > @@ -710,6 +710,54 @@ static const struct i2c_algorithm drm_dp_i2c_algo = {
> > > > .master_xfer = drm_dp_i2c_xfer,
> > > > };
> > > >
> > > > +struct drm_dp_aux_node {
> > > > + struct klist_node list;
> > > > + struct drm_dp_aux *aux;
> > > > +};
> > > > +
> > > > +static DEFINE_KLIST(drm_dp_aux_list, NULL, NULL);
> > > > +
> > > > +static BLOCKING_NOTIFIER_HEAD(aux_notifier);
> > > > +
> > > > +int drm_dp_aux_register_notifier(struct notifier_block *nb)
> > > > +{
> > > > + return blocking_notifier_chain_register(&aux_notifier, nb);
> > > > +}
> > > > +EXPORT_SYMBOL(drm_dp_aux_register_notifier);
> > >
> > > Why is this notifier stuff needed? Why not just register/unregister the
> > > aux-dev directly?
> > >
> >
> > I am not sure it's needed, I was just looking for the best way of
> > informing aux-dev that a new aux channel was added.
> >
> > By register/unregister the aux-dev directly, do you mean making this
> > drm_dp_helper aware of the aux dev, when it's registered, and directly
> > calling some callback to inform that new aux channels were added?
>
> That was my thought, yes. It would mean the auxdev module can't be
> unloaded like i2c-dev, but I'm not sure that's a use case worth
> worrying about.
Yeah imo notifiers are evil because of locking inversion issues that
usually grop up, and having a use-case to unload dp_aux-dev seems
unlikely. Maybe we can do a Kconfig knob or something like that if some
people want to make the dev nodes optional.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/dri-devel
