[Intel-gfx] [PATCH 1/2] drm/property: Enforce more lifetime rules

[Thierry Reding]

On Thu, Oct 24, 2019 at 12:40:55PM +0200, Thierry Reding wrote:
> On Wed, Oct 23, 2019 at 04:49:52PM +0200, Daniel Vetter wrote:
> > Properties can't be attached after registering, userspace would get
> > confused (no one bothers to reprobe really).
> > 
> > - Add kerneldoc
> > - Enforce this with some checks. This needs a somewhat ugly check
> >   since connectors can be added later on, but we still need to attach
> >   all properties before they go public.
> > 
> > Note that we already enforce that properties themselves are created
> > before the entire device is registered.
> > 
> > Cc: Jani Nikula <jani.nikula at linux.intel.com>
> > Cc: Rajat Jain <rajatja at google.com>
> > Signed-off-by: Daniel Vetter <daniel.vetter at intel.com>
> > ---
> >  drivers/gpu/drm/drm_mode_object.c | 14 ++++++++++++++
> >  1 file changed, 14 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/drm_mode_object.c b/drivers/gpu/drm/drm_mode_object.c
> > index 6a23e36ed4fe..35c2719407a8 100644
> > --- a/drivers/gpu/drm/drm_mode_object.c
> > +++ b/drivers/gpu/drm/drm_mode_object.c
> > @@ -224,12 +224,26 @@ EXPORT_SYMBOL(drm_mode_object_get);
> >   * This attaches the given property to the modeset object with the given initial
> >   * value. Currently this function cannot fail since the properties are stored in
> >   * a statically sized array.
> > + *
> > + * Note that all properties must be attached before the object itself is
> > + * registered and accessible from userspace.
> >   */
> >  void drm_object_attach_property(struct drm_mode_object *obj,
> >  				struct drm_property *property,
> >  				uint64_t init_val)
> >  {
> >  	int count = obj->properties->count;
> > +	struct drm_device *dev = property->dev;
> > +
> > +
> > +	if (obj->type == DRM_MODE_OBJECT_CONNECTOR) {
> > +		struct drm_connector *connector = obj_to_connector(obj);
> > +
> > +		WARN_ON(!dev->driver->load &&
> > +			connector->registration_state == DRM_CONNECTOR_REGISTERED);
> > +	} else {
> > +		WARN_ON(!dev->driver->load && dev->registered);
> > +	}
> 
> I'm not sure I understand why dev->driver->load needs to be a special
> case. Don't the same rules apply for those drivers as well?

Nevermind, I just noticed that drm_dev_register() sets dev->registered
to true before calling the driver's ->load() implementation, so makes
sense:

Reviewed-by: Thierry Reding <treding at nvidia.com>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20191024/4bbdd798/attachment.sig>