On 02/23/16 17:32, Tomi Valkeinen wrote: > On 23/02/16 17:26, Jyri Sarha wrote: > >>> You didn't comment on why this is not an error? Why should the driver >>> continue even if crtc->port is missing? >>> >> >> At least for the time being if the drm_of_find_possible_crtcs() fails >> the tda998x driver assumes the first crtc with a warning. So for that >> part everything will work just fine still. >> >> Then it is another question how priv->is_componentized could be set and >> probing has gotten this far while there is no port node to be found. The >> WARN_ON() should really never happen as long as the code is the way it >> currently is. > > Ok. But I think it's either ok to not have crtc->port, and in that case > no print is needed, or it's not ok, and it's better to print an error > and fail. > > Now it's kind of vague: the driver continues without crtc->port, but > gives a scary WARN. >
The scary WARN is not for not having crtc->port initialized, but for breached internal sanity when a componentized probe has somehow reached this point without a port node to be found.