Hi Vladimir, On Thu, Apr 19, 2018 at 02:18:30PM +0300, Vladimir Zapolskiy wrote: > Hi Jacopo, > > On 04/10/2018 01:53 PM, Jacopo Mondi wrote: > > Add DRM bridge driver for Thine THC63LVD1024 LVDS to digital parallel > > output converter. > > > > Signed-off-by: Jacopo Mondi > > Reviewed-by: Andrzej Hajda > > Reviewed-by: Niklas Söderlund > > Reviewed-by: Vladimir Zapolskiy > Thanks. FYI I sent v9 yesterday with a minimal change compared to v8. > Generally I have only one pretty ignorable comment. > > > + > > +enum thc63_ports { > > + THC63_LVDS_IN0, > > + THC63_LVDS_IN1, > > + THC63_RGB_OUT0, > > + THC63_RGB_OUT1, > > +}; > > + > > The driver uses only THC63_RGB_OUT0 value, or port@2, and MODE{0,1,2} IC > configuration is ignored. > > I don't know if right from the beginning it would be better to support > dual-out modes, preferably both single-in and dual-in ones. Will it > impact port enumeration? The bindings have been designed to support dual in/out modes, as you can see there are 4 possible ports described there: Required video port nodes: - port@0: First LVDS input port - port@2: First digital CMOS/TTL parallel output Optional video port nodes: - port@1: Second LVDS input port - port@3: Second digital CMOS/TTL parallel output Future extension should not require changing the port enumeration, just add a property to specify the selected mode. > > I do understand that the extension is possible, and likely only hardware > accessibility postpones it. Yes, hardware on one side, but also what I think is a shortcoming of DRM (which exists in other sub-systems, say v4l2) that matches devices on their OF device nodes and makes cumbersome handling drivers wanting to register on 'port' nodes instead, as it would happen if you have 2 input endpoints. See my [1] note here: https://lkml.org/lkml/2018/3/9/422 And this reply to Archit's comment which has been left floating as it is not a real issue (yet): https://lkml.org/lkml/2018/3/10/214 Thanks j > > -- > With best wishes, > Vladimir