Hi Nicolas, On 9/6/2024 7:49 PM, Nicolas Dufresne wrote: > Hi, > > Le lundi 02 septembre 2024 à 03:02 +0300, Dmitry Baryshkov a écrit : >> On Sat, Aug 31, 2024 at 04:18:35PM GMT, Bryan O'Donoghue wrote: >>>> The result of v4l2-compliance test on SM8250: >>>> >>>> v4l2-compliance 1.29.0-5239, 64 bits, 64-bit time_t >>>> v4l2-compliance SHA: a1ebb4dad512 2024-08-08 20:00:17 >>>> >>>> Compliance test for iris_driver device /dev/video0: >>>> >>>> Driver Info: >>>> Driver name : iris_driver >>>> Card type : iris_decoder >>> >>> Hmm, so this is decoder only ? >>> >>> What's the intention here for encoding support ? >>> >>> I've verified your results on the test branch but I just noticed that sm8250 >>> with the iris driver is decoder only - whereas the venus driver does both, >>> which strikes me as a bit odd. >> >> I think we all have discussed this during the review of the previous >> series: complete driver becomes very huge and complicated to review. So >> the recommendation was to submit the limited features driver (decoding, >> 1 codec) and get more features (more codecs, encoding support, etc) >> after getting the Iris driver in. Of course sm8250 support in Venus >> driver will stay in until Iris driver reaches feature parity. > > And demonstrated backward compatibility. Just mentioning, since flipping default > driver could theoretically break userspace, and that is a big no no. As > maintainers won't run your HW, it will work on trust, and users report. But you > have to be convincing. We are committed to bring this driver at feature parity with venus. This will be in phased approach and started with (single codec)decoder. >>> >>> Is your intention to publish more patches to enable the encoder in another >>> series ? > > I think this was the real question, any reason not immediately replying ? What > I'm hoping is an answer in the form "yes" + some ETA. Encoder changes were already available and were dropped from this series to make it easier to review. ETA would be as soon as this series which adds the decoder part, gets accepted. > > Nicolas > >>> >>> --- >>> bod >> Regards, Vikash
