On Mon, Apr 08, 2019 at 06:17:18AM +0000, Mickael GUENE wrote: > Hi Sakari, > > On 4/6/19 13:01, Sakari Ailus wrote: > > On Tue, Mar 26, 2019 at 02:36:41PM +0000, Mickael GUENE wrote: > >> Sakari, > >> > >>>> +static int bpp_from_code(__u32 code) > >>>> +{ > >>>> + switch (code) { > >>>> + case MEDIA_BUS_FMT_SBGGR8_1X8: > >>>> + case MEDIA_BUS_FMT_SGBRG8_1X8: > >>>> + case MEDIA_BUS_FMT_SGRBG8_1X8: > >>>> + case MEDIA_BUS_FMT_SRGGB8_1X8: > >>>> + return 8; > >>>> + case MEDIA_BUS_FMT_SBGGR10_1X10: > >>>> + case MEDIA_BUS_FMT_SGBRG10_1X10: > >>>> + case MEDIA_BUS_FMT_SGRBG10_1X10: > >>>> + case MEDIA_BUS_FMT_SRGGB10_1X10: > >>>> + return 10; > >>>> + case MEDIA_BUS_FMT_SBGGR12_1X12: > >>>> + case MEDIA_BUS_FMT_SGBRG12_1X12: > >>>> + case MEDIA_BUS_FMT_SGRBG12_1X12: > >>>> + case MEDIA_BUS_FMT_SRGGB12_1X12: > >>>> + return 12; > >>>> + case MEDIA_BUS_FMT_UYVY8_2X8: > >>> > >>> This is good for the parallel bus, but on CSI-2 side you should have > >>> MEDIA_BUS_FMT_UYVY8_1X16 instead. This isn't technically correct for a > >>> serial bus, but the custom is to use the one sample / pixel formats on the > >>> serial busses. > >>> > >> Should MEDIA_BUS_FMT_BGR888_1X24 be something like > >> MEDIA_BUS_FMT_BGR888_3X8 for parallel output bus ? > > > > Good point. Yes. > > > > Could you add that format to > > Documentation/media/uapi/v4l/subdev-formats.rst, please, in a separate > > patch? > > > I also need to add it to include/uapi/linux/media-bus-format.h I > suppose ? Yes, that's correct. > I was also wondering if I should add MEDIA_BUS_FMT_BGR888_3X8_BE and > MEDIA_BUS_FMT_BGR888_3X8_LE instead ? If it's either, then "BE" but in this case I think it's better be without the explicit endianness as the endianness change results in pixel order change --- the format itself remains valid. I wonder what others think, too. Hans, any opinion? > > So if you are ok with the above I will add MEDIA_BUS_FMT_BGR888_3X8 to > subdev-formats.rs and media-bus-format.h as a first parch of my serie. -- Regards, Sakari Ailus sakari.ailus@xxxxxxxxxxxxxxx