Well, I gave two alternatives :-) Both are fine as far as I am concerned, but it would be nice to hear what others think. Regards, Hans On 11/04/14 11:45, Boris Brezillon wrote: > Hi Hans, > > On Tue, 04 Nov 2014 11:20:40 +0100 > Hans Verkuil <hansverk@xxxxxxxxx> wrote: > >> Hi Boris, >> >> On 11/04/14 10:54, Boris Brezillon wrote: >>> Rename mediabus formats and move the enum into a separate header file so >>> that it can be used by DRM/KMS subsystem without any reference to the V4L2 >>> subsystem. >>> >>> Old V4L2_MBUS_FMT_ definitions are now referencing MEDIA_BUS_FMT_ value. >> >> I missed earlier that v4l2-mediabus.h contained a struct as well, so it can't be >> deprecated and neither can a #warning be added. >> >> The best approach, I think, is to use a macro in media-bus-format.h >> that will either define just the MEDIA_BUS value when compiled in the kernel, or >> define both MEDIA_BUS and V4L2_MBUS values when compiled for userspace. >> >> E.g. something like this: >> >> #ifdef __KERNEL__ >> #define MEDIA_BUS_FMT_ENTRY(name, val) MEDIA_BUS_FMT_ # name = val >> #else >> /* Keep V4L2_MBUS_FMT for backwards compatibility */ >> #define MEDIA_BUS_FMT_ENTRY(name, val) \ >> MEDIA_BUS_FMT_ # name = val, \ >> V4L2_MBUS_FMT_ # name = val >> #endif > > Okay, but this means we keep adding V4L2_MBUS_FMT_ definitions even for > new formats (which definitely doesn't encourage people to move on). > Moreover, we add a V4L2 prefix in what was supposed to be a subsystem > neutral header. > > Anyway, these are just nitpicks, and if you prefer this approach > I'll rework my series :-). > >> >> An alternative approach is to have v4l2-mediabus.h include media-bus-format.h, >> put #ifndef __KERNEL__ around the enum v4l2_mbus_pixelcode and add a big comment >> there that applications should use the defines from media-bus-format.h and that >> this enum is frozen (i.e. new values are only added to media-bus-format.h). >> >> But I think I like the macro idea best. > > As you wish, my only intent is to use those bus format definitions in a > DRM driver :-). > > Thanks, > > Boris > > _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
