Hi Sakari, On Wed, Mar 18, 2020 at 12:02:33AM +0200, Sakari Ailus wrote: > On Tue, Mar 17, 2020 at 03:06:01PM +0200, Laurent Pinchart wrote: > ... > > > I have not followed CAP_IO_MC closely, but I > > > wonder if that could help catching that situation in the core and > > > return an error earlier. Maybe there could be a way for drivers to > > > advertise support for that feature to the core and fail earlier if > > > mbus_code is set and they don't claim to support it ? > > > > I was thinking of linking the two, making this extension mandatory for > > drivers that advertise CAP_IO_MC, in which case we could indeed drop the > > flag. > > > > Hans, what's your preference ? > > My mild preference would be towards binding this to CAP_IO_MC flag. If that's Hans' preference too, I'll rework it that way. It however means that both will need to be merged in the same kernel release. -- Regards, Laurent Pinchart
