Hi Sakari, On Thu, Nov 26, 2020 at 12:22:05PM +0200, Sakari Ailus wrote: > Hi Jacopo, > > On Thu, Nov 26, 2020 at 11:12:51AM +0100, Jacopo Mondi wrote: > > Hi Niklas, Sakari, > > > > On Wed, Nov 25, 2020 at 05:44:45PM +0100, Niklas Söderlund wrote: > > > Hello, > > > > > > This series aims to remove a V4L2 helper that provides a too simple > > > implementation, v4l2_async_notifier_parse_fwnode_endpoints_by_port(). > > > Instead drivers shall implement what the helper does directly to get > > > greater control of the process. There is only one user left in tree of > > > this interface, R-Car VIN. > > > > What is the plan going forward ? > > removing v4l2_async_notifier_parse_fwnode_endpoints_by_port() here > > then remove v4l2_async_notifier_parse_fwnode_endpoints() as it has a > > single user in mainline too ? > > > > Are we standardizing all platform drivers to use > > v4l2_async_notifier_add_fwnode_subdev() and perform fwnode.match > > initialization by themselves or is there a plan to replace > > Yes, please. > > > v4l2_async_notifier_parse_fwnode_endpoints*() with something else ? > > That's always been somewhat clunky and required special cases. The other > option, i.e. what this patchset does, is straightforward as well as allows > setting defaults in drivers, and admittedly, comes with a little bit of > extra code in drivers in areas that are driver specific. The old functions > such as v4l2_async_notifier_parse_fwnode_endpoints() just pretended they > were not... Agreed in full :) (At the expense of a little extra code in drivers) > > -- > Regards, > > Sakari Ailus
