Hi Sakari, On 6/20/19 5:26 PM, Sakari Ailus wrote: > Hi Hugues, > > On Tue, Jun 11, 2019 at 10:48:30AM +0200, Hugues Fruchet wrote: >> Add a new "sensor" field to dcmi struct instead of >> reusing entity->subdev to address sensor subdev. As discussed on IRC, fixed in v3, > > The purpose of the struct binding image source's async subdev as well as > related information is to allow associating the two. This patch breaks > that. If your device can support a single sensor, it might not be a big > deal. The end result remains somewhat inconsistent as subdev specific > information is spread across struct stm32_dcmi and struct > dcmi_graph_entity. As discussed on IRC, fixed in v3, > > In general you don't need to know the sensor as you can always find it > using media_entity_remote_pad(). This driver is a little different though > as it could presumably continue to work without MC. Was that the intent? > > On a side note: struct dcmi_graph_entity does NOT have struct > v4l2_async_subdev as its first member. Please fix that and prepend the fix > to this set. > As discussed on IRC, fixed in v3, BR, Hugues.
