This patch series is based on this earlier RFC series: https://www.mail-archive.com/linux-media@xxxxxxxxxxxxxxx/msg83415.html This patch series starts the work of removing several video ops that are duplicated in as pad ops in the subdev kernel API. This kills the whole point of the subdev API which is to enable reuse of subdevices by different bridge drivers. With duplicate ops bridge drivers now either have to check which one to use, or (and that's what happens today) they make an assumption and just use one of the two. Duplicate ops should never have been allowed. A lesson for the future. This patch series gets rid of two duplicate ops (the enum_framesizes and enum_frameintervals video ops) and prepares the way to remove the remaining crop video ops. The patch that does that isn't part of this patch series as it needs more work (testing primarily). The first patch of this series lays the foundation: the pad ops use the v4l2_subdev_fh for the try formats. However, bridge drivers that want to call the pad ops do not have a v4l2_subdev_fh struct available. It's only used when accessing the 'try' formats that are part of struct v4l2_subdev_fh. This patch splits off those fields into a separate struct and uses that instead of v4l2_subdev_fh. This breaks the link between a file handle and the try formats. If a bridge driver needs to deal with try formats, then it is the responsibility of the bridge driver to create a v4l2_subdev_pad_config array. In practice, bridge drivers for simple, non-MC devices only need to use the active format, and that doesn't need a v4l2_subdev_pad_config array, so NULL can be used there instead. However, this might change in the future. The second and third patch add a 'which' field to the enum struct used in the pad ops. Currently all enum pad ops assume that it is against the 'try' formats, but simple bridge drivers will want to enum against the active formats. So add support to select what you want. And frankly, I think it is more consistent and actually useful for application to see what the enumeration is against the active format. The next two patches add support for the new 'which' field where appropriate, the sixth patch removes the uses of the video enum ops and the final patch documents the new 'which' field. Regards, Hans -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
