Hello, This small patch series addresses a bug related to the usage of v4l2_async_notifier_add_subdev() that is widespread among drivers. The issue is explained in 1/4, which improves the documentation of the v4l2_async_notifier_add_subdev() function by stating explicitly that the asd argument needs to be allocated dynamically. Among the 13 drivers that use that function, only one gets it right. The rest of the patches fix the problem in several Renesas-related drivers, with an unrelated fwnode reference leak fix for the rcar-drif driver in 2/5 that made the v4l2_async_notifier_add_subdev() fix easier to implement in that driver. I'm lacking hardware to test 2/5 and 3/5. Ramesh, would you be able to test that ? What development board do you use to test the DRIF driver ? I don't see any DT integration upstream. I also haven't dug up my MAX9286 development kit to test 5/5. I would thus appreciate if someone could test it, but worst case I can do so myself. Laurent Pinchart (5): media: v4l2-async: Document asd allocation requirements media: rcar_drif: Fix fwnode reference leak when parsing DT media: rcar_drif: Allocate v4l2_async_subdev dynamically media: rcar-csi2: Allocate v4l2_async_subdev dynamically media: i2c: max9286: Allocate v4l2_async_subdev dynamically drivers/media/i2c/max9286.c | 38 +++++++++++---------- drivers/media/platform/rcar-vin/rcar-csi2.c | 24 ++++++------- drivers/media/platform/rcar_drif.c | 30 +++++----------- include/media/v4l2-async.h | 5 +-- 4 files changed, 42 insertions(+), 55 deletions(-) -- Regards, Laurent Pinchart
