The v4l2-async framework's subdev notifiers do not behave correctly if a
device in the middle of the pipeline is unbound and then rebound. The
v4l2-subdevices upstream from the device being rebound gets confused as
they receive no notification of the device unbound and can't cleanup
their state and when they are rebound to the new v4l2-subdev notifier
they try to reinitialize their internal state, this may include things
as trying to create links that already exists and in some cases crash
the system, for example the adv748x.
This should be solved in the v4l2-async framework, but as a stop-gap
measure suppress the bind and unbind sysfs nodes for the rcar-csi2
driver so it can't be used to crash the system.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
---
drivers/media/platform/rcar-vin/rcar-csi2.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
index 11848d0c4a55cb4c..37cc46b9201e0546 100644
--- a/drivers/media/platform/rcar-vin/rcar-csi2.c
+++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
@@ -1498,6 +1498,7 @@ static struct platform_driver rcar_csi2_pdrv = {
.probe = rcsi2_probe,
.driver = {
.name = "rcar-csi2",
+ .suppress_bind_attrs = true,
.of_match_table = rcar_csi2_of_table,
},
};
--
2.34.1
