On Thu, Nov 14, 2019 at 03:07:16PM +0100, Jan Kara wrote: > I'd just note that e.g. drivers/scsi/sr.c or drivers/scsi/sd.c already call > revalidate_disk() on device open so it seems a bit stupid to call it again > just a bit later. But that's not really a new thing, this patch just makes > it universal. That whole area is a bit of a mess, including how the whole ->revalidate_disk callback works, and how we have a complicated events mechanism, but then also md as the only remaining user of the legacy ->media_changed method. I have a few idea on how to sort some of that out, but it's going to take a while.
