Marc--

Thanks for the report.

On 11/25/2017 03:11 PM, Marc MERLIN wrote:

Howdy,

This has been an issue for a while, but it's not fatal.
Nonetheless, I guess I should report it.

I trigger it like this:
echo 1 > /sys/block/md5/bcache/stop
sleep 5
mdadm --stop /dev/md5
(this was backed by bcache64)

bcache: bcache_device_free() bcache64 stopped
ida_remove called for id=64 which is not allocated.

I believe this is fixed by 1dbe32ad0a82f39c6 which is in Linus's tree
for 4.15.  Unfortunately it's not a candidate for -stable because it
changes device numbering.

Can you easily try a 4.15 RC to see if this is fixed for you?

Thanks,

Mike