[ENBD] Problem with enbd 2.2.27

[Peter T. Breuer]

"A month of sundays ago Richard Sharpe wrote:"
> That enbd module is really badly behaved. The system refuses to reboot
> using the reboot command if things go wrong ... And the driver refuses to

You have oopsed. The kernel can do anything now.

> unload, even though nothing is using it.

If nothing is using it (lsmod) then it can unload when outstanding requests
are served. If you don't want to serve them, error them out.
Send it a USR1 signal (i.e. echo -n 0 > /proc/nbdinfo if your clients are
dead).

If the usage count is incorrect, use a USR2 signal (i.e. a 1 if echoing to the
proc file).

> It must have some filesystem lock because it refuses to unmount the file
> systems.

No. It knows nothing about file systems. You are seeing the kernel
trying to write some bytes (such as last unmount time) to a device
that doesn't exist. Error out the requests. Show the outoput of
proc/nbdinfo.


Peter