[Bug 212337] scsi_debug: race at module load and module unload

[bugzilla-daemon@bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=212337

--- Comment #19 from Luis Chamberlain (mcgrof@kernel.org) ---
(In reply to d gilbert from comment #17)
> I'm working on the rm_all_hosts parameter angle which lets me get in "under"
> modprobe, as well as after scsi_debug_init() finishes ... lots of races :-)

OK so this would be a new generic sysfs knob, ie, independent of scsi devices
it creates? So created and attached synchronously at init.

> If I can solve all of those, the reported problem may disappear.

If the solution includes a state machine to prevent *future* additions, *or* it
uses try_module_get() before allowing any new possible generic knobs (not sure
if you are going to add some) I agree that this should work.

FWIW, just one minor consideration, just recall that distributions are now
using scsi_debug to do tons of testing both on fstests and blktests... So...
they'd benefit from fixing these races as a backport with the least amount of
changes as possible, and so if a fix is possible with the least amount of lines
possible they can backport this fix so that tests don't run into so many false
positives with tests.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.