[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[lvm-devel] [RFC][PATCH 0/4] dm-raid1: fix deadlock at suspend after suspend was interrupted (v2)



Hi,

This is an update patch set to fix deadlock on suspending of mirror device.
Based on the Ueda-san's suggestion, I updated the patch set so that a target's
resume handler is used instead of introducing new handler (cancel_presuspend).


ISSUE
=====

Suspend procedure on a dm-mirror device could cause deadlock on recovery_count
semaphore.

When mirror_presuspend is called, recovery_count semaphore is acquired in
dm_rh_stop_recovery() to stop recovery routine, but when an signal is caught
in dm_wait_for_completion() or an error occurred in in dm_suspend(),
the suspend process is interrupted without releasing recovery_count semaphore
of a mirror device. This means that another suspend is executed, and then
the suspend process gets stuck at dm_rh_stop_recovery().

When suspend procedure is interrupted, the device should work properly since
the status of the device is not "suspended."


SOLUTION
========

Restore the target's state change by calling a target's specific resume handler
when its suspend procedure was interrupted after its presuspend handler completed.


PATCH SET
=========
    1/4: dm: restore presuspend status
    2/4: dm-log: update resume method for interruption of presuspend
    3/4: dm-crypt: update resume method for interruption of presuspend
    4/4: cmirror: update resume method for interruption of presuspend

    NOTE: The cmirror patch (4/4) hasn't been tested yet.


I appreciate your comments.

Thanks,
Taka


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]