drbd: properly call drbd_rs_cancel_all() in drbd_disconnected()
authorLars Ellenberg <lars.ellenberg@linbit.com>
Fri, 17 Aug 2012 13:09:13 +0000 (15:09 +0200)
committerPhilipp Reisner <philipp.reisner@linbit.com>
Fri, 9 Nov 2012 13:08:19 +0000 (14:08 +0100)
commit08332d73250eec349b055843a503d45a9b5c13b6
treeec23064ccd2e5f0e21ee883793d0682c5e9a784a
parent155522df5b8ac24ee66a903e51d5b3023b2a76f9
drbd: properly call drbd_rs_cancel_all() in drbd_disconnected()

drbd_disconnected() is supposed to clear the resync lru cache,
by calling drbd_rs_cancel_all().

We must do so before we call drbd_flush_workqueue(), as at least the
callback w_restart_disk_io() may wait for resync progres, and would
otherwise deadlock.

drbd_finish_peer_reqs() may again populate that cache, which will
then potentially be stale after the next resync handshake and bitmap
exchange, we have to do it again after that.

A stale resync lru cache causes no harm but ugly messages like this:
 BAD! sector=196608s enr=6 rs_left=-256 rs_failed=0 count=256 cstate=SyncTarget

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
drivers/block/drbd/drbd_receiver.c