drbd: Finish requests that completed while IO was frozen
authorPhilipp Reisner <philipp.reisner@linbit.com>
Tue, 31 Jul 2012 09:22:58 +0000 (11:22 +0200)
committerPhilipp Reisner <philipp.reisner@linbit.com>
Thu, 16 Aug 2012 15:14:45 +0000 (17:14 +0200)
commit509fc019e534bdf5f3969d78c53184db4cf7ff48
tree2db80b5650709fd1e0a538506a3e94007a273b3c
parent227f052f4711caf432b9a7dbcfe1a2857d3c0def
drbd: Finish requests that completed while IO was frozen

Requests of an acked epoch are stored on the barrier_acked_requests list. In
case the private bio of such a request completes while IO on the drbd device
is suspended [req_mod(completed_ok)] then the request stays there.

When thawing IO because the fence_peer handler returned, then we use
tl_clear() to apply the connection_lost_while_pending event to all requests
on the transfer-log and the barrier_acked_requests list.

Up to now the connection_lost_while_pending event was not applied
on requests on the barrier_acked_requests list. Fixed that.

I.e. now the connection_lost_while_pending and resend events are
applied to requests on the barrier_acked_requests list. For that
it is necessary that the resend event finishes (local only)
READS correctly.

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