drbd: fix WRITE_ACKED_BY_PEER_AND_SIS to not set RQ_NET_DONE
authorLars Ellenberg <lars.ellenberg@linbit.com>
Mon, 26 Mar 2012 14:51:11 +0000 (16:51 +0200)
committerPhilipp Reisner <philipp.reisner@linbit.com>
Thu, 8 Nov 2012 15:58:23 +0000 (16:58 +0100)
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
drivers/block/drbd/drbd_req.c

index 771f29c..938a57b 100644 (file)
@@ -587,28 +587,23 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,
                }
                break;
 
-       case WRITE_ACKED_BY_PEER_AND_SIS:
-               req->rq_state |= RQ_NET_SIS;
        case DISCARD_WRITE:
                /* for discarded conflicting writes of multiple primaries,
                 * there is no need to keep anything in the tl, potential
                 * node crashes are covered by the activity log. */
                req->rq_state |= RQ_NET_DONE;
                /* fall through */
+       case WRITE_ACKED_BY_PEER_AND_SIS:
        case WRITE_ACKED_BY_PEER:
+               if (what == WRITE_ACKED_BY_PEER_AND_SIS)
+                       req->rq_state |= RQ_NET_SIS;
                D_ASSERT(req->rq_state & RQ_EXP_WRITE_ACK);
                /* protocol C; successfully written on peer.
-                * Nothing to do here.
+                * Nothing more to do here.
                 * We want to keep the tl in place for all protocols, to cater
-                * for volatile write-back caches on lower level devices.
-                *
-                * A barrier request is expected to have forced all prior
-                * requests onto stable storage, so completion of a barrier
-                * request could set NET_DONE right here, and not wait for the
-                * P_BARRIER_ACK, but that is an unnecessary optimization. */
+                * for volatile write-back caches on lower level devices. */
 
                goto ack_common;
-               /* this makes it effectively the same as for: */
        case RECV_ACKED_BY_PEER:
                D_ASSERT(req->rq_state & RQ_EXP_RECEIVE_ACK);
                /* protocol B; pretends to be successfully written on peer.