staging: lustre: lnet: make connection more stable with packet loss
authorAlexander Boyko <alexander.boyko@seagate.com>
Tue, 16 Aug 2016 20:19:31 +0000 (16:19 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 21 Aug 2016 13:57:39 +0000 (15:57 +0200)
IB network may lose last connection handshake packet.
This problem isn't Lustre specific and described at
https://oss.oracle.com/pipermail/rds-devel/2007-December/000271.html
for example. Solution is to make conection established if any packet
is received for it.

Signed-off-by: Alexander Boyko <alexander.boyko@seagate.com>
Signed-off-by: Alexey Lyashkov <alexey.lyashkov@seagate.com>
Seagate-bug-id: MRP-2883
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-8303
Reviewed-on: http://review.whamcloud.com/20874
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: Alexander Zarochentsev <alexander.zarochentsev@seagate.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd_cb.c

index 19c90fcceb0d2566e81be52fb75991a9eeccb13e..6cd78ea279a7d7dd06a2a55cdbb6d15d40abaa45 100644 (file)
@@ -3419,6 +3419,12 @@ kiblnd_qp_event(struct ib_event *event, void *arg)
        case IB_EVENT_COMM_EST:
                CDEBUG(D_NET, "%s established\n",
                       libcfs_nid2str(conn->ibc_peer->ibp_nid));
+               /*
+                * We received a packet but connection isn't established
+                * probably handshake packet was lost, so free to
+                * force make connection established
+                */
+               rdma_notify(conn->ibc_cmid, IB_EVENT_COMM_EST);
                return;
 
        default: