Fixes another bug in packet retransmissions.

When slave was in full state, it didn't retransmit
lost packets. That leads to OSPF connection locked
in states loading-full.

proto/ospf/dbdes.c

@@ -344,6 +344,11 @@ ospf_dbdes_receive(struct ospf_dbdes_packet *ps,
       /* Only duplicate are accepted */
     {
       OSPF_TRACE(D_PACKETS, "Received duplicate dbdes from %I.", n->ip);
+      if (n->myimms.bit.ms == 0)
+      {
+	/* Slave should retransmit dbdes packet */
+	ospf_dbdes_send(n);
+      }
       return;
     }
     else