OpenSourceRouting / quagga-snap / 7ef4221c3f8

Commits

Donald Sharp authored and Paul Jakma committed 7ef4221c3f8 30 Mar 2015

bgpd: Fix race in clearing completion

When a peer that is Established goes down, it is moved into the Clearing
state to facilitate clearing of the routes received from the peer - remove
from the RIB, reselect best path, update/delete from Zebra and to other
peers etc. At the end of this, a Clearing_Completed event is generated to
the FSM which will allow the peer to move out of Clearing to Idle.

The issue in the code is that there is a possibility of multiple Clearing
Completed events being generated for a peer, one per AFI/SAFI. Upon the
first such event, the peer would move to Idle. If other events happened
(e.g., new connection got established) before the last Clearing_Completed
event is received, bad things can happen.

Fix to ensure only one Clearing_Completed event is generated.

Signed-off-by: Vivek Venkataraman <vivek@cumulusnetworks.com>

bgpd/bgp_fsm.c

Modified

-399
+   read/write and timer thread. */
-400
+void
-401
+bgp_fsm_change_status (struct peer *peer, int status)
-402
+{
-403
+  bgp_dump_state (peer, peer->status, status);
-404
-405
+  /* Transition into Clearing or Deleted must /always/ clear all routes..
-406
+   * (and must do so before actually changing into Deleted..
-407
+   */
-408
+  if (status >= Clearing)
--
+    bgp_clear_route_all (peer);
-+
+    {
-+
+      bgp_clear_route_all (peer);
-+
-+
+      /* If no route was queued for the clear-node processing, generate the
-+
+       * completion event here. This is needed because if there are no routes
-+
+       * to trigger the background clear-node thread, the event won't get
-+
+       * generated and the peer would be stuck in Clearing. Note that this
-+
+       * event is for the peer and helps the peer transition out of Clearing
-+
+       * state; it should not be generated per (AFI,SAFI). The event is
-+
+       * directly posted here without calling clear_node_complete() as we
-+
+       * shouldn't do an extra unlock. This event will get processed after
-+
+       * the state change that happens below, so peer will be in Clearing
-+
+       * (or Deleted).
-+
+       */
-+
+      if (!peer->clear_node_queue->thread)
-+
+        BGP_EVENT_ADD (peer, Clearing_Completed);
-+
+    }
-426
-427
+  /* Preserve old status and change into new status. */
-428
+  peer->ostatus = peer->status;
-429
+  peer->status = status;
-430
-431
+  if (BGP_DEBUG (normal, NORMAL))
-432
+    zlog_debug ("%s went from %s to %s",
-433
+		peer->host,
-434
+		LOOKUP (bgp_status_msg, peer->ostatus),
-435
+		LOOKUP (bgp_status_msg, peer->status));

quagga-snap

Commits

Jira Issues

Add shortcut