tipc: add trace_events for tipc link

The commit adds the new trace_events for TIPC link object:

trace_tipc_link_timeout()
trace_tipc_link_fsm()
trace_tipc_link_reset()
trace_tipc_link_too_silent()
trace_tipc_link_retrans()
trace_tipc_link_bc_ack()
trace_tipc_link_conges()

And the traces for PROTOCOL messages at building and receiving:

trace_tipc_proto_build()
trace_tipc_proto_rcv()

Note:
a) The 'tipc_link_too_silent' event will only happen when the
'silent_intv_cnt' is about to reach the 'abort_limit' value (and the
event is enabled). The benefit for this kind of event is that we can
get an early indication about TIPC link loss issue due to timeout, then
can do some necessary actions for troubleshooting.

For example: To trigger the 'tipc_proto_rcv' when the 'too_silent'
event occurs:

echo 'enable_event:tipc:tipc_proto_rcv' > \
      events/tipc/tipc_link_too_silent/trigger

And disable it when TIPC link is reset:

echo 'disable_event:tipc:tipc_proto_rcv' > \
      events/tipc/tipc_link_reset/trigger

b) The 'tipc_link_retrans' or 'tipc_link_bc_ack' event is useful to
trace TIPC retransmission issues.

In addition, the commit adds the 'trace_tipc_list/link_dump()' at the
'retransmission failure' case. Then, if the issue occurs, the link
'transmq' along with the link data can be dumped for post-analysis.
These dump events should be enabled by default since it will only take
effect when the failure happens.

The same approach is also applied for the faulty case that the
validation of protocol message is failed.

Acked-by: Ying Xue <ying.xue@windriver.com>
Tested-by: Ying Xue <ying.xue@windriver.com>
Acked-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Tuong Lien <tuong.t.lien@dektech.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 7 insertions, 1 deletions

diff --git a/net/tipc/node.c b/net/tipc/node.c
index 4fd6e2887818..1e13ea98b96c 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -784,6 +784,7 @@ static void __tipc_node_link_down(struct tipc_node *n, int *bearer_id,
 		if (tipc_link_peer_is_down(l))
 			tipc_node_fsm_evt(n, PEER_LOST_CONTACT_EVT);
 		tipc_node_fsm_evt(n, SELF_LOST_CONTACT_EVT);
+		trace_tipc_link_reset(l, TIPC_DUMP_ALL, "link down!");
 		tipc_link_fsm_evt(l, LINK_RESET_EVT);
 		tipc_link_reset(l);
 		tipc_link_build_reset_msg(l, xmitq);
@@ -801,6 +802,7 @@ static void __tipc_node_link_down(struct tipc_node *n, int *bearer_id,
 	tipc_node_fsm_evt(n, NODE_SYNCH_END_EVT);
 	n->sync_point = tipc_link_rcv_nxt(tnl) + (U16_MAX / 2 - 1);
 	tipc_link_tnl_prepare(l, tnl, FAILOVER_MSG, xmitq);
+	trace_tipc_link_reset(l, TIPC_DUMP_ALL, "link down -> failover!");
 	tipc_link_reset(l);
 	tipc_link_fsm_evt(l, LINK_RESET_EVT);
 	tipc_link_fsm_evt(l, LINK_FAILOVER_BEGIN_EVT);
@@ -1022,6 +1024,7 @@ void tipc_node_check_dest(struct net *net, u32 addr,
 			*respond = false;
 			goto exit;
 		}
+		trace_tipc_link_reset(l, TIPC_DUMP_ALL, "link created!");
 		tipc_link_reset(l);
 		tipc_link_fsm_evt(l, LINK_RESET_EVT);
 		if (n->state == NODE_FAILINGOVER)
@@ -1599,8 +1602,11 @@ static bool tipc_node_check_state(struct tipc_node *n, struct sk_buff *skb,
 		}
 	}
 
-	if (!tipc_link_validate_msg(l, hdr))
+	if (!tipc_link_validate_msg(l, hdr)) {
+		trace_tipc_skb_dump(skb, false, "PROTO invalid (2)!");
+		trace_tipc_link_dump(l, TIPC_DUMP_NONE, "PROTO invalid (2)!");
 		return false;
+	}
 
 	/* Check and update node accesibility if applicable */
 	if (state == SELF_UP_PEER_COMING) {