mctpd: update setup peer logic for bridged endpoints

For bridged endpoints, route setup is already done once pool space
is allocated successfully. Those are destined to be routed from
the bridge themselves without need of neighbours.

Add appropiate check to avoid extra route creation for setting up
bridged endpoints.

Signed-off-by: Faizan Ali <[email protected]>

Author: faizana-nvidia

src/mctpd.c

@@ -2856,13 +2856,18 @@ static int peer_route_update(struct peer *peer, uint16_t type)
 static int setup_added_peer(struct peer *peer)
 {
 	int rc;
+	struct net *n;
+	n = lookup_net(peer->ctx, peer->net);
 
 	// Set minimum MTU by default for compatibility. Clients can increase
 	// this with .SetMTU as needed
 	peer->mtu = mctp_nl_min_mtu_byindex(peer->ctx->nl, peer->phys.ifindex);
 
-	// add route before querying
-	add_peer_route(peer);
+	// add route before querying for non-bridged endpoints.
+	// bridged endpoints will be routed via gateway route from bridge itself
+	// without need of neighbour
+	if (!is_eid_in_bridge_pool(n, peer->ctx, peer->eid))
+		add_peer_route(peer);
 
 	rc = query_peer_properties(peer);
 	if (rc < 0)