bgpd: Allow bgp to know when to actually add/delete agg route

The aggregated route was being sent in updates to peers every
time a route changed that we were aggregating.  Modify
the code such that we only send aggregated route updates
if we actually have something different to tell the peer.

Signed-off-by: Donald Sharp <sharpd@cumulusnetworks.com>

diff --git a/bgpd/bgp_route.c b/bgpd/bgp_route.c
index ed71f4197f79f50c6366aca85a7cb5b9145e0611..567805d91929dfc31a30e993bd37719df1d23f7b 100644 (file)
--- a/bgpd/bgp_route.c
+++ b/bgpd/bgp_route.c
@@ -5433,6 +5433,26 @@ static void bgp_aggregate_free(struct bgp_aggregate *aggregate)
        XFREE(MTYPE_BGP_AGGREGATE, aggregate);
 }
 
+static int bgp_aggregate_info_same(struct bgp_info *ri, struct aspath *aspath,
+                                  struct community *comm)
+{
+       static struct aspath *ae = NULL;
+
+       if (!ae)
+               ae = aspath_empty();
+
+       if (!ri)
+               return 0;
+
+       if (!aspath_cmp(ri->attr->aspath, (aspath) ? aspath : ae))
+               return 0;
+
+       if (!community_cmp(ri->attr->community, comm))
+               return 0;
+
+       return 1;
+}
+
 static void bgp_aggregate_install(struct bgp *bgp, afi_t afi, safi_t safi,
                                  struct prefix *p, uint8_t origin,
                                  struct aspath *aspath,
@@ -5447,7 +5467,34 @@ static void bgp_aggregate_install(struct bgp *bgp, afi_t afi, safi_t safi,
        table = bgp->rib[afi][safi];
 
        rn = bgp_node_get(table, p);
+
+       for (ri = rn->info; ri; ri = ri->next)
+               if (ri->peer == bgp->peer_self && ri->type == ZEBRA_ROUTE_BGP
+                   && ri->sub_type == BGP_ROUTE_AGGREGATE)
+                       break;
+
        if (aggregate->count > 0) {
+               /*
+                * If the aggregate information has not changed
+                * no need to re-install it again.
+                */
+               if (bgp_aggregate_info_same(rn->info, aspath, community)) {
+                       bgp_unlock_node(rn);
+
+                       if (aspath)
+                               aspath_free(aspath);
+                       if (community)
+                               community_free(community);
+
+                       return;
+               }
+
+               /*
+                * Mark the old as unusable
+                */
+               if (ri)
+                       bgp_info_delete(rn, ri);
+
                new = info_make(
                        ZEBRA_ROUTE_BGP, BGP_ROUTE_AGGREGATE, 0, bgp->peer_self,
                        bgp_attr_aggregate_intern(bgp, origin, aspath,
@@ -5682,8 +5729,6 @@ void bgp_aggregate_increment(struct bgp *bgp, struct prefix *p,
                if ((aggregate = rn->info) != NULL
                    && rn->p.prefixlen < p->prefixlen) {
                        bgp_aggregate_delete(bgp, &rn->p, afi, safi, aggregate);
-                       bgp_aggregate_install(bgp, afi, safi, &rn->p, 0, NULL,
-                                             NULL, 0, aggregate);
                        bgp_aggregate_route(bgp, &rn->p, ri, afi, safi, NULL,
                                            aggregate);
                }
@@ -5714,8 +5759,6 @@ void bgp_aggregate_decrement(struct bgp *bgp, struct prefix *p,
                if ((aggregate = rn->info) != NULL
                    && rn->p.prefixlen < p->prefixlen) {
                        bgp_aggregate_delete(bgp, &rn->p, afi, safi, aggregate);
-                       bgp_aggregate_install(bgp, afi, safi, &rn->p, 0, NULL,
-                                             NULL, 0, aggregate);
                        bgp_aggregate_route(bgp, &rn->p, NULL, afi, safi, del,
                                            aggregate);
                }