bgpd: Add Support for rfc 8050 MRT add-path

 - Rfc 8050 adds support for BGP NLRI that carry path identifiers. this commit adds that support to FRR
 - Updated bgp_dump.h to include new sub-type values
 - Updated bgp_dump.c to check for add_path af_caps in the peer struct.
 - Updated bgp_dump.c to use the proper sub-type values upon detection of add-path af_caps
 - Updated bgp_dump.c to properly dump the path_id wen present.

Signed-off-by: David Teach <dteach@routeviews.org>

1 files changed, 39 insertions, 3 deletions

diff --git a/bgpd/bgp_dump.c b/bgpd/bgp_dump.c
index 975bba9314..944a5848ec 100644
--- a/bgpd/bgp_dump.c
+++ b/bgpd/bgp_dump.c
@@ -300,6 +300,13 @@ static void bgp_dump_routes_index_table(struct bgp *bgp)
 	fflush(bgp_dump_routes.fp);
 }
 
+static int bgp_addpath_encode_rx(struct peer *peer, afi_t afi, safi_t safi)
+{
+
+	return (CHECK_FLAG(peer->af_cap[afi][safi], PEER_CAP_ADDPATH_AF_RX_ADV)
+		&& CHECK_FLAG(peer->af_cap[afi][safi],
+			      PEER_CAP_ADDPATH_AF_TX_RCV));
+}
 
 static struct bgp_path_info *
 bgp_dump_route_node_record(int afi, struct bgp_dest *dest,
@@ -308,16 +315,27 @@ bgp_dump_route_node_record(int afi, struct bgp_dest *dest,
 	struct stream *obuf;
 	size_t sizep;
 	size_t endp;
+	int addpath_encoded;
 	const struct prefix *p = bgp_dest_get_prefix(dest);
 
 	obuf = bgp_dump_obuf;
 	stream_reset(obuf);
 
+	addpath_encoded = bgp_addpath_encode_rx(path->peer, afi, SAFI_UNICAST);
+
 	/* MRT header */
-	if (afi == AFI_IP)
+	if (afi == AFI_IP && addpath_encoded)
+		bgp_dump_header(obuf, MSG_TABLE_DUMP_V2,
+				TABLE_DUMP_V2_RIB_IPV4_UNICAST_ADDPATH,
+				BGP_DUMP_ROUTES);
+	else if (afi == AFI_IP)
 		bgp_dump_header(obuf, MSG_TABLE_DUMP_V2,
 				TABLE_DUMP_V2_RIB_IPV4_UNICAST,
 				BGP_DUMP_ROUTES);
+	else if (afi == AFI_IP6 && addpath_encoded)
+		bgp_dump_header(obuf, MSG_TABLE_DUMP_V2,
+				TABLE_DUMP_V2_RIB_IPV6_UNICAST_ADDPATH,
+				BGP_DUMP_ROUTES);
 	else if (afi == AFI_IP6)
 		bgp_dump_header(obuf, MSG_TABLE_DUMP_V2,
 				TABLE_DUMP_V2_RIB_IPV6_UNICAST,
@@ -361,6 +379,11 @@ bgp_dump_route_node_record(int afi, struct bgp_dest *dest,
 		/* Originated */
 		stream_putl(obuf, time(NULL) - (bgp_clock() - path->uptime));
 
+		/*Path Identifier*/
+		if (addpath_encoded) {
+			stream_putl(obuf, path->addpath_rx_id);
+		}
+
 		/* Dump attribute. */
 		/* Skip prefix & AFI/SAFI for MP_NLRI */
 		bgp_dump_routes_attr(obuf, path->attr, p);
@@ -528,19 +551,32 @@ static void bgp_dump_packet_func(struct bgp_dump *bgp_dump, struct peer *peer,
 				 struct stream *packet)
 {
 	struct stream *obuf;
-
+	int addpath_encoded = 0;
 	/* If dump file pointer is disabled return immediately. */
 	if (bgp_dump->fp == NULL)
 		return;
+	if (peer->su.sa.sa_family == AF_INET) {
+		addpath_encoded =
+			bgp_addpath_encode_rx(peer, AFI_IP, SAFI_UNICAST);
+	} else if (peer->su.sa.sa_family == AF_INET6) {
+		addpath_encoded =
+			bgp_addpath_encode_rx(peer, AFI_IP6, SAFI_UNICAST);
+	}
 
 	/* Make dump stream. */
 	obuf = bgp_dump_obuf;
 	stream_reset(obuf);
 
 	/* Dump header and common part. */
-	if (CHECK_FLAG(peer->cap, PEER_CAP_AS4_RCV)) {
+	if (CHECK_FLAG(peer->cap, PEER_CAP_AS4_RCV) && addpath_encoded) {
+		bgp_dump_header(obuf, MSG_PROTOCOL_BGP4MP,
+				BGP4MP_MESSAGE_AS4_ADDPATH, bgp_dump->type);
+	} else if (CHECK_FLAG(peer->cap, PEER_CAP_AS4_RCV)) {
 		bgp_dump_header(obuf, MSG_PROTOCOL_BGP4MP, BGP4MP_MESSAGE_AS4,
 				bgp_dump->type);
+	} else if (addpath_encoded) {
+		bgp_dump_header(obuf, MSG_PROTOCOL_BGP4MP,
+				BGP4MP_MESSAGE_ADDPATH, bgp_dump->type);
 	} else {
 		bgp_dump_header(obuf, MSG_PROTOCOL_BGP4MP, BGP4MP_MESSAGE,
 				bgp_dump->type);