bgpd: set ipv4-mapped ipv6 for ipv4 with ipv6 nexthop

The code was expected that no IPv6 global address was present but the previous commit was replacing nexthop.v6global by the link-local address instead of un-setting it in case of removal of the IPv6 global. Set also ipv4-mapped ipv6 address as nexthop when a link-local is found and it is an ipv4 prefix over ipv6 nexthop. Signed-off-by: Louis Scalbert <[email protected]>
opensourcerouting · Apr 23, 2024 · fc5a738 · fc5a738
1 parent 04c220b
commit fc5a738
Showing 1 changed file with 7 additions and 5 deletions.
diff --git a/bgpd/bgp_updgrp_packet.c b/bgpd/bgp_updgrp_packet.c
@@ -523,11 +523,13 @@ struct stream *bpacket_reformat_for_peer(struct bpacket *pkt,
 			gnh_modified = 1;
 		}
 
-		if (IN6_IS_ADDR_UNSPECIFIED(mod_v6nhg)) {
-			if (peer->nexthop.v4.s_addr != INADDR_ANY) {
-				ipv4_to_ipv4_mapped_ipv6(mod_v6nhg,
-							 peer->nexthop.v4);
-			}
+		if (peer->nexthop.v4.s_addr != INADDR_ANY &&
+		    (IN6_IS_ADDR_UNSPECIFIED(mod_v6nhg) ||
+		     (IN6_IS_ADDR_LINKLOCAL(mod_v6nhg) &&
+		      peer->connection->su.sa.sa_family == AF_INET6 &&
+		      paf->afi == AFI_IP))) {
+			ipv4_to_ipv4_mapped_ipv6(mod_v6nhg, peer->nexthop.v4);
+			gnh_modified = 1;
 		}
 
 		if (IS_MAPPED_IPV6(&peer->nexthop.v6_global)) {