Anthropic Oss-Scanner

In FRRouting 8.4 and later, bgpd frees the same memory twice when dynamic BGP neighbors disconnect, if the peer-group used by a "bgp listen range" also has a Site-of-Origin configured ("neighbor PG soo"). Each dynamic neighbor shares the peer-group's SoO value instead of getting its own copy, but frees it as its own when the neighbor is deleted, which happens as soon as its connection closes. The first disconnect leaves the peer-group pointing at freed memory, and the second disconnect frees it again, causing bgpd to abort and drop all BGP sessions on the router. Any host whose address falls inside the listen range can trigger this by opening and closing two TCP connections to the BGP port, without sending any BGP message and without knowing the AS number, and can repeat it each time bgpd restarts. If the peer-group has a TCP-MD5 password, only hosts that know the password can trigger it. Two ordinary flaps of legitimate dynamic neighbors have the same effect. Between the two frees, displaying the configuration ("show running-config") or establishing a dynamic neighbor also reads the freed memory, which can crash bgpd after a single disconnect. Only configurations that combine a listen range with a SoO on the same peer-group are affected.

The `bgpd` process fails to correctly close a BGP session upon the expiration of the Send Hold Timer (RFC 9687), which occurs when a neighbor stops accepting data for twice the hold time. While the connection is closed, `bgpd` continues to treat the session as active. This can be triggered by a neighbor that continues sending KEEPALIVEs but stops reading data, particularly when the router is sending large amounts of data such as a full routing table. If the neighbor supports Graceful Restart, `bgpd` crashes shortly after the session closes, disrupting all BGP sessions on the router. If Graceful Restart is not supported, the session remains stuck in the Established state, causing the neighbor's routes to remain installed and advertised to other peers, preventing reconnection until the session is manually cleared. This behavior can be triggered by an overloaded or stuck peer and does not require malicious intent.

In all prior released versions of FRRouting, the OSPFv3 (for IPv6) implementation does not check that OSPFv3 packets have a link-local source address and either a link-local destination or one of the protocol multicast destination groups. This may enable malicious actors to establish OSPFv3 adjacencies from unexpectedly, distant IPv6 hosts. Proper usage of OSPFv3 authentication trailers (RFC 7166, as is best practice) prevents full exploitation of this issue, though may allow a malicious actor to waste CPU on authentication checks. Note the linked PR #23935 is not sufficient to fix the issue as it only verifies the source address. It is the destination address that must be checked. Required configuration: The OSPFv3 for IPv6 routing protocol must be running (ospf6d process) and in use (enabled globally with "router ospf6" and on specific interfaces with "ipv6 ospf6 area".) Workaround: either: * configure OSPFv3 authentication trailers ("ipv6 ospf6 authentication") * or ensure through firewall configuration that only packets with link-local or multicast destination addresses are accepted on input for IPv6 protocol 89 (OSPF). Note that checking the source address is insufficient, as some routers may (unexpectedly) forward packets with global-scoped IPv6 destination addresses even if they have link-local source addresses.