DEBIAN-CVE-2026-129456: High severity debian/frr vulnerability
In FRRouting, bgpd frees the sub-TLV chain of a Tunnel Encapsulation attribute (type 23) while routes stored by bgpd still point to it, if the attribute is received from a neighbor that has "soft-reconfiguration inbound" configured. A cache added to speed up attribute interning while processing the prefixes of an UPDATE checks whether an interned attribute set may be reused by comparing the parsed attributes with themselves, so the check always succeeds. Interning the parsed attributes for the Adj-RIB-In removes the Tunnel Encapsulation sub-TLVs from them, and the copy that bgpd then interns for the route reuses the cached attribute set, which still holds the sub-TLVs, without taking a reference on them. When the route is later replaced, withdrawn or cleared, for example when the session goes down, the sub-TLV chain is freed while still in use, and bgpd then reads it, decrements a reference count inside it and frees it a second time. In the reporter's build without AddressSanitizer this ended in an assertion failure in bgpd's memory accounting, aborting bgpd and dropping all BGP sessions on the router, and the neighbor can repeat it each time bgpd restarts. Because freed heap memory is written to and freed again, consequences beyond a crash cannot be ruled out. An established BGP session is required, and the attribute does not need to be malformed. bgpd processes attribute 23 from every neighbor, including eBGP neighbors, unless it is explicitly discarded, although RFC 9012 requires it to be filtered on eBGP sessions by default. Neighbors without "soft-reconfiguration inbound" are not affected.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
debian/frrto a version that resolves this vulnerability.Fixed in 10.7.0-1 - Configuration
Disable “soft-reconfiguration inbound” for affected neighbors.
FRRouting bgpd soft-reconfiguration inbound = disabled
Event History
Frequently Asked Questions
Which peers can trigger the issue?
A peer needs an established BGP session and the ability to send BGP UPDATE messages. bgpd processes the attribute from every neighbor, including eBGP peers, unless the attribute is explicitly discarded.
Are eBGP peers affected by default?
Yes. The described behavior processes Tunnel Encapsulation attribute type 23 from eBGP neighbors unless it is explicitly filtered or discarded, despite RFC 9012 requiring filtering on eBGP.
What can be done if patching is not immediately possible?
Explicitly discard or filter Tunnel Encapsulation attribute type 23 from eBGP neighbors. This prevents those neighbors from supplying the malformed attribute through the affected path.
How can an attempted exploit be identified?
bgpd logs a mismatch between the outer TLV length and the attribute length, but continues processing the attribute. A successful trigger may later cause an assertion failure in bgpd memory accounting, abort bgpd, and drop all BGP sessions.