FreeBSD EUVD-2026-101867

Requires an attacker-controlled process running inside a FreeBSD jail that has received a directory file descriptor from a DIFFERENT jail via SCM_RIGHTS over a Unix domain socket; the process then re-sends that descriptor to itself, which fails to preserve FD_RESOLVE_BENEATH and clears the jail filesystem-root restriction. Limiting factors: the attacker must first obtain code execution within a jail, and a cross-jail FD-passing channel must exist - single-jail or default deployments without inter-jail FD sharing are not exploitable, and the flaw is not reachable remotely despite the input AV:N rating. Fixed FreeBSD patch levels are not vulnerable.

The input CVSS of 9.8 (AV:N/AC:L/PR:N/UI:N) overstates real-world risk and directly conflicts with the description, which requires an attacker to already have code execution inside a FreeBSD jail AND to have received a directory file descriptor from another jail - this is a local, precondition-heavy jail escape, not a remotely reachable unauthenticated flaw. The CVSS network vector should be treated as questionable; verify with the FreeBSD advisory, as the vulnerability is realistically AV:L with specific attack requirements. Corroborating the lower practical risk, EPSS is only 0.19% (8th percentile) and CISA SSVC lists Exploitation as 'none', with no KEV listing and no public exploit identified at time of analysis. However, SSVC rates Automatable as 'yes' and Technical Impact as 'total', consistent with a complete jail filesystem-root escape (CWE-273, improper check for dropped privileges). Net assessment: a genuinely serious containment-boundary defect for multi-jail FreeBSD hosts that should be patched promptly, but not an emergency mass-exploitation threat - the CVSS 9.8 label alone overstates urgency given the local, cross-jail FD prerequisite and absence of exploitation signals.

An attacker who already controls a process inside a FreeBSD jail - for example via a compromised jailed service - and who can receive a directory file descriptor passed from another jail over a Unix domain socket, re-sends that descriptor to itself to strip the FD_RESOLVE_BENEATH restriction. With the restriction cleared, the process traverses outside the jail's filesystem root to read and modify host files, escaping containment. No public exploit code is identified at time of analysis, and exploitation depends on an existing cross-jail FD-sharing setup.