Impact
During the unmount process, the kernel incorrectly allows a busy submount to be detached. The first unmount changes the mount tree, so a subsequent propagation check may misinterpret the busy mount as free and remove it. This behavior can surface as EBUSY errors and may result in the removal of a mount that still has active references, potentially corrupting the filesystem state or leading to kernel instability.
Affected Systems
All Linux kernel releases that use the unmount logic described prior to the patch commit are affected. The vendor list indicates generic Linux kernels; no specific version ranges are provided, so any kernel built with the pre‑patch code is potentially vulnerable. The issue involves automounted submounts of NFS, AFS, CIFS, tracefs, and any bind mounts inheriting the MNT_SHRINKABLE flag.
Risk and Exploitability
Location: requires local privileged access to perform unmount operations, typically CAP_SYS_ADMIN. The CVSS score of 4.1 denotes moderate severity. EPSS <1% indicates low likelihood of exploitation in the wild. The vulnerability is not listed in the CISA KEV. Attackers would need to safely trigger the race condition through coordinated unmount requests on busy MNT_SHRINKABLE mounts to cause mount removal errors or instability.
OpenCVE Enrichment