Technical impact
The vulnerability originates in the Linux kernel’s handling of mount namespace locks during copy_tree() operations. When a copy is performed without including CL_COPY_MNT_NS_FILE, the function may omit a mount of a mount‑namespace file, allowing a new mount to be placed under a previously locked mount. A process that is root in its own user namespace can exploit this to move the missing mount beneath the lock, take over the lock, and unmount the covering mount without a race. The attacker can then access or modify files that were previously protected, effectively escalating privileges locally. This flaw is related to improper handling of lock states, enabling unauthorized removal of a mounting guard.
Affected Systems
All Linux kernel versions that lack the fixes introduced by commits c62a4766937e, df7342b24018, and f78ddcc87224 are affected. Common distribution kernels released before the merge of these commits remain vulnerable. The issue specifically impacts operations involving mount namespace manipulation such as bind mounts, MOVE_MOUNT_BENEATH, and unshare(CLONE_NEWNS) or clone(CLONE_NEWNS). Kernels that have incorporated these commits are no longer vulnerable.
Risk and Exploitability
The CVSS score of 8.8 classifies the flaw as high severity. The EPSS score indicates a very low probability of exploitation at present (< 1%), and it is not listed in the CISA KEV catalog. The exploit requires local root within a user namespace and the ability to perform mount operations; it does not provide remote code execution or system‑wide compromise. The most likely attack surface is an environment that allows untrusted code to create or manipulate user namespaces, such as container runtimes or applications that spawn namespaces (inferred from the described requirements). Organizations that run such workloads should consider this risk.