Executive brief
A flaw in the Linux kernel's networking component could allow a local user to cause a system deadlock. This occurs when specific BPF programs interact with UNIX domain sockets in a way that triggers a recursive locking error. An exploit would result in a denial-of-service, potentially freezing the affected system and disrupting operations.
Technical details
A deadlock vulnerability exists in the Linux kernel's af_unix BPF iterator. The root cause is in `bpf_iter_unix_seq_show()`, which may deadlock when `lock_sock_fast()` takes the fast path while a BPF program simultaneously attempts to update a sockmap via `sock_map_update_elem()`. This leads to a recursive lock acquisition of the `slock-AF_UNIX` spinlock. An attacker with the ability to run BPF programs (typically requiring local access) can trigger this condition to cause a kernel hang. The issue has been resolved by replacing `lock_sock_fast()` with `lock_sock()` and updating the socket state checks.
Affected products
- Linux Linux Kernel 2c860a43dd77 to 4d328dd69538
Timeline
- 2026-04-14: disclosed: Initial patch submission by Michal Luczaj
- 2026-04-15: patched: Mainline kernel patch committed
- 2026-06-24: advisory: CVE-2026-53035 published
References
- https://git.kernel.org/stable/c/3cef33b9813b78f227942572fb317afcd5c9ac94
- https://git.kernel.org/stable/c/4d328dd695383224aa750ddee6b4ad40c0f8d205
- https://git.kernel.org/stable/c/527057ebe8076dfbcaef51195ff1b7508646be2c
- https://git.kernel.org/stable/c/66d9fab4565eafe1afe7ba0581f79b76073b60fa
- https://git.kernel.org/stable/c/87828b380956d4986f59f2c086e0b09b3e6cdaae
- https://git.kernel.org/stable/c/bd3592129f24243713673a07225cf1f15a9bb835