The flaw in the Linux kernel’s vsock implementation causes a stale error flag (sk_err) to persist after a failed self‑connect that sets it to EPROTO. When the socket is subsequently set to listen, the kernel fails to clear this flag, so any future accept() call reads the stale EPROTO and rejects normally‑formed incoming connections. The rejected path also leaves the child socket allocated on certain vsock transports, producing a resource leak. The result is that a legitimate client cannot establish a connection and the kernel may exhaust resources tied to the leaked sockets.

The vulnerability is confined to the vsock driver within the Linux kernel. All Linux kernel releases that include vsock prior to the patch that removed the stale sk_err check are potentially affected. No specific version numbers are supplied, but any kernel using the vsock subsystem is at risk.

The EPSS score for this issue is reported as less than one percent, and it is not listed in the CISA KEV catalog, indicating that exploitation is currently expected to be rare. Nevertheless, the bug can be triggered by local or virtual machine to host communication paths, so systems that rely on vsock for internal or privileged connections may experience denial of service or intentional resource exhaustion. Because the attacker does not need special privileges to trigger a self‑connect, the risk profile is moderate, though the impact can compromise the stability of virtualized workloads.