Fixed
Created: Dec 7, 2025
Updated: Jun 1, 2026
Resolved Date: May 25, 2026
Found In Version: 10.22.33.1
Fix Version: 10.22.33.24
Severity: Standard
Applicable for: Wind River Linux LTS 22
Component/s: Kernel
In the Linux kernel, the following vulnerability has been resolved:[EOL][EOL]vsock: Ignore signal/timeout on connect() if already established[EOL][EOL]During connect(), acting on a signal/timeout by disconnecting an already[EOL]established socket leads to several issues:[EOL][EOL]1. connect() invoking vsock_transport_cancel_pkt() ->[EOL] virtio_transport_purge_skbs() may race with sendmsg() invoking[EOL] virtio_transport_get_credit(). This results in a permanently elevated[EOL] `vvs->bytes_unsent`. Which, in turn, confuses the SOCK_LINGER handling.[EOL][EOL]2. connect() resetting a connected socket's state may race with socket[EOL] being placed in a sockmap. A disconnected socket remaining in a sockmap[EOL] breaks sockmap's assumptions. And gives rise to WARNs.[EOL][EOL]3. connect() transitioning SS_CONNECTED -> SS_UNCONNECTED allows for a[EOL] transport change/drop after TCP_ESTABLISHED. Which poses a problem for[EOL] any simultaneous sendmsg() or connect() and may result in a[EOL] use-after-free/null-ptr-deref.[EOL][EOL]Do not disconnect socket on signal/timeout. Keep the logic for unconnected[EOL]sockets: they don't linger, can't be placed in a sockmap, are rejected by[EOL]sendmsg().[EOL][EOL][1]: https://lore.kernel.org/netdev/e07fd95c-9a38-4eea-9638-133e38c2ec9b@rbox.co/[EOL][2]: https://lore.kernel.org/netdev/20250317-vsock-trans-signal-race-v4-0-fc8837f3f1d4@rbox.co/[EOL][3]: https://lore.kernel.org/netdev/60f1b7db-3099-4f6a-875e-af9f6ef194f6@rbox.co/
