GPU VulnDB

Database/Kernel, userspace & hypervisor

Linux kernel nfsd: replayed SEQUENCE requests race on a session slot and crash the NFS server

UnscoredCVE-2026-99366Kernel, userspace & hypervisorcurated

Impact

A regression means a reply served from a slot's cache acts as if it owns the slot. It stores its reply and clears the in-use flag, so two replays can run on one slot at once. One frees and rebuilds the slot credentials while the other reads them. The result is a NULL dereference and kernel panic, seen on a production 6.12 server, or a double free. From v6.14 on, when the slot table shrinks it can also become a use-after-free write. A panicked NFS server stalls every client mount on the cluster, along with the training jobs reading from it.

Who can reach it

An NFSv4.1+ client with a session, sending retransmissions on a slot. Ordinary retries under load can trigger it, as the production crash shows.

What to do

Install a kernel with the stable fixes and reboot the NFS server, which interrupts client mounts. The record names no workaround. Kernels without commit cc028a10a48c do not have the regression.

References

Related entries

All Kernel, userspace & hypervisor entries

This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.