GPU VulnDB

Database/Kernel, userspace & hypervisor

Linux kernel nfsd: use-after-free reading NFSv4 compound ops from the rpc_status netlink dump

CVSS 7.8CVE-2026-89690Kernel, userspace & hypervisorcurated

Impact

The rpc_status netlink dumpit walks in-flight svc_rqst structures under RCU and reads NFSv4 opnums out of args->ops, but that buffer is a plain vmalloc allocation freed synchronously at the end of every compound. RCU protects the svc_rqst, not the ops buffer, so a compound completing concurrently with a status dump frees vmalloc memory that the dump is still reading. On a busy NFS server backing a GPU cluster, compounds complete constantly, so the window is open whenever anything queries rpc_status - including routine monitoring. The fix switches to kvfree_rcu_mightsleep(), at the cost of holding completed ops buffers for one grace period.

Who can reach it

Local. A user able to issue the nfsd rpc_status netlink dump (root or a process with the matching capability in the namespace) racing against normal NFSv4 client traffic on the same server.

What to do

Update to a stable kernel containing commit 19413ccc4507 or its backport and reboot the NFS server nodes. Until then, avoid running nfsd rpc_status queries (including monitoring agents that poll it) on affected kernels.

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.