GPU VulnDB

Database/Control plane, storage & DevOps

Linux kernel nfsd: stale opcnt after compound release leaks adjacent slab memory through the RPC status netlink dump

CVSS 7.1CVE-2026-89691Control plane, storage & DevOpscurated

Impact

nfsd4_release_compoundargs() resets args->ops to the inline 8-element iops[] array but leaves args->opcnt at its original value, which can be up to 200 for NFSv4.1+ compounds. If rq_status_counter is left odd, the RPC status dumpit handler reads min(opcnt, 16) entries, so indices 8 through 15 read adjacent slab memory past the end of the structure and hand it to userspace over netlink. That is a kernel heap disclosure to whoever can query the nfsd status interface on the storage server - useful for defeating KASLR or lifting neighbouring kernel data. Local only: it requires a process on the NFS server itself, not a remote client, which limits it to operators and anything running on the storage head.

Who can reach it

A local user or process on the NFS server able to issue the nfsd RPC-status netlink dump. Not reachable from remote NFS clients.

What to do

Update the server kernel to a build that zeroes opcnt on compound release (stable commits 58bcdfb2b2e4, ae4c38555e81, c1a4f7b1848f, e879148867bd) and reboot the node. In the meantime, keep unprivileged workloads off the NFS server itself - a dedicated storage head with no tenant processes is not meaningfully exposed.

References

Related entries

All Control plane, storage & DevOps 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.