CVE-2026-89688 PUBLISHED

nfsd: drop the stateid, not the stateowner, on seqid_op replay retry

Assigner: Linux
Reserved: 11.09.2026 Published: 11.09.2026 Updated: 13.09.2026

In the Linux kernel, the following vulnerability has been resolved:

nfsd: drop the stateid, not the stateowner, on seqid_op replay retry

In nfs4_preprocess_seqid_op() the stateid is obtained from nfsd4_lookup_stateid(), which holds a reference on the nfs4_stid (sc_count) but takes no reference on the stateowner. openlockstateid() merely casts that stid and likewise takes no reference.

When nfsd4_cstate_assign_replay() returns -EAGAIN (the replay owner is being torn down, RP_UNHASHED) it has not taken a stateowner reference on that path. The error handling nevertheless called nfs4_put_stateowner(stp->st_stateowner), dropping an so_count reference the function never acquired -- risking a stateowner refcount underflow and use-after-free -- while leaking the sc_count reference held on the stid. The leaked stid reference can also stall a concurrent nfsd4_close_open_stateid() waiting for sc_count to drop.

Drop the reference actually held -- the stid -- before retrying. The stateowner stays alive through the reference held by the stid. This mirrors the open path in nfsd4_process_open1(), where the put balances a reference that path explicitly holds on the stateowner.

Metrics

CVSS Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
CVSS Score: 9.8

AV:N - nfsd processes NFSv4 COMPOUND ops (CLOSE, OPEN_DOWNGRADE, LOCK, OPEN_CONFIRM) in the in-kernel NFS server on TCP/UDP 2049; those handlers call nfs4_preprocess_seqid_op(), so a remote peer reaches the bug with no local access. AC:L - A remote NFSv4.0 client drives both sides by sending concurrent CLOSE and another seqid-mutating op on the same openowner; CLOSE's move_to_close_lru() sets RP_UNHASHED and waits on sc_count, so the window is attacker-controlled and retryable. PR:N - Typical nfsd exports use AUTH_SYS/AUTH_NULL, so RPC uids are self-asserted and not cryptographically verified; any host allowed by the export list can SETCLIENTID, OPEN, then CLOSE/OPEN_DOWNGRADE with no local Linux account. UI:N - The attacker issues the NFSv4 COMPOUND requests itself; no victim on the server must mount a filesystem, open a file, or take any other action. S:U - The underflow frees an nfsd nfs4_stateowner in the NFS server kernel; impact stays in that kernel and does not cross a VM, IOMMU, or sandbox boundary. C:H - nfs4_put_stateowner() drops an so_count the seqid path never took, freeing the stateowner while the stid and retry loop still use it; that UAF enables slab reclaim and kernel memory disclosure. I:H - The same UAF exposes so_ops->so_unhash/so_free function pointers and list writes under cl_lock, yielding a kernel write and control-flow hijack primitive. A:H - The UAF oopses or panics nfsd; the leaked stid sc_count also leaves move_to_close_lru() waiting forever for sc_count==2, hanging an nfsd thread, and the -EAGAIN retry can spin while RP_UNHASHED stays set.

Product Status

Vendor Linux
Product Linux
Versions Default: unaffected
  • affected from eec7620800081e27dbf8019ac2e66259f0d5bf6f to 69ed78b6b947c9257213164678b4edf17533093b (excl.)
  • affected from eec7620800081e27dbf8019ac2e66259f0d5bf6f to 00843074d9b84824552c9679d423d29500ca5de0 (excl.)
  • affected from eec7620800081e27dbf8019ac2e66259f0d5bf6f to f7cb90ddc021747fc9abfd4cf5252d425fd34eec (excl.)
  • affected from eec7620800081e27dbf8019ac2e66259f0d5bf6f to 5e4627d3513e60accfce9d5f4c7fa95251ef93d6 (excl.)
Vendor Linux
Product Linux
Versions Default: affected
  • Version 6.10 is affected
  • unaffected from 0 to 6.10 (excl.)
  • unaffected from 6.12.109 to 6.12.* (incl.)
  • unaffected from 6.18.50 to 6.18.* (incl.)
  • unaffected from 7.2.4 to 7.2.* (incl.)
  • unaffected from 7.3-rc1 to * (incl.)

References