[CVE-2026-89681][MODERATE 7.0] nfsd: fix layout fence worker double-reference race

[CVE-2026-89681][MODERATE 7.0] nfsd: fix layout fence worker double-reference race

[AL-KERNEL]

CVE: CVE-2026-89681
Priority: MODERATE 7.0
AL-KERNEL base severity: MODERATE
KPANIC flag: YES
Patch: nfsd: fix layout fence worker double-reference race
Commit: a278d361e0e8f242211891193d56483a7a9f47a8
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a278d361e0e8f242211891193d56483a7a9f47a8
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89681
Analysis date: Sat, 12 Sep 2026 06:38:11 -0400
ActionableScore: 4
ActionableScore lower bound: 3
Actionable bucket: Borderline / manual review recommended
Manual review required: YES

Summary:
A race in NFSD pNFS layout fencing can leak layout stateid references when a delayed worker is already running but appears non-pending, allowing repeated client activity to cause gradual server resource exhaustion.

======================================================================
ABOUT THIS REPORT
======================================================================

The original Linux kernel CVE announcement for CVE-2026-89681 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89681

The original announcement does not normally provide a security severity
estimate, CVSS assessment, or enough information to determine whether the
reported kernel bug represents a practically relevant security issue.

This report was generated by AL-KERNEL, an AI-assisted Linux kernel
vulnerability analysis system developed by Alexander Larkin. It combines
an autonomous classifier with LLM-assisted technical analysis and a
separate ActionableScore mechanism.

The purpose of this report is to prioritize Linux kernel CVEs before
manual review, identify cases that require prompt investigation, and
support automatic closure of issues that are unlikely to have meaningful
security impact.

Published priority for this report: MODERATE 7.0
Manual review required: YES

A detailed explanation of the methodology and priority rules is included
at the end of this message.

======================================================================
AL-KERNEL CLASSIFICATION RESULT
======================================================================

CVE-2026-89681	MODERATE	https://www.kernelcve.org/list/?q=CVE-2026-89681 CHECK WITH IMPACT FROM ORIG NN LOW	Maybe valid. Check manually. Hints by AL-KERNEL: The best (paranoid) CVSS is 'AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H';CWE-362;*CWE-401;*CWE-772;Other CVSS 'AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L';BEST CVSS score: '5.3';DESCR 'A race in NFS server layout fencing can occur because the workqueue clears WORK_STRUCT_PENDING before the fence worker callback starts. The lease manager timeout path can then treat the worker as not pending, take an additional layout stateid reference, and schedule work that coalesces with the worker already in progress. The additional reference is not released, causing a layout stateid and related NFS server resources to leak. For the CVSS the PR:L reflects that an attacker normally needs access to an NFSv4 export that permits pNFS layout operations. The issue is network reachable through NFSv4 client activity, but reliable triggering requires winning a timing race and repeatedly exercising the affected layout fencing path. Impact is primarily denial of service through gradual resource exhaustion. There is no demonstrated use after free, invalid memory access, information disclosure, integrity impact, or privilege escalation primitive in this patch.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Borderline / manual review recommended (with actual score 4)  SKIP CVE-2026-89681 UNKNOWN SKIP The Fixes patch not applied yet, so unlikely that actual: url YES  NO NO unknown 	MAYBE	LOCK DISK RACE LEAK NETWORK KPANIC 	-	-	checked

======================================================================
ACTIONABLESCORE ANALYSIS
======================================================================

ActionableScore=4
ActionableScoreLower=3
PrimitiveClass=NONE
RealUAF=NO
HostAdminRequired=UNKNOWN
NamespaceScopedPrivilege=UNKNOWN
ImportantCandidate=NO
ManualReview=RECOMMENDED
AutoDowngradeAllowed=YES

## 1. ActionableScore

* Raw conservative score: 3
* Conservative score: 3
* Raw paranoid score: 4
* Paranoid score: 4
* Final recommended bucket: **Borderline / manual review recommended**::

No floor or cap was applied.

## 2. Signal breakdown

Positive signals:

* +2 Remote reachable / directly network-triggerable. An NFSv4 client can reach the affected NFSD layout fencing path through network protocol activity when the server provides the relevant pNFS layout functionality and export access.
* +1 Concrete persistent kernel resource/reference leak. The patch directly establishes that `refcount_inc_not_zero()` can take an extra `sc_count` reference on the layout stateid, while coalescing prevents a matching worker execution and corresponding `nfs4_put_stid()`.
* +1 paranoid only: plausible persistent resource exhaustion. If an eligible NFS client can repeatedly induce the timing condition across many layout-state objects, leaked stateid references and associated NFSD resources may accumulate beyond a single harmless leak.

Negative signals:

* No hard-race penalty was applied. The bug is a race, but the supplied evidence does not establish a rare one-shot or externally uncontrollable timing window. The workqueue pending-state transition and lease-manager timeout processing provide a concrete repeatable concurrent condition.
* No host-admin penalty was applied. The input does not establish that host-root-equivalent privilege is required by the complete attacker path.
* No niche-subsystem penalty was applied. pNFS layout support is conditional deployment functionality, but the prompt does not establish that it is unusually rare or unreachable once enabled.

Important non-awarded signals:

* No UAF signal. The duplicate operation is `refcount_inc_not_zero()`, not a release, put, free, or post-free dereference.
* No concrete repeated release/refcount mutation exists. The defect leaks an additional reference because the extra reference is never put. It does not decrement or mutate a potentially freed refcount.
* Controlled reclaim, object replacement, arbitrary write, and callback influence are not demonstrated.
* Confidentiality and integrity impact are not awarded. The patch establishes a reference leak, not a disclosure, unauthorized modification, or memory corruption primitive.
* Strong LPE plausibility is not awarded. There is no evidence of credential, policy, ownership, or active-object corruption.
* Reliable immediate kernel crash is not demonstrated. The stated consequence is resource leakage and possible gradual exhaustion.
* Additional availability points are not awarded beyond the paranoid accumulation interpretation, because host-wide exhaustion is not directly demonstrated.

## 3. Reachability analysis

The relevant attacker is an NFSv4 client communicating with an NFSD server that exposes the affected pNFS layout-management path. The client must be able to access an export and exercise operations that create layout state and lead to lease-manager fencing behavior.

The supplied patch does not prove whether this requires ordinary authenticated export access, a delegated service identity, special NFS authorization, or host-administrator-equivalent privileges. Therefore `HostAdminRequired=UNKNOWN` is appropriate. It would be incorrect to assume host administration is required solely because the server must already be configured for NFS and pNFS.

No namespace capability path is established. The patch contains no evidence that a client-created namespace, namespace-scoped capability, or delegated network namespace can perform the complete server-side trigger sequence. Therefore `NamespaceScopedPrivilege=UNKNOWN`.

The NFS service and affected pNFS functionality must already be enabled and exposed by an administrator. That is deployment setup, but it is not evidence that an attacker who can access the export needs host administrative privilege. No special hardware, boot parameter, or firmware interaction is indicated.

The race occurs when `WORK_STRUCT_PENDING` is cleared before the delayed-work callback executes, allowing `lm_breaker_timedout()` to schedule coalescing work and acquire another reference. Repeated attempts may be possible through repeated layout and fencing activity, but the supplied patch does not quantify the rate or demonstrate practical host-memory exhaustion.

## 4. Primitive and severity interpretation

The exact primitive is a race-induced persistent reference leak in an active `nfs4_layout_stateid`. The lease-manager timeout path can observe the delayed work as non-pending while a fence worker is already executing. It then successfully increments `ls_stid.sc_count`, marks a new operation as pending, and schedules work that coalesces with the already-running worker. The extra reference has no matching `nfs4_put_stid()`.

This is neither an active structured-object UAF nor fixed stale storage. The affected object remains valid because the defect adds an unmatched reference, extending rather than prematurely ending its lifetime. The patch does not show a stale dereference, use after free, double free, refcount decrement through freed memory, object replacement, or post-free structured-field consumption.

There is no concrete repeated release/refcount operation:

A. Ordinary stale cleanup dereference is not present.

B. A second lifetime/refcount mutation through a potentially dangling object is not present. The problematic operation is an additional reference increment on a still-live object. The failure is that this reference is later leaked.

The demonstrated consequence is gradual memory or NFSD state-resource retention. The speculative consequence is denial of service through sufficiently sustained resource exhaustion. The patch does not establish a kernel crash, confidentiality impact, integrity impact, privilege escalation, or a security-boundary bypass.

`PrimitiveClass=NONE` is appropriate because a concrete resource-management flaw exists, but no meaningful corruption, stale-lifetime use, authorization bypass, or other exploitation primitive is established. `ImportantCandidate=NO` is appropriate because the leak may be operationally relevant but lacks unresolved corruption leverage. `AutoDowngradeAllowed=YES` is appropriate because this is a resource/reference leak without demonstrated UAF, memory corruption, policy bypass, or post-lifetime mutation.

Likely operational severity is Low to ordinary Moderate for most deployments, rising to a manual-review case when an exposed pNFS service permits untrusted or numerous clients to repeatedly exercise layout fencing.

## 5. One-sentence report phrase

A race in NFSD pNFS layout fencing can leak layout stateid references when a delayed worker is already running but appears non-pending, allowing repeated client activity to cause gradual server resource exhaustion.

## 6. Manual review recommendation

MANUAL CHECK RECOMMENDED

Confirm whether pNFS layouts are enabled, whether NFSv4 exports are reachable by untrusted clients, and whether repeated layout fencing can accumulate leaked stateids quickly enough to cause material memory or NFSD resource exhaustion.

## 7. Call-site confidence

Call-site confidence: medium

======================================================================
UPSTREAM PATCH SUMMARY
======================================================================

Patch: nfsd: fix layout fence worker double-reference race
Commit: a278d361e0e8f242211891193d56483a7a9f47a8
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a278d361e0e8f242211891193d56483a7a9f47a8

Commit description:

The workqueue core clears WORK_STRUCT_PENDING before the callback
is invoked, so delayed_work_pending() in lm_breaker_timedout() can
return false while the fence worker is already running. This lets
the breaker take a duplicate sc_count reference and schedule a new
worker that coalesces with the in-progress one. The extra reference
is never put, leaking the layout stateid.

Replace the racy delayed_work_pending() check with an
ls_fence_inflight boolean set atomically with
refcount_inc_not_zero() under ls_lock, and cleared under ls_lock
before the final nfs4_put_stid() on the dispose path; the retry
path intentionally retains it.  Remove the self-rearm
mod_delayed_work() at the top of the worker.

Fixes: f52792f ("NFSD: Enforce timeout on layout recall and integrate lease manager fencing")
Cc: stable@vger.kernel.org
Assisted-by: kres:claude-opus-4-7
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Link: https://patch.msgid.link/20260531-nfsd-testing-v1-6-7bfa481b0540@kernel.org
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Changed files:
  fs/nfsd/nfs4layouts.c
  fs/nfsd/state.h

Diff excerpt:

Not included in this email. See the upstream URL for the full patch.

Full patch:
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a278d361e0e8f242211891193d56483a7a9f47a8

======================================================================
DETAILED REPORT METHODOLOGY
======================================================================

The original Linux kernel CVE announcement for CVE-2026-89681 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89681

The original CVE announcement normally does not include a security-level
estimate. In particular, it may not contain a CVSS assessment, an impact
level, or enough information to determine whether the reported bug is a
practically relevant security issue. One purpose of this parallel CVE list
is to provide that missing technical and prioritization information.

The original goal of the AL-KERNEL project was to prioritize Linux kernel
CVE analysis automatically before manual review. The system can also help
identify non-security issues that may be suitable for automatic closure.

This report was generated by AL-KERNEL, an AI-assisted Linux kernel
vulnerability analysis system developed by Alexander Larkin.

The first analysis stage combines an autonomous classifier with additional
LLM-based analysis. The autonomous classifier runs locally on a CPU and is
based on a backpropagation neural network. Together, these mechanisms
produce a technical vulnerability description, identify likely weakness
types, estimate CVSS severity, and provide input for ActionableScore.

Two CVSS estimates are retained because incomplete kernel vulnerability
information often permits more than one defensible interpretation:

  Conservative CVSS vector: AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:L
  The Best / paranoid CVSS vector: AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H
  The Best / paranoid CVSS score: 5.3

The conservative vector represents a lower-impact interpretation.
The Best/paranoid vector intentionally represents a plausible upper-bound
interpretation and should not automatically be treated as demonstrated
real-world impact.

CVSS may also need to be adjusted for a particular Linux deployment,
because actual reachability, privileges, enabled kernel configuration,
hardware, namespaces, exposed device nodes, and other environmental
conditions can differ significantly between systems.

A separate ActionableScore mechanism evaluates practical remediation
urgency. Its analysis may include reachability, attack prerequisites,
subsystem exposure, memory-corruption characteristics, denial-of-service
reliability, and possible confidentiality, integrity, or
privilege-escalation impact.

  Conservative ActionableScore: 3
  Paranoid ActionableScore: 4

The final base severity is taken directly from the second tab-separated
field of the AL-KERNEL classification result. ActionableScore does not
replace or independently override that final AL-KERNEL decision, and
if ActionableScore adjusted impact level of ALKERNEL, then you would see
self-readable flags above like INCREASED_TO_HIGH_BASED_ON_ACTIONABLESCOREHIGHEREQTHAN7.

For an AL-KERNEL result of MODERATE, this report uses the following
additional presentation split:

  ActionableScore below 5   -> MODERATE REGULAR
  ActionableScore 5 or more -> MODERATE 7.0

The distinction between MODERATE REGULAR and MODERATE 7.0 makes it
possible to identify Moderate issues that should receive manual analysis
and fixes before lower-priority MODERATE REGULAR issues. In many cases,
MODERATE REGULAR fixes may wait for a later rebase or routine update.

There is one override in which MODERATE REGULAR becomes MODERATE 7.0
even when the ActionableScore is below 5. When the AL-KERNEL result
contains the KPANIC flag, a MODERATE result is always presented as
MODERATE 7.0. The KPANIC flag selected with few regexps without
usage of AI at all, so it helps to detect cases when Kernel Crash happens
and similar (to filter False-Negative results from the LLM usage).

KPANIC indicates that a reliable kernel crash, kernel panic, or similarly
serious kernel availability impact was identified by the classification
workflow.

AL-KERNEL base severity for this report: MODERATE
KPANIC detected for this report: YES
Published priority for this report (same as in Subject): MODERATE 7.0

These results are intended to support engineering triage. They are
machine-generated estimates, and cases marked for manual review should
be validated by a human security engineer before final disposition.
For more info read docs linked from here: https://kernelcve.org/
(and you can submit you own patch there to generate such a report
for non-existant CVE-id yet).

Note that in many cases this AI tool selects higher severity, than
real is (means you can expect Importants instead of Moderate 7.0 or
Moderates 7.0 instead of regular Moderates). If you see such cases,
please use reply email interface to add additional manual analyses
info to this particular CVE.
And please, please, let me know when you see Lows instead of Importants
or Important instead of Low (because particular for such cases I
need to tune this AI tool to make it better for this one and next similar).
My contact email for such notifications is alexanjelausa@gmail.com (and both
send reply to CVE record itself too and see "reply" button below for howto reply).