[CVE-2026-98429][MODERATE REGULAR] RDMA/efa: Keep EQ resources alive while IRQ is registered [ Upstream

[AL-KERNEL <alkernel@kernelcve.org>]

CVE: CVE-2026-98429
Priority: MODERATE REGULAR
AL-KERNEL base severity: MODERATE
KPANIC flag: NO
Patch: RDMA/efa: Keep EQ resources alive while IRQ is registered [ Upstream
Commit: 3cd652dfcd822af26ba75d6d2d245ea7f73b499e
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=3cd652dfcd822af26ba75d6d2d245ea7f73b499e
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98429
Analysis date: Thu, 15 Oct 2026 09:42:44 -0400
ActionableScore: 3
ActionableScore lower bound: unknown
Actionable bucket: Borderline / manual review recommended
Manual review required: YES

Summary:
No one-sentence report phrase was found.

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

The original Linux kernel CVE announcement for CVE-2026-98429 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98429

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 REGULAR
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-2026-98429	MODERATE	https://www.kernelcve.org/list/?q=CVE-2026-98429 CHECK WITH IMPACT FROM ORIG NN MODERATE | AEGIS IMPACT MODERATE_KPANIC AEGIS_AS=2 LEGACY=MODERATE RESULT=DIFF (independent Legacy and Aegis differ; FINAL: MODERATE, Aegis: MODERATE_KPANIC)	Maybe valid. Check manually. Hints by AL-KERNEL: The best (paranoid) CVSS is 'AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H';*CWE-416;CWE-362;*CWE-664;Other CVSS 'AV:L/AC:H/PR:H/UI:N/S:U/C:L/I:L/A:H';BEST CVSS score: '7';DESCR 'The EFA completion IRQ handler can access EQ state and its DMA buffer before initialization is complete or after those resources have been released. During teardown the previous ordering destroyed the EQ before free_irq synchronized with active handlers, creating a race and a use after free window that can cause a kernel crash and may permit memory corruption. For the CVSS the PR:L reflects the paranoid case where a reduced capability process or service with delegated RDMA device access can generate completion activity without holding full host administrative privileges. AC:H applies because exploitation depends on interrupt timing and overlap with EQ teardown. The issue is not proven to be remotely exploitable from an arbitrary network. A connected RDMA peer may influence completion activity, but reliable control of the teardown race requires manual verification. Impact is at least denial of service and in the worst defensible case may include confidentiality and integrity impact or privilege escalation due to use after free access to kernel and DMA backed state. Automated impact decision path. The original SimpleNet classifier selected MODERATE_KPANIC. The primary LLM estimated paranoid CVSS 7 and the production ActionableScore returned 3 with recommendation Borderline / manual review recommended. The production rules then evaluated CVSS thresholds and ActionableScore thresholds. Recorded rule transitions were ActionableScore thresholds stage MODERATE_KPANIC to MODERATE because the ActionableScore threshold rule was applied. The legacy KP specialist has no review artifact for this run. The legacy FD or DH downgrade specialist has no review artifact for this run. Exact triggered rule markers were DECREASED_TO_MODERATEREG_BASED_ON_ACTIONABLESCORELESSTHAN4. The independent weaker Aegis shadow evaluated CVSS 4.8 and ActionableScore 2 with primitive class CLASS_B and manual review RECOMMENDED. Its independently reconciled impact was MODERATE_KPANIC. The models disagreed but the aggregation rules retained the production result. The final reported impact is MODERATE.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Borderline / manual review recommended (with actual score 3)	MAYBE	INIT UAF NETWORK DMAorINTERRUPT LINUS  DECREASED_TO_MODERATEREG_BASED_ON_ACTIONABLESCORELESSTHAN4	-	-	checked

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

ActionableScore=3
PrimitiveClass=CLASS_B
RealUAF=YES
HostAdminRequired=UNKNOWN
NamespaceScopedPrivilege=UNKNOWN
ImportantCandidate=NO
ManualReview=RECOMMENDED
AutoDowngradeAllowed=YES
RequiresHostUID0=UNKNOWN
AuthorizationDecisionBypass=NO
ProtectedResourceImpact=NONE
B4PendingReview=NO
B4PendingConfirmed=NO

1. ActionableScore

* Raw conservative score: 3
* Conservative score: 3
* Raw paranoid score: 3
* Paranoid score: 3
* No floor or cap was applied.
* Final recommended bucket: **Borderline / manual review recommended**::

2. Signal breakdown

* +2 Real UAF / lifetime violation: teardown destroyed the EQ state and DMA buffer before `free_irq()` synchronized with the completion IRQ handler, allowing post-lifetime access.
* +1 Reliable kernel crash / strong DoS: an IRQ handler accessing freed EQ state or its DMA buffer can realistically cause an Oops or host crash.
* +0 Initialization-order defect: registering the IRQ before initializing and arming the EQ permits access to incomplete state, but this is part of the same ordering defect and is not counted separately.
* +0 Firmware-mediated or remote reachability: completion events are device-mediated, but the evidence does not establish that a remote RDMA peer controls the complete vulnerable sequence.
* +0 Structured stale-object leverage: the supplied code does not show reclaimed data controlling callbacks, pointers, offsets, or another useful post-free operation.
* +0 Strong or weak LPE leverage: controlled reclaim, object substitution, arbitrary write, callback control, disclosure, and privilege escalation are not demonstrated.
* +0 Concrete repeated release/refcount: no second `put`, `free`, or refcount mutation through the same dangling reference exists. The dedicated +1 and FLOOR 5 do not apply.
* +0 Extra Availability: not awarded because the same possible kernel crash is already counted.
* +0 Race penalty: the required IRQ overlap is evident, but the supplied evidence does not establish that it is sufficiently narrow, one-shot, or difficult to repeat to justify -1.
* +0 Hardware rarity penalty: EFA hardware is required, but driver specificity alone does not establish the rare or physical-only condition required for a penalty.
* NEW52 B5-R does not apply. There is no retry-accumulated heap OOB write.
* NEW53 B3-PENDING does not apply. There is no shared-backing or file-backed write.
* NEW54 B4-PENDING does not apply. No authorization decision or DENY-to-ALLOW transition is involved.

3. Reachability analysis

The vulnerable lifetime sequence occurs while creating or destroying a device-level EFA completion EQ. The IRQ handler can run before initialization completes or concurrently with teardown after EQ resources have been released.

The supplied context does not establish who can initiate the complete teardown sequence. Host administration may be required for device removal or driver teardown, but reset, error-recovery, or device-failure paths may also initiate it. Therefore `HostAdminRequired=UNKNOWN`, and no privilege penalty is applied.

A process with RDMA access or a connected peer may influence completion activity, but the evidence does not establish control over both completion generation and EQ teardown. Direct remote reachability and below-host-admin triggering are therefore not awarded. No namespace-scoped complete trigger is shown.

EFA-capable hardware and an initialized driver are prerequisites. This is prior environmental setup, but it is not independently scored as rare hardware. Race repeatability remains unresolved because the interrupt handler and complete teardown call paths are not supplied.

4. Primitive and severity interpretation

The teardown defect is a real fixed-storage asynchronous UAF: `efa_com_eq_destroy()` can release EQ state and its DMA buffer while the registered completion IRQ handler can still access them. The affected storage is a fixed driver EQ context and associated DMA buffer rather than a demonstrated attacker-controlled replacement object.

The stale use occurs around initialization and teardown. There is no demonstrated controlled reclaim, attacker-shaped replacement, repeated refcount mutation, callback dispatch through reclaimed contents, or directed kernel read/write. It therefore remains `CLASS_B`, not `CLASS_A`.

The demonstrated practical consequence is invalid post-lifetime access with plausible kernel crash or DoS. Memory corruption beyond that access is possible in the generic sense, but no independently useful corruption primitive is established.

`ImportantCandidate=NO` because this is presently a constrained asynchronous driver UAF without meaningful demonstrated leverage beyond crash. `AutoDowngradeAllowed=YES` because no unresolved repeated lifetime mutation, controlled reuse, authorization bypass, shared-page violation, or other Important-preserving primitive remains. Operationally, this is a borderline Moderate driver-lifetime issue that merits confirmation of the IRQ handler behavior and teardown reachability.

5. One-sentence report phrase

The EFA driver could destroy EQ state and its DMA buffer before synchronizing the registered completion IRQ handler, creating a constrained teardown use-after-free with a plausible kernel crash consequence.

6. Manual review recommendation

MANUAL CHECK RECOMMENDED

Confirm whether an ordinary RDMA user or remote peer can reliably generate completion IRQ activity during a reachable reset or teardown path, and inspect which EQ fields the stale handler consumes.

7. Call-site confidence

Call-site confidence: medium

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

Patch: RDMA/efa: Keep EQ resources alive while IRQ is registered [ Upstream
Commit: 3cd652dfcd822af26ba75d6d2d245ea7f73b499e
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=3cd652dfcd822af26ba75d6d2d245ea7f73b499e

Changed files:
  drivers/infiniband/hw/efa/efa_com.c
  drivers/infiniband/hw/efa/efa_com.h
  drivers/infiniband/hw/efa/efa_main.c

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=3cd652dfcd822af26ba75d6d2d245ea7f73b499e

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

The original Linux kernel CVE announcement for CVE-2026-98429 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-98429

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:L/AC:H/PR:H/UI:N/S:U/C:L/I:L/A:H
  The Best / paranoid CVSS vector: AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H
  The Best / paranoid CVSS score: 7

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: unknown
  Paranoid ActionableScore: 3

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: NO
Published priority for this report (same as in Subject): MODERATE REGULAR

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).