[CVE-2026-89811][IMPORTANT] drm/amdkfd: Add TLB flush after MES queue eviction/suspension

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

CVE: CVE-2026-89811
Priority: IMPORTANT
AL-KERNEL base severity: IMPORTANT
KPANIC flag: YES
Patch: drm/amdkfd: Add TLB flush after MES queue eviction/suspension
Commit: e230c546ed93741833ab5babb0b202f9c2052b45
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=e230c546ed93741833ab5babb0b202f9c2052b45
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89811
Analysis date: Wed, 16 Sep 2026 13:46:31 -0400
ActionableScore: 6
ActionableScore lower bound: 5
Actionable bucket: Actionable Moderate / Strong Important candidate requiring manual review
Manual review required: YES

Summary:
No one-sentence report phrase was found.

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

The original Linux kernel CVE announcement for CVE-2026-89811 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89811

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: IMPORTANT
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-2026-89811	IMPORTANT	https://www.kernelcve.org/list/?q=CVE-2026-89811 CHECK WITH IMPACT FROM ORIG NN MODERATE	Maybe valid. Check manually. Hints by AL-KERNEL: The best (paranoid) CVSS is 'AV:L/AC:H/PR:L/UI:N/S:U/C:L/I:L/A:H';CWE-362;CWE-667;*CWE-664;Other CVSS 'AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:N/A:H';BEST CVSS score: '5.8';DESCR 'A race in AMD KFD with MES can leave stale GPU TLB translations after queues are evicted or suspended. In flight DMA descriptors may continue to access memory that was already unmapped, freed, or migrated during SVM page migration, which can lead to GPU page faults and compute queue hangs. For the CVSS the PR:L is used because a local user or process with access to AMD KFD GPU compute queues can exercise the affected queue and SVM migration paths. The issue is not network reachable and requires local GPU compute access on affected AMD hardware with MES enabled. Impact is at least denial of service through GPU queue hang. In the paranoid interpretation, stale DMA translations after unmap may also create limited confidentiality or integrity risk and should be reviewed manually.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Actionable Moderate / Strong Important candidate requiring manual review (with actual score 6)  SKIP CVE-2026-89811 UNKNOWN SKIP No affected files built, so skip this CVE NO  - - unknown 	MAYBE	INIT RACE UAF DMAorINTERRUPT HARDWARE KPANIC PACKET  KPANIC INCREASED_TO_HIGH_BASED_ON_GUESSCVSS	-	-	checked

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

ActionableScore=6
ActionableScoreLower=5
PrimitiveClass=CLASS_B
RealUAF=YES
HostAdminRequired=NO
NamespaceScopedPrivilege=NO
ImportantCandidate=YES
ManualReview=REQUIRED
AutoDowngradeAllowed=NO

1. ActionableScore

* Raw conservative score: 5
* Conservative score: 5. No floor or cap changed the score.
* Raw paranoid score: 6
* Paranoid score: 6. No floor or cap changed the score.
* Final recommended bucket: **Actionable Moderate / Strong Important candidate requiring manual review**::

2. Signal breakdown

Conservative positive signals:

* +2 Real stale lifetime / use-after-unmap primitive. MES removes or suspends queues without a heavy TLB flush, so in-flight DMA descriptors may continue using translations for memory that has already been unmapped, freed, or migrated.
* +1 Local below-host-admin reachability. A local process with AMD KFD GPU compute access can exercise queues and SVM migration paths. Host administrator privilege is not established as required.
* +1 Availability impact. The demonstrated consequence is GPU page faults and compute queue hangs with submitted packets not consumed.
* +1 Active translation synchronization violation. The fix restores a required heavy-weight TLB flush after queue removal or suspension, showing that stale translations may remain operationally active outside the immediate software queue state.

Paranoid additional signal:

* +1 Unresolved DMA ownership / memory-access concern. The stale GPU translation can point at memory after unmap, free, or migration. This is stronger than a simple device hang, although controlled target selection, controlled payload, or arbitrary host memory access is not demonstrated.

Non-awarded signals:

* +0 Remote reachability. No network trigger is established.
* +0 Controlled reclaim or controlled replacement. Not demonstrated.
* +0 Arbitrary or directed kernel write. Not demonstrated.
* +0 Strong LPE / strong exploitation leverage. Not established.
* +0 Extra availability. The queue hang is already counted once.
* +0 Concrete repeated release/refcount. No second put, free, or refcount mutation on the same dangling object is shown.
* +0 Race penalty. The issue is a race, but the evidence does not establish a narrow one-shot timing window that should reduce the score.
* +0 Rare hardware penalty. The affected path is AMD GPU/MES/XNACK specific, but common GPU hardware is not automatically rare physical-only hardware.

3. Reachability analysis

The evaluated attacker is a local user or local process with access to AMD KFD GPU compute queues. This is below host administrator in common deployments where render/GPU compute access is delegated to users or workloads.

The complete trigger requires affected AMD GPU hardware with MES enabled and SVM page migration activity. The supplied evidence mentions gfx1151 Strix Point with XNACK mode 1, but this is not enough to classify the issue as physical-only or rare hardware for scoring.

The issue is not namespace-scoped. No network namespace, user namespace, or namespaced capability chain is relevant.

No prior host-admin setup is proven as a required part of exploitation beyond the system already having the affected GPU driver and feature path available. The race appears tied to normal queue eviction, suspension, debug/CRIU suspension, and SVM migration ordering rather than to a single externally uncontrollable timing edge.

4. Primitive and severity interpretation

The exact primitive is stale GPU TLB translation use after MES queue eviction or suspension. In-flight DMA descriptors may keep accessing memory after the software path has unmapped the queues and before memory is freed or migrated.

This is best treated as an active translation lifetime violation, not a simple NULL dereference and not a pure device compatibility quirk. It is also not a demonstrated arbitrary write or controlled host memory corruption primitive.

The demonstrated consequence is GPU page faults and GPU compute queue hang. The speculative but technically defensible consequence is limited confidentiality or integrity risk if stale DMA translations access memory after it has been migrated or reused. That uncertainty is why the paranoid score is higher and manual review is required.

PrimitiveClass is CLASS_B because a real active-state lifetime/translation primitive exists, but CLASS_A is not justified without controlled target selection, controlled DMA payload, arbitrary read/write, callback influence, or demonstrated privilege escalation.

ImportantCandidate=YES because stale DMA access after unmap/free/migration is meaningful unresolved kernel/device memory-access behavior that should not be auto-closed as ordinary DoS.

AutoDowngradeAllowed=NO because automatic downgrade would discard a concrete stale translation/DMA lifetime primitive with unresolved exploitation impact.

5. One-sentence report phrase

A race in AMD KFD MES queue eviction or suspension can leave stale GPU TLB translations active after queue memory is unmapped, allowing in-flight DMA descriptors to access freed or migrated memory and causing GPU page faults or queue hangs, with limited confidentiality or integrity impact requiring manual review.

6. Manual review recommendation

MANUAL CHECK REQUIRED

Reason: the demonstrated impact is DoS, but the primitive involves stale DMA translations after unmap/free/migration, which is stronger than a crash-only invalid pointer and may have unresolved memory-access consequences.

7. Call-site confidence

Call-site confidence: medium

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

Patch: drm/amdkfd: Add TLB flush after MES queue eviction/suspension
Commit: e230c546ed93741833ab5babb0b202f9c2052b45
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=e230c546ed93741833ab5babb0b202f9c2052b45

Commit description:

MES (Micro Engine Scheduler) does not perform heavy-weight TLB
invalidation after unmapping queues, unlike HWS which does this
automatically. This causes a race condition where in-flight DMA
descriptors can access memory that has been unmapped, leading to page
faults and GPU queue hangs during SVM page migration.

The issue manifests as KFDSVMRangeTest.MultiThreadMigrationTest
failures on gfx1151 (Strix Point) with XNACK mode 1 enabled - the GPU
compute queue hangs with packets submitted but never consumed.

Add kfd_flush_tlb() calls after MES queue removal in two locations:
- evict_process_queues_cpsch(): after all queues removed during eviction
- suspend_queues(): after debug/criu queue suspension (with mem_fence barrier)

This ensures all in-flight memory accesses from unmapped queues are
flushed before memory is freed or migrated.

Signed-off-by: Priya Hosur <Priya.Hosur@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit f5c4f88e0f9c45a8fb9dfac0c1df726c95e41b77)
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Changed files:
  drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.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=e230c546ed93741833ab5babb0b202f9c2052b45

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

The original Linux kernel CVE announcement for CVE-2026-89811 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-89811

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:L/UI:N/S:U/C:N/I:N/A:H
  The Best / paranoid CVSS vector: AV:L/AC:H/PR:L/UI:N/S:U/C:L/I:L/A:H
  The Best / paranoid CVSS score: 5.8

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: 5
  Paranoid ActionableScore: 6

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: IMPORTANT
KPANIC detected for this report: YES
Published priority for this report (same as in Subject): IMPORTANT

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).