[CVE-2026-68103][MODERATE 7.0] drm/amdgpu: reject mapping a reserved doorbell to a new queue

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

CVE: CVE-2026-68103
Priority: MODERATE 7.0
AL-KERNEL base severity: MODERATE
KPANIC flag: NO
Patch: drm/amdgpu: reject mapping a reserved doorbell to a new queue
Commit: 1050d258c7c56066d2dcaedf8d0ef66364062adc
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=1050d258c7c56066d2dcaedf8d0ef66364062adc
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-68103
Analysis date: Mon, 10 Aug 2026 09:07:16 -0400
ActionableScore: 5
ActionableScore lower bound: 4
Actionable bucket: Actionable Moderate at minimum
Manual review required: NO

Summary:
AMDGPU user queue creation could overwrite an existing doorbell mapping for the same BO and offset, causing wrong queue fence IRQ routing and cleanup confusion, with local PR:L access to the GPU render device and practical impact mainly as GPU driver state corruption or DoS.

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

The original Linux kernel CVE announcement for CVE-2026-68103 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-68103

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: NO

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

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

CVE-2026-68103	MODERATE	CHECK WITH IMPACT FROM ORIG NN LOW	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-20;CWE-284;*CWE-664;Other CVSS 'AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L';BEST CVSS score: '5.8';DESCR 'AMDGPU user queue creation can overwrite an existing doorbell mapping when a later queue is created with the same doorbell BO and offset. This can misroute fence IRQ processing to the wrong queue and can make cleanup for one queue remove the mapping that belongs to another queue. For the CVSS the PR:L is used because a local user with access to the AMDGPU render device can create user queues without full administrator privileges in many deployments. The issue is not network reachable and requires local GPU device access. Impact is mainly denial of service or incorrect GPU queue and fence handling, with limited confidentiality or integrity impact kept only for paranoid triage because the patch shows state confusion but not a direct arbitrary write primitive.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Actionable Moderate at minimum (with actual score 5)	MAYBE	HARDWARE  INCREASED_FROM_LOW_BASED_ON_REQUIREMANUALCHECK DECREASED_TO_MODERATEREG_BASED_ON_FALSEPOSCHECKOFKP	-	-	guess

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

ActionableScore=5
ActionableScoreLower=4

## 1. ActionableScore

* Conservative score: 4
* Paranoid score: 5
* Final recommended bucket: **Actionable Moderate at minimum**::

## 2. Signal breakdown

Conservative signals:

* Local unprivileged trigger: +1
  A local process with access to the AMDGPU DRM render device can create user queues in many deployments. This does not usually require full host root.
* Memory corruption, weak/indirect corruption candidate: +1
  The bug allows an existing xarray doorbell mapping to be overwritten with another queue pointer. This is kernel object state confusion, but the patch does not show an arbitrary write, UAF reclaim, type confusion, or callback hijack primitive.
* Integrity impact plausible: +1
  Fence IRQ processing can be routed to the wrong queue, and cleanup for one queue can erase another queue mapping.
* Availability impact realistic: +1
  Incorrect queue and doorbell ownership can disrupt GPU queue operation and cause GPU workload failure or driver malfunction.
* Corruption primitive highly constrained or metadata-only: -1
  The overwritten object is an xarray mapping entry for a doorbell index. The payload is a kernel queue pointer selected through queue creation, not an attacker-controlled arbitrary value.

Conservative total: 4

Paranoid additional signal:

* Weak LPE concern: +1
  The affected state participates in IRQ fence routing and cleanup ownership. There is no demonstrated privilege escalation primitive, but the wrong-queue mapping and wrong cleanup target make this worth manual review.

Paranoid total: 5

Not awarded:

* Remote reachable: +0
  No network path.
* Firmware-mediated external influence: +0
  The trigger is user queue creation, not firmware input.
* Strong memory corruption primitive: +0
  No UAF reclaim, double free, OOB write, arbitrary write, refcount takeover, or controlled overwrite target is shown.
* High-control GPU API bonus: +0
  GPU object creation alone is not enough. The patch shows object mapping confusion, but not attacker control over memory layout, callbacks, DMA ownership, or reclaim behavior.

## 3. Reachability analysis

The realistic trigger is local access to the AMDGPU user queue creation interface through DRM device access. In many systems render nodes are available to non-root users in the video or render group, so PR:L is the practical assumption rather than PR:H. Containers may matter if GPU render devices are passed through into a container, because a container user could potentially exercise the same DRM interface against the host driver.

The path is not network reachable. It depends on AMDGPU hardware and support for the user queue feature, so exposure is narrower than core kernel networking or filesystems. However, AMDGPU is common enough that this should not be treated as a rare research-only path when render device access is available.

## 4. Severity interpretation

This behaves like an actionable Moderate issue rather than a clear Important vulnerability. The bug is not a simple crash-only issue because it can corrupt kernel driver ownership state for doorbell to queue mappings and can misroute fence IRQ processing. At the same time, the patch does not demonstrate a strong memory corruption primitive, arbitrary write, UAF reclaim, controlled object replacement, or direct privilege escalation path.

Theoretical impact includes limited confidentiality or integrity concern from wrong queue association and cleanup confusion. Realistic evidence supports GPU queue disruption, incorrect fence handling, and local DoS more strongly than LPE. Manual review is justified because IRQ and cleanup state confusion in a GPU driver can have subtle consequences not visible from the small diff alone.

## 5. One-sentence report phrase

AMDGPU user queue creation could overwrite an existing doorbell mapping for the same BO and offset, causing wrong queue fence IRQ routing and cleanup confusion, with local PR:L access to the GPU render device and practical impact mainly as GPU driver state corruption or DoS.

## 6. Manual review recommendation

MANUAL CHECK RECOMMENDED

Reason: the issue is not proven LPE or strong memory corruption, but it affects kernel GPU queue ownership, fence IRQ routing, and cleanup of another queue mapping. This is above ordinary auto-close Moderate quality and should receive at least targeted driver review.

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

Patch: drm/amdgpu: reject mapping a reserved doorbell to a new queue
Commit: 1050d258c7c56066d2dcaedf8d0ef66364062adc
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=1050d258c7c56066d2dcaedf8d0ef66364062adc

Commit description:

When creating an user-queue, the user space
provides a doorbell BO handle and an offset within
the bo to obtain a doorbell.

However current implementation using xa_store_irq()
to store a doorbell, which allows a later queue created
with the same BO and offset parameters to overwrite an
existing queue and doorbell mapping.

This can cause problems like misrouting fence IRQ
processing to a wrong queue, and mislead the cleanup
process of one queue erasing the mapping of another queue.

This commit fixes this issue by replacing xa_store_irq with
xa_insert_irq, which rejects mapping a reserved
doorbell to a newly created queue

Signed-off-by: Zhu Lingshan <lingshan.zhu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 6244eae )
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Changed files:
  drivers/gpu/drm/amd/amdgpu/amdgpu_userq.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=1050d258c7c56066d2dcaedf8d0ef66364062adc

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

The original Linux kernel CVE announcement for CVE-2026-68103 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-68103

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:L/PR:L/UI:N/S:U/C:N/I:L/A:L
  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: 4
  Paranoid ActionableScore: 5

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 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).