[CVE-2026-90240][MODERATE 7.0] iommu/vt-d: Flush context cache with correct SID when tearing down aliases [ Upstream

[CVE-2026-90240][MODERATE 7.0] iommu/vt-d: Flush context cache with correct SID when tearing down aliases [ Upstream

[AL-KERNEL]

CVE: CVE-2026-90240
Priority: MODERATE 7.0
AL-KERNEL base severity: MODERATE
KPANIC flag: YES
Patch: iommu/vt-d: Flush context cache with correct SID when tearing down aliases [ Upstream
Commit: a803141597d61dc0511f7cd849625e17d042ca8a
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a803141597d61dc0511f7cd849625e17d042ca8a
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-90240
Analysis date: Fri, 18 Sep 2026 03:13:30 -0400
ActionableScore: 6
ActionableScore lower bound: 5
Actionable bucket: Actionable Moderate / Strong Important candidate requiring manual review
Manual review required: YES

Summary:
Intel VT-d may invalidate the wrong context-cache SID while tearing down DMA aliases, leaving stale IOMMU translation state that can reference a freed PASID directory and may cause host DoS or unresolved DMA translation memory corruption.

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

The original Linux kernel CVE announcement for CVE-2026-90240 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-90240

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-2026-90240	MODERATE	https://www.kernelcve.org/list/?q=CVE-2026-90240 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:H/I:H/A:H';*CWE-416;*CWE-672;*CWE-664;Other CVSS 'AV:L/AC:H/PR:L/UI:N/S:U/C:L/I:L/A:H';BEST CVSS score: '7';DESCR 'Intel VT-d may fail to invalidate the correct IOMMU context cache entry when tearing down a DMA alias because the flush used the device requester ID instead of the SID of the alias being cleared. The stale cached context can remain usable after the in memory context entry is cleared, and in scalable mode it may still reference a PASID directory that has already been freed. This is a use after free candidate in IOMMU translation state rather than a simple warning or resource leak. For the CVSS the PR:L value is used because practical triggering can be possible from a local actor that can drive device assignment, device reset, detach, or other delegated device management paths, even if full host root is not always required in virtualized environments. The issue is not directly network reachable. Impact is at least local denial of service via IOMMU faults or memory corruption, and in the paranoid case may include confidentiality and integrity impact if stale DMA translation state can be abused.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Actionable Moderate / Strong Important candidate requiring manual review (with actual score 6)	YES	INIT UAF DMAorINTERRUPT HARDWARE KPANIC MEMORY  KPANIC INCREASED_TO_HIGH_BASED_ON_GUESSCVSS DECREASED_TO_MODERATE7_BASED_ON_FALSEPOSCHECKOFKP	YES	NO	checked

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

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

## 1. ActionableScore

* Raw conservative score: 5
* Conservative score: 5
* Raw paranoid score: 6
* Paranoid score: 6
* Final recommended bucket: **Actionable Moderate / Strong Important candidate requiring manual review**::

No floor or cap was applied. The higher paranoid score is emitted as `ActionableScore` because the stale IOMMU translation state may continue referencing freed PASID memory and the exploitation leverage is not fully resolved.

## 2. Signal breakdown

Positive conservative signals:

* +2 Real UAF / lifetime violation: the commit directly states that a stale IOMMU context-cache entry may still reference a PASID directory after `intel_pasid_free_table()` frees it, allowing the IOMMU to walk freed memory.
* +1 Active-state retirement / synchronization bypass: the normal teardown clears the context entry and requires device-selective invalidation, but the vulnerable path invalidates the wrong SID for DMA aliases.
* +1 Active translation synchronization signal: the stale entry may remain operational in hardware context cache after the software context entry was cleared, so the old translation state can remain active outside the immediate software structure.
* +1 Reliable host DoS / strong availability concern: stale IOMMU translation or freed PASID-table traversal can plausibly cause IOMMU faults, memory corruption symptoms, or host instability.

Additional paranoid signal:

* +1 Privileged kernel / DMA translation-state corruption concern: the affected state is IOMMU/PASID/DMA translation lifetime state, and stale hardware use of freed translation tables leaves unresolved DMA isolation and memory-corruption implications.

Non-awarded signals:

* +0 Remote reachability: no packet or network-triggered path is established.
* +0 Local unprivileged trigger: below-host-admin reachability is not positively established.
* +0 Strong LPE / controlled corruption: no controlled reclaim, arbitrary write, callback control, or demonstrated privilege escalation is shown.
* +0 Confidentiality / integrity direct impact: stale DMA translation could theoretically affect C/I, but the patch does not establish a concrete read/write primitive against attacker-selected host memory.
* +0 Race penalty: this is a stale invalidation / teardown bug, not a proven hard timing race requiring a narrow uncontrollable window.
* +0 Rare hardware penalty: Intel VT-d and PASID/scalable-mode deployments are hardware/config dependent, but the supplied evidence does not justify treating the path as rare or physical-only.
* +0 Extra availability: not added separately because the same host instability/crash concern is already counted once.

Concrete repeated release/refcount check:

* First release can end lifetime: yes, `intel_pasid_free_table()` can free PASID directory state.
* Exact same software reference receives second put/release: no evidence of a second refcount or release operation on the same reference.
* Second operation mutates lifetime/refcount: no.
* Concrete repeated release/refcount +1 awarded: no.
* Paranoid repeated-refcount +1 awarded: no.
* FLOOR 5 from repeated release: not applicable.

## 3. Reachability analysis

The vulnerable paths are `domain_context_clear_one()` and `device_pasid_table_teardown()`, both invoked during teardown of Intel VT-d context/PASID state for PCI DMA aliases. Practical triggering requires an affected Intel IOMMU configuration, a device with DMA aliases, and teardown of the relevant context entries. The freed-PASID-directory path is specifically relevant to scalable mode / PASID handling.

The likely operational triggers are device detach, reset, hot-unplug, domain teardown, PASID-table teardown, or virtualization/device-assignment management. These operations are commonly controlled by host administration or a privileged device-management stack, but the supplied input does not positively prove that host-admin-equivalent privilege is always required for the complete vulnerable sequence. Therefore `HostAdminRequired=UNKNOWN`.

No namespace-scoped capability path is established. The bug is not directly network reachable. There is no evidence that ordinary packet traffic can trigger it. Prior setup may include enabling Intel VT-d/scalable mode and using a device/topology with DMA aliases, but the supplied evidence does not establish this as a rare or attacker-inaccessible configuration penalty.

Race repeatability is not the main issue. The primitive follows from invalidating the wrong SID during teardown, leaving stale hardware cache state for an alias.

## 4. Primitive and severity interpretation

The exact primitive is a stale active IOMMU context-cache entry for a DMA alias. The software context entry is cleared, but the hardware cache invalidation targets the device requester ID instead of the alias SID. For aliases other than the device RID, the cleared entry may remain cached and usable by hardware.

This is an active translation-state lifetime violation, not fixed stale storage and not a simple warning. In scalable mode, the stale context entry may still point to a PASID directory that has been freed, so hardware may walk freed memory. That is a directly implied UAF-like condition in IOMMU translation state.

Primitive class is `CLASS_B`: a real lifetime/corruption primitive exists and exploitation leverage requires manual analysis. It is not `CLASS_A` because the supplied evidence does not establish controlled reclaim, attacker-directed DMA read/write, arbitrary kernel write, callback control, or a demonstrated privilege-escalation chain.

Demonstrated consequence is stale hardware translation state and possible walk of freed memory. Practical consequence is at least host DoS or instability. Speculative consequence is stronger DMA isolation failure or memory corruption if stale translation state can be shaped or abused, but this is not proven from the supplied patch alone.

`ImportantCandidate=YES` because the primitive involves active IOMMU/PASID translation state and freed memory, which is security-boundary relevant and should not be reduced to ordinary DoS without review. `AutoDowngradeAllowed=NO` because a concrete active translation lifetime violation with unresolved DMA/IOMMU memory-corruption leverage would be lost by automatic closure.

## 5. One-sentence report phrase

Intel VT-d may invalidate the wrong context-cache SID while tearing down DMA aliases, leaving stale IOMMU translation state that can reference a freed PASID directory and may cause host DoS or unresolved DMA translation memory corruption.

## 6. Manual review recommendation

MANUAL CHECK REQUIRED

The patch establishes stale active IOMMU cache state and a freed PASID-directory walk, which is a real Class B lifetime/corruption primitive in DMA translation state even though controlled exploitation is not demonstrated.

## 7. Call-site confidence

Call-site confidence: medium

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

Patch: iommu/vt-d: Flush context cache with correct SID when tearing down aliases [ Upstream
Commit: a803141597d61dc0511f7cd849625e17d042ca8a
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a803141597d61dc0511f7cd849625e17d042ca8a

Changed files:
  drivers/iommu/intel/iommu.c
  drivers/iommu/intel/iommu.h
  drivers/iommu/intel/pasid.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=a803141597d61dc0511f7cd849625e17d042ca8a

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

The original Linux kernel CVE announcement for CVE-2026-90240 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-90240

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: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: 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: 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).