[CVE-2026-74346][MODERATE REGULAR] RDMA/irdma: Fix OOB read during CQ MR registration [ Upstream

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

CVE: CVE-2026-74346
Priority: MODERATE REGULAR
AL-KERNEL base severity: MODERATE
KPANIC flag: NO
Patch: RDMA/irdma: Fix OOB read during CQ MR registration [ Upstream
Commit: a80b3b13786e9ab1c52b31a1f16c7d6708fa9220
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a80b3b13786e9ab1c52b31a1f16c7d6708fa9220
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-74346
Analysis date: Sun, 16 Aug 2026 09:16:51 -0400
ActionableScore: 4
ActionableScore lower bound: 2
Actionable bucket: Borderline manual review recommended
Manual review required: YES

Summary:
An OOB read in RDMA irdma CQ memory registration can occur when CQ resize state is checked too late, causing `arr[req->cq_pages]` to be read one element past the array and potentially leading to local driver instability or limited DMA metadata misuse.

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

The original Linux kernel CVE announcement for CVE-2026-74346 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-74346

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-2026-74346	MODERATE	CHECK WITH IMPACT FROM ORIG NN LOW	Maybe valid. Check manually. Hints by AL-KERNEL: The best (paranoid) CVSS is 'AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L';CWE-125;CWE-665;CWE-754;Other CVSS 'AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L';BEST CVSS score: '5.3';DESCR 'An OOB read in RDMA irdma CQ memory registration occurs because cqmr split is false during MR registration even when the device supports CQ resize. For GEN_2 hardware and later, req cq_pages can be equal to the allocated arr size, so reading arr(req cq_pages) reads one entry past the array while populating the CQ shadow address. For the CVSS the PR:L is used because practical triggering requires a local process that can access RDMA verbs or uverbs and create CQs on an affected irdma device. The issue is not directly network reachable because the vulnerable path is a local RDMA control path rather than packet processing. Impact is mainly a local driver memory safety issue with possible limited information exposure or instability, and it should be reviewed manually due to the kernel OOB read primitive.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Borderline manual review recommended (with actual score 3)	YES	OOB NETWORK HARDWARE MEMORY  DECREASED_TO_MODERATE_REGULAR_FROM_MODERATE7_BASED_ON_GUESSCVSS_AND_CIANNH	NO	NO	checked

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

ActionableScore=4
ActionableScoreLower=2

## 1. ActionableScore

* Conservative score: 2
* Paranoid score: 4
* Final recommended bucket: **Borderline manual review recommended**::

## 2. Signal breakdown

Conservative signals:

* Local unprivileged or low-privileged trigger: +1. A local process with access to RDMA verbs or uverbs can likely reach CQ creation and CQ MR registration.
* Privileged kernel/device-management DMA-adjacent path: +1. The bug affects RDMA CQ memory registration and populates DMA-related CQ shadow metadata.
* Availability impact realistic: +1. A one-entry OOB read of a DMA address array can plausibly cause driver instability or crash depending on adjacent memory and later use.
* Rare hardware/subsystem exposure: -1. This requires an affected Intel irdma device and GEN_2 or later behavior with CQ resize support.

Paranoid additional signals:

* Weak or indirect corruption candidate: +1. The OOB read is not just a passive stack read. The value is assigned to `cqmr->shadow`, a DMA-address-like field, which may influence later device or CQ setup behavior.
* Integrity impact plausible but limited: +1. If the bogus shadow address is consumed in a later hardware setup path, limited device-memory or DMA metadata corruption is technically possible, but not demonstrated.

Not awarded:

* No remote reachable signal. This is a local RDMA control path, not packet receive processing.
* No strong memory corruption signal. The patch shows OOB read by one, not OOB write, UAF, double-free, type confusion, or arbitrary write.
* No strong LPE signal. There is no shown attacker-controlled reclaim, overwrite, callback control, refcount abuse, or object replacement.
* No confidentiality signal in the conservative score. The patch does not show a user-visible disclosure path for the out-of-bounds value.

## 3. Reachability analysis

The likely trigger is a local process that can open and use the affected RDMA device and create completion queues through RDMA verbs or uverbs. This is PR:L in practical terms when RDMA device nodes are available to non-root users or service accounts, but it may become effectively privileged on systems where RDMA access is restricted by device permissions or policy.

Namespaces and containers may matter if RDMA device access is delegated into a container. Container root or a service account with RDMA device access should not be treated as full host root, but the bug still requires local access to the device API.

The path is not broadly default-exposed on ordinary systems. It is relevant mainly to machines with Intel irdma hardware and RDMA userspace access enabled.

Call-site confidence: medium. The patch includes the faulty read and nearby CQ creation context, but not the full later use of `cqmr->shadow`.

## 4. Severity interpretation

This behaves more like a borderline actionable Moderate than an Important-class issue. The concrete primitive is a one-element OOB read in kernel RDMA driver code during CQ memory registration. Realistic impact is likely local driver instability or DoS.

The theoretical concern is higher because the OOB value is stored into a DMA-address-like field, which makes this more security-relevant than a harmless read-only bounds bug. However, the patch does not demonstrate attacker-controlled data, a write primitive, a disclosure path, or a reliable privilege-escalation chain.

## 5. One-sentence report phrase

An OOB read in RDMA irdma CQ memory registration can occur when CQ resize state is checked too late, causing `arr[req->cq_pages]` to be read one element past the array and potentially leading to local driver instability or limited DMA metadata misuse.

## 6. Manual review recommendation

MANUAL CHECK REQUIRED

Reason: this is a kernel OOB read in an RDMA driver and the read value is stored into DMA-related CQ shadow metadata. It is not a clear Important-class issue from the patch alone, but it should not be auto-closed without confirming whether `cqmr->shadow` can influence later DMA or CQ setup in an attacker-useful way.

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

Patch: RDMA/irdma: Fix OOB read during CQ MR registration [ Upstream
Commit: a80b3b13786e9ab1c52b31a1f16c7d6708fa9220
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=a80b3b13786e9ab1c52b31a1f16c7d6708fa9220

Changed files:
  drivers/infiniband/hw/irdma/verbs.c
  drivers/infiniband/hw/irdma/verbs.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=a80b3b13786e9ab1c52b31a1f16c7d6708fa9220

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

The original Linux kernel CVE announcement for CVE-2026-74346 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-74346

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:L/I:N/A:L
  The Best / paranoid CVSS vector: AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L
  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: 2
  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: 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).