* [CVE-2026-93109][MODERATE REGULAR] RDMA/mlx5: Drain RCU callbacks during module teardown [ Upstream
@ 2026-09-18 10:27 21% AL-KERNEL
0 siblings, 0 replies; 1+ results
From: AL-KERNEL @ 2026-09-18 10:27 UTC (permalink / raw)
To: kernel-cve
CVE: CVE-2026-93109
Priority: MODERATE REGULAR
AL-KERNEL base severity: MODERATE
KPANIC flag: NO
Patch: RDMA/mlx5: Drain RCU callbacks during module teardown [ Upstream
Commit: ca428e597f12b278bf158356e34641e6386cfa82
Upstream patch: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=ca428e597f12b278bf158356e34641e6386cfa82
Original CVE announcement: https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-93109
Analysis date: Fri, 18 Sep 2026 06:27:54 -0400
ActionableScore: 3
ActionableScore lower bound: 2
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-2026-93109 is available here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-93109
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-2026-93109 MODERATE https://www.kernelcve.org/list/?q=CVE-2026-93109 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-416;CWE-362;*CWE-664;Other CVSS 'AV:L/AC:H/PR:H/UI:N/S:U/C:N/I:N/A:H';BEST CVSS score: '5.8';DESCR 'A lifetime race in RDMA mlx5 can leave devx_free_subscription queued as an RCU callback after the last DevX event file drops its module reference or after auxiliary devices are detached. If mlx5_ib teardown continues before the queued callback runs, later RCU execution may touch code or data that has already been unloaded or cleaned up, which is a use-after-free candidate and can crash the kernel. For the CVSS the PR:L is used in the paranoid score because a local user with access to mlx5 DevX or RDMA user verbs may be able to create and drop event file state while teardown is triggered by module unload, driver detach, hotplug, or service management. The base score still uses PR:H because reliable standalone triggering usually requires administrative control over module unload or device teardown. The issue is not directly network reachable and is not triggered by normal remote RDMA traffic alone. Impact is at least local denial of service via kernel crash and in worst case may allow limited confidentiality or integrity impact due to RCU lifetime memory corruption, so it should be reviewed manually.';YES REQUIRES MANUAL CHECK; ,and ActionableScore result is Borderline / manual review recommended (with actual score 3) YES SIMPLEFIX NETWORK HARDWARE LINUS DECREASED_TO_MODERATEREG_BASED_ON_ACTIONABLESCORELESSTHAN4 NO NO checked
======================================================================
ACTIONABLESCORE ANALYSIS
======================================================================
ActionableScore=3
ActionableScoreLower=2
PrimitiveClass=CLASS_B
RealUAF=YES
HostAdminRequired=YES
NamespaceScopedPrivilege=NO
ImportantCandidate=NO
ManualReview=RECOMMENDED
AutoDowngradeAllowed=YES
1. ActionableScore
* Raw conservative score: 2
* Conservative score: 2
* Raw paranoid score: 3
* Paranoid score: 3
* Final recommended bucket: **Borderline / manual review recommended**::
No floor was applied. Real UAF floor 4 is not applied because host-admin-equivalent teardown is positively established for the complete reliable trigger. No cap was applied.
2. Signal breakdown
Conservative positive signals:
* +2 Real UAF / RCU lifetime violation. The commit directly states that devx_free_subscription can remain queued as an RCU callback while mlx5_ib unloads before the callback runs.
* +1 Reliable kernel crash / strong DoS. A queued callback executing after module teardown can touch unloaded module code or cleaned-up module state and plausibly crash the kernel.
* +1 RCU teardown synchronization defect. The fix adds rcu_barrier() in both registration error unwind and module cleanup, showing that teardown previously failed to wait for outstanding RCU callbacks before destroying later module state.
Conservative negative signals:
* -2 Host-admin-equivalent trigger. The complete reliable trigger is module teardown or registration error unwind, which is tied to module unload/load, auxiliary driver unregister, or device detach workflows.
Paranoid difference:
* +1 Paranoid unresolved RCU lifetime risk. A local RDMA or DevX user may influence event-file state and queue the callback before an admin-driven unload or detach, so the callback lifetime issue is worth manual triage even though the complete trigger still appears host-admin dependent.
Non-awarded signals:
* Controlled reclaim not demonstrated.
* Arbitrary write not demonstrated.
* Confidentiality and integrity not awarded because no disclosure, controlled write, callback hijack from attacker-shaped memory, or privilege escalation primitive is established.
* Remote reachability not established. Normal remote RDMA traffic alone is not shown to trigger the teardown sequence.
* Structured stale-object leverage not awarded because the demonstrated stale use is an RCU callback during teardown, not attacker-shaped active object reuse.
* Concrete repeated release/refcount signal not awarded. The patch does not show first release may destroy an object followed by a second lifetime/refcount mutation on the exact same dangling reference.
* Extra Availability not awarded because the same crash is already counted once.
Repeated release/refcount checklist:
* First release can end lifetime: not established for the exact same reference.
* Exact same reference survives to a second release: not established.
* Second operation is a real lifetime/refcount mutation: not established.
* +1 Concrete repeated release/refcount awarded: no.
* Paranoid repeated-refcount leverage +1 awarded: no.
* FLOOR 5 applied: no.
3. Reachability analysis
The reliable complete trigger appears to require host-admin-equivalent control over mlx5_ib module teardown, driver unregister, auxiliary device detach, or registration error unwind. A local user with RDMA DevX access may plausibly create and close event-file state that queues devx_free_subscription, but the supplied evidence does not show that the same below-host-admin actor can force the full module unload or auxiliary detach sequence.
Namespace-scoped privilege is not established. RDMA DevX access may be delegated in some deployments, but the complete teardown path is not shown to be namespace-scoped or attacker-created. Prior host setup exists in practice because mlx5 hardware and the mlx5_ib module must be present, but this is not by itself a scoring penalty beyond the proven admin teardown requirement. Special hardware is involved, but mlx5 RDMA hardware is not treated as rare physical-only hardware for scoring.
The race is an RCU teardown ordering issue. I do not subtract a race penalty merely because RCU timing is involved. However, the exploitable window depends on callback queuing before unload and teardown continuing before callback execution, so it remains a constrained teardown scenario rather than a broadly repeatable unprivileged data-plane trigger.
4. Primitive and severity interpretation
The exact primitive is an RCU lifetime violation: an RCU callback associated with DevX subscription cleanup can remain pending after the mlx5_ib module proceeds through teardown. This is a real UAF or post-lifetime execution/use candidate because the callback may execute after code or data it depends on has been unloaded or cleaned up.
This is an active module lifetime bug, but the stale use is tied to teardown/unwind rather than a general attacker-controlled active-object reuse path. The demonstrated consequence is kernel crash or host DoS. The speculative consequence is broader memory corruption if callback execution touches freed or repurposed module state, but the patch does not establish controlled reclaim, attacker-shaped replacement contents, arbitrary read/write, function-pointer control, or LPE.
PrimitiveClass is CLASS_B because a real lifetime/RCU violation is established. ImportantCandidate is NO because the available evidence shows a constrained teardown UAF with host-admin-dependent complete triggering and no meaningful independent exploitation leverage beyond crash. AutoDowngradeAllowed is YES because this should not be treated as an Important-preserving primitive on the supplied evidence, although manual review is still recommended due to real RCU lifetime misuse.
Likely operational severity is Borderline to Actionable Moderate only in environments where untrusted users have RDMA DevX access and module/device teardown can be induced by service workflows. In ordinary deployments it is primarily a host-admin-triggered teardown crash risk.
5. One-sentence report phrase
An RCU lifetime bug in RDMA mlx5 can leave devx_free_subscription queued past mlx5_ib module teardown, allowing a later callback to touch unloaded or cleaned-up module state and crash the kernel under module unload or driver detach conditions.
6. Manual review recommendation
MANUAL CHECK RECOMMENDED
Reason: real RCU post-lifetime callback behavior is established, but the complete reliable trigger appears host-admin dependent and no controlled reclaim, repeated refcount mutation, arbitrary write, disclosure, or privilege escalation primitive is shown.
7. Call-site confidence
Call-site confidence: medium
======================================================================
UPSTREAM PATCH SUMMARY
======================================================================
Patch: RDMA/mlx5: Drain RCU callbacks during module teardown [ Upstream
Commit: ca428e597f12b278bf158356e34641e6386cfa82
Upstream URL: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=ca428e597f12b278bf158356e34641e6386cfa82
Changed files:
drivers/infiniband/hw/mlx5/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=ca428e597f12b278bf158356e34641e6386cfa82
======================================================================
DETAILED REPORT METHODOLOGY
======================================================================
The original Linux kernel CVE announcement for CVE-2026-93109 can be found here:
https://lore.kernel.org/linux-cve-announce/?q=CVE-2026-93109
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: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: 2
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 [email protected] (and both
send reply to CVE record itself too and see "reply" button below for howto reply).
^ permalink raw reply [relevance 21%]
Results 1-1 of 1 | reverse | sort options + mbox downloads above
-- links below jump to the message on this page --
2026-09-18 10:27 21% [CVE-2026-93109][MODERATE REGULAR] RDMA/mlx5: Drain RCU callbacks during module teardown [ Upstream AL-KERNEL
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox