Skip to main content

Linux Kernel CVE-2026-98152

| EUVDEUVD-2026-86983 MEDIUM
Memory Leak (CWE-401)
2026-09-25 Linux GHSA-vg6q-7wg8-hxrg
5.5
CVSS 3.1 · NVD
View

Attack ChainAIDerived

Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain

  1. Access
    Reach NVMe-oF RDMA target listener
  2. Delivery
    Flood repeated connect/disconnect attempts
  3. Exploit
    Exceed pending-disconnect backlog limit
  4. Execution
    Leak queue and IB resources per attempt
  5. Impact
    Exhaust kernel/IB memory and degrade availability

Vulnerability Assessment

Weakness
CWE
ASVS 5.0
V15 - Secure Coding and Architecture Resource cleanup, leak prevention
Exploitation Conditions

Exploitable ONLY on hosts configured as an NVMe-over-Fabrics RDMA target (the nvmet-rdma driver loaded with an active RDMA listener/port); systems without NVMe-oF RDMA target configured are not affected. The specific trigger is driving pending disconnecting queues past the connect backlog limit, i.e. rapid repeated connect/disconnect churn from one or more initiators. Limiting factors: requires RDMA fabric reachability to the target, the leak is incremental (slow resource exhaustion rather than instant crash), and impact is confined to availability (A:H) with no confidentiality or integrity effect; no user interaction is required.

Impact

This is a genuine but low-priority availability issue, not a widespread emergency. The signals are consistent and point to minor risk: CVSS 5.5 (MEDIUM) with A:H but C:N/I:N means pure resource exhaustion with no data or integrity impact; EPSS is very low (0.16%, 5th percentile); the CVE is NOT on CISA KEV; and no public exploit has been identified. The one point worth flagging is the CVSS attack vector: NVD/the kernel CNA assigned AV:L (local), yet the flaw is in the nvmet-rdma connect path, which is exercised by a remote RDMA initiator over the storage fabric - so the realistic trigger is network/adjacent, not a local console user. This AV:L vs. fabric-reachable discrepancy should be verified against deployment topology. Even accounting for that, impact is a slow memory/IB-resource leak (CWE-401) that degrades availability over time rather than granting code execution, so this is a patch-in-normal-maintenance item, heavily gated by the requirement that the host run an NVMe-oF RDMA target at all.

Exploit ScenarioDerived

An attacker (or a misbehaving initiator) with reachability to a host acting as an NVMe-over-Fabrics RDMA target repeatedly opens and tears down connections so that pending disconnecting queues exceed the backlog limit. Each rejected connect in this state leaks a freshly allocated queue and its associated InfiniBand resources, and sustained abuse gradually exhausts kernel memory and IB resources, degrading or crashing the storage target. No public exploit code has been identified at time of analysis.

This Month Plan within 30 days. Why: MEDIUM severity (CVSS 4.0-6.9).
Patch available 0.2% exploitation probability (EPSS)

Get an e-mail when this CVE is exploited, patched or re-scored. Free account.

Alert me

AnalysisAIv1 – Oct 02, 2026

A resource leak in the Linux kernel's nvmet-rdma driver - the RDMA transport for NVMe-over-Fabrics targets - causes a newly allocated queue and its associated InfiniBand resources to be leaked whenever incoming disconnecting queues exceed the configured connect backlog limit. The issue is reachable without authentication (CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H per independent assessment) but only on hosts actually configured as NVMe-oF RDMA targets with an active RDMA listener, and only via the adjacent RDMA fabric; an attacker or misbehaving initiator must generate rapid repeated connect/disconnect churn. Impact is limited to availability through slow, incremental resource exhaustion rather than an immediate crash, EPSS is very low (0.16%, 5th percentile), and no public exploit code or confirmed active exploitation (CISA KEV) has been identified at time of analysis.

RemediationAI

Vendor-released patch: upgrade to Linux 6.12.111, 6.18.53 or 7.2.7 (or the 7.3-rc2 code), which contain the upstream fixes referenced at https://git.kernel.org/stable/c/186414a6a1a34e081b07e8873622f90402346235, https://git.kernel.org/stable/c/e48f9d1076f8c62c3969588d638602b94aaeff12, https://git.kernel.org/stable/c/32e598324edc3ebb1ac9362d5b9fc30ce0de4873, https://git.kernel.org/stable/c/287420cde9d6669abcd2878c344db67423eb7df6, https://git.kernel.org/stable/c/4f7cf573cdf0ee857448b9b1967d686b07c71e7d, https://git.kernel.org/stable/c/60d56bf0b14d3c545bacb9aeef92a7e6f2cf0caa and https://git.kernel.org/stable/c/fb1ed67788e21832b614c23767a088c08cfdd2f2, and plan a maintenance window because loading a new kernel on an NVMe-oF target disrupts active fabrics. If patching must be deferred, load the nvmet-rdma module only on hosts that genuinely need NVMe-oF RDMA exports - unloading/blacklisting nvmet-rdma (modprobe -r nvmet_rdma, with the trade-off that all RDMA-backed NVMe namespaces on that host become unavailable) removes the exposure entirely. Additionally, rate-limit or isolate initiators so pending disconnecting queues cannot be driven past the backlog limit: restrict RDMA fabric reachability to trusted initiators via InfiniBand partition keys (pkeys) or RoCE VLAN segmentation, and cap reconnect churn in multipath/failover configurations (fewer aggressive reconnect retries, longer reconnect delays), accepting that aggressive failover convergence will be slower. Monitor the target for creeping kernel memory growth and unbounded nvmet-rdma queue/QP allocation via slab and RDMA resource counters as an early warning that the leak is being triggered; there is no userspace configuration knob that fixes the leak itself, only the kernel patch does.

Affected VersionsENISA

Linux 712f3268a62d0df98a7ee991cba963ced2e58007 <32e598324edc3ebb1ac9362d5b9fc30ce0de4873
Linux a8803c4f0ac3fa7df5551bbb5a8800c434a94357 <4f7cf573cdf0ee857448b9b1967d686b07c71e7d
Linux 00eaa58988d35fd47ca8811f7f72871591f61ffb <e48f9d1076f8c62c3969588d638602b94aaeff12
Linux badc53620fe813b3a9f727ef9526f98567c2c898 <fb1ed67788e21832b614c23767a088c08cfdd2f2
Linux 6.12.96 <6.12.111
Linux 7.2
Linux 6.18.39 <6.18.53
Linux d65fe42820b890a6a4644de0a95a812471f79ad3 <287420cde9d6669abcd2878c344db67423eb7df6
Linux patch: 7.3-rc2
Linux 204b9645536bb4a77a2cb35e6519dbf9f4ea1665 <186414a6a1a34e081b07e8873622f90402346235
Linux badc53620fe813b3a9f727ef9526f98567c2c898 <60d56bf0b14d3c545bacb9aeef92a7e6f2cf0caa
Linux patch: 0
Linux patch: 6.12.111
Linux patch: 6.18.53
Linux patch: 7.2.7
Linux 5828517d17eda27f21d29ea14800c9e0a57bad11
Linux 7.1.4 <7.2

Affected Packages

euvd linux fix: 6.12.111, 6.18.53, 7.2

Source: OSV (Open Source Vulnerabilities)

Affects your stack

Affects 1 package in euvd

Checking this CVE against your own stacks and servers is part of the Basic plan and up.

Technical ContextAI · show details

CWE-401 (Missing Release of Memory after Effective Lifetime) in the Linux kernel's nvmet-rdma module, which implements the NVMe-over-Fabrics target-side RDMA transport. When an RDMA initiator connects, nvmet-rdma allocates an nvmet_rdma_queue plus the underlying InfiniBand resources (queue pairs, completion queues and related memory registrations). If the number of pending disconnecting queues exceeds the driver's connect backlog limit, the connect path takes the error branch and drops only the device reference, leaving the freshly allocated queue and its IB resources unreachable - a classic leak. Because each leaked queue is a small, finite object, the defect does not produce a single deterministic crash but instead degrades the target over time as repeated connect/disconnect cycles accumulate; the fixed code path returns/releases the queue and its IB resources on the backlog-exceeded branch. The upstream fixes are recorded across multiple stable trees via git.kernel.org commits (186414a6a1a34e081b07e8873622f90402346235, e48f9d1076f8c62c3969588d638602b94aaeff12, 32e598324edc3ebb1ac9362d5b9fc30ce0de4873, 287420cde9d6669abcd2878c344db67423eb7df6, 4f7cf573cdf0ee857448b9b1967d686b07c71e7d, 60d56bf0b14d3c545bacb9aeef92a7e6f2cf0caa, fb1ed67788e21832b614c23767a088c08cfdd2f2), and the affected product is uniformly identified by CPE as linux:linux - i.e. this is a kernel subsystem issue, not a userspace nvme-cli or storage-array problem.

Know when the next one hits your stack

  • Watch products and get an email when something new is published
  • Check this CVE against your own stack and servers
  • Exploit and PoC references, threat intelligence from 22 sources

No credit card · 7-day full trial

Severity by source

NVD PRIMARY 5.5 MEDIUM
AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
vuln.today AI 6.5 MEDIUM

Triggered by a remote RDMA initiator on the storage fabric (AV:A, contrasting NVD's AV:L), no auth before the leak, only availability impacted via resource leak.

3.1 AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
4.0 AV:A/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
SUSE 6.5 MEDIUM
AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Red Hat 5.5 MEDIUM

Primary rating from NVD.

CVSS VectorNVD

Attack Vector
Local
Attack Complexity
Low
Privileges Required
Low
User Interaction
None
Scope
Unchanged
Confidentiality
None
Integrity
None
Availability
High

Lifecycle Timeline

4
CVSS changed
Oct 02, 2026 - 15:22 NVD
5.5 (MEDIUM)
Patch available
Sep 25, 2026 - 12:02 EUVD
CVE Published
Sep 25, 2026 - 10:36 nvd
MEDIUM 5.5
CVE Published
Sep 25, 2026 - 10:36 cve.org
UNKNOWN (no severity yet)

DescriptionNVD

In the Linux kernel, the following vulnerability has been resolved:

nvmet-rdma: fix queue leak when connect backlog is exceeded

When pending disconnecting queues exceed the backlog limit, the connect path only drops the device reference and leaks the newly allocated queue and its IB resources.

Threat Intelligence Overview

EPSS Score
0.2%
Top 95% exploitable
Prevalence
Low
nvmet-rdma NVMe-oF RDMA target is a niche storage-server subsystem, rarely enabled
Data Quality
4 sources
Confidence: Medium

How this decision was reached

The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.

SignalValue
Severity MEDIUM
CISA KEV No
Public POC No
EPSS 0.2%
Patch available Yes
Active threat No
Deciding rule
MEDIUM severity (CVSS 4.0-6.9)
What every action level means
Emergency Immediate action required. Criteria: CRITICAL + in CISA KEV (actively exploited) without patch, OR CRITICAL + EPSS > 30% without patch, OR CRITICAL + active threat signals without patch.
Act Now Prioritize within 24-48h. Criteria: in KEV with patch available, OR CRITICAL + public POC, OR HIGH + EPSS > 10%, OR CRITICAL + active threat, OR any CRITICAL severity.
This Week Schedule remediation this week. Criteria: HIGH severity (CVSS 7.0-8.9). Patch within normal change window.
This Month Plan within 30 days. Criteria: MEDIUM severity (CVSS 4.0-6.9), lower exploitation probability. Include in regular patch cycle.
Monitor Track and reassess. Criteria: LOW severity (CVSS 0.1-3.9), minimal exploitation risk. No immediate action needed.

SSVC AssessmentENISA

SSVC Decision· derived
Defer
Exploitation· derived
None
System Exposure· derived
Small
Automatable· derived
No
Human Impact· derived
Medium

Priority Score

28
Low Medium High Critical
KEV: 0
EPSS: 0
CVSS: +28
PoC: 0
Weaponized: 0
Threat intel: 0
Gov advisory: 0

Vendor StatusVendor

SUSE

Severity: Moderate
Product Status
SUSE Linux Enterprise Desktop 15 SP7 Not-Affected
SUSE Linux Enterprise Desktop 15 SP7 Not-Affected
SUSE Linux Enterprise High Availability Extension 15 SP7 Not-Affected
SUSE Linux Enterprise High Availability Extension 15 SP7 Not-Affected
SUSE Linux Enterprise High Availability Extension 16.0 Not-Affected

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy