Back to Intelligence

USN-9175-1: Ubuntu Linux Kernel (Oracle) Vulnerabilities — Detection, Patching, and Hardening Guide

SA
Security Arsenal Team
November 19, 2026
9 min read

Canonical has published USN-9175-1, a security update addressing multiple vulnerabilities in the Linux kernel as shipped for Oracle Cloud Infrastructure (OCI) kernels on Ubuntu. The advisory is notable not for a single headline-grabbing bug, but for its breadth: flaws span more than twenty kernel subsystems, from the block layer and XDP (eXpress Data Path) to multiple CPU architectures (x86, ARM64, PowerPC, RISC-V, S390), CXL drivers, DMA engines, the ublk userspace block driver, and the Intel NPU driver.

For defenders, advisories like this are the bread and butter of kernel risk management. Individually, most of these issues are local privilege escalation or denial-of-service class bugs — but in a cloud and multi-tenant context, "local" is a much lower bar than many teams assume. Any attacker with a foothold — a compromised container, a rogue process, a low-privileged service account — can weaponize a local kernel flaw to break out, escalate to root, and own the host. That makes timely kernel patching one of the highest-leverage defensive actions you can take this week.

Technical Analysis

What USN-9175-1 covers

Per the advisory, the update corrects flaws in the following subsystems:

  • Multiple devices driver and Drivers core
  • eXpress Data Path (XDP) — a high-risk attack surface given the rise of eBPF-related privilege escalation primitives
  • Kernel build system
  • Architecture-specific code: ARM64, PowerPC, RISC-V, S390, and x86 — meaning the flaws are cross-platform and affect virtually any Ubuntu-on-OCI deployment
  • Block layer subsystem and ublk userspace block driver — historically a rich source of use-after-free and race conditions
  • Intel NPU Driver and Compute Acceleration Framework — increasingly relevant as AI/ML workloads bring accelerator drivers into the kernel TCB
  • ACPI drivers, Bus devices, Clock framework/drivers, Clocksource drivers
  • Hardware random number generator core and Hardware crypto device drivers
  • CXL (Compute Express Link) drivers
  • Buffer Sharing and Synchronization framework (dma-buf) — implicated in numerous past Linux privilege escalation chains
  • DMA engine subsystem

The advisory does not enumerate individual CVE identifiers in its summary, but the subsystem list tells an experienced analyst a great deal. The concentration of fixes in XDP/eBPF-adjacent code, dma-buf, ublk, and the block layer is significant: these subsystems have produced some of the most reliable local privilege escalation primitives in recent years. Race conditions in buffer-sharing and block I/O paths are particularly attractive to attackers because they can often be won deterministically with techniques like heap grooming and CPU pinning.

Affected systems

  • Ubuntu systems running the Oracle (OCI) kernel variant (linux-oracle). If your fleet includes Ubuntu instances on Oracle Cloud Infrastructure — or on-prem systems using the Oracle-tuned kernel — you are in scope.
  • Architecture coverage (x86, ARM64, PowerPC, RISC-V, S390) means heterogeneous environments, including ARM-based Ampere instances on OCI, are affected.

Exploitation assessment

  • Exploitation requirements: The majority of multi-subsystem kernel fixes in advisories of this type require local access — an unprivileged process on the host. In containerized environments, a container escape or a shared-kernel sandbox escape scenario elevates these from "local" to "remote-adjacent."
  • In-the-wild status: At publication, USN-9175-1 does not indicate confirmed active exploitation or CISA KEV listing. Treat this as pre-emptive remediation — the window between advisory publication and public PoC for kernel LPEs has historically been days to weeks, not months.
  • Urgency: Any workload running untrusted code (multi-tenant Kubernetes, CI runners, shared dev hosts, container platforms) should treat this as high priority. Single-tenant, tightly controlled hosts can follow a standard patch cycle but should not defer past the next maintenance window.

Detection & Response

You cannot write a signature for "the kernel bugs in USN-9175-1" — but you can hunt for the observable behaviors that local kernel exploitation reliably produces: unexpected privilege transitions, suspicious eBPF and kernel module activity, and post-exploitation artifacts like tampered audit subsystems. The rules below target that exploitation behavior class, which is what a mature SOC should be doing for kernel LPEs generally.

YAML
---
title: Unprivileged Process Enabling eBPF or Loading Kernel Object
description: Detects unprivileged or unexpected processes interacting with bpf() tooling, insmod/modprobe, or writing to /sys/kernel — common prerequisites and artifacts of local kernel privilege escalation against subsystems like XDP, ublk, and dma-buf.
id: 9c1f4a72-6b3d-4e58-a2f1-7d8c9b0e1f2a
status: experimental
references:
  - https://ubuntu.com/security/notices/USN-9175-1
  - https://attack.mitre.org/techniques/T1068/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.privilege_escalation
  - attack.t1068
logsource:
  category: process_creation
  product: linux
detection:
  selection_tool:
    Image|endswith:
      - '/insmod'
      - '/modprobe'
      - '/bpftool'
      - '/kmod'
  selection_bpf_syscall:
    CommandLine|contains:
      - 'bpf '
      - 'map create'
      - 'prog load'
  condition: selection_tool or selection_bpf_syscall
falsepositives:
  - Legitimate kernel module management by administrators and configuration management (Ansible, Chef)
  - Container runtimes and Cilium/Calico using eBPF legitimately — tune by parent process and user
level: medium
---
title: Audit Logging Subsystem Tampering on Linux
description: Detects attempts to stop, disable, or flush auditd or to delete audit rules — a frequent post-exploitation step after successful local privilege escalation to root.
id: 2e7b5c91-3a4f-4d62-9c8a-5f6e7d8a9b0c
status: experimental
references:
  - https://ubuntu.com/security/notices/USN-9175-1
  - https://attack.mitre.org/techniques/T1562.001/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.defense_evasion
  - attack.t1562.001
logsource:
  category: process_creation
  product: linux
detection:
  selection:
    CommandLine|contains:
      - 'auditctl -D'
      - 'auditctl -e 0'
      - 'systemctl stop auditd'
      - 'systemctl disable auditd'
      - 'service auditd stop'
  condition: selection
falsepositives:
  - Controlled maintenance windows; correlate with change tickets
level: high
---
title: Unexpected Setuid to Root From Non-Privileged Parent
description: Detects a shell or interpreter spawning as root from a non-root parent process — the hallmark outcome of a successful local kernel privilege escalation.
id: 4d8a2f63-1c5e-4b79-8a3d-6e9f0a1b2c3d
status: experimental
references:
  - https://ubuntu.com/security/notices/USN-9175-1
  - https://attack.mitre.org/techniques/T1068/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.privilege_escalation
  - attack.t1068
logsource:
  category: process_creation
  product: linux
detection:
  selection:
    Image|endswith:
      - '/bash'
      - '/sh'
      - '/dash'
      - '/python'
      - '/python3'
      - '/perl'
  filter_known_root_services:
    ParentImage|endswith:
      - '/sshd'
      - '/sudo'
      - '/su'
      - '/systemd'
      - '/cron'
      - '/crond'
  condition: selection and not filter_known_root_services
falsepositives:
  - Legitimate use of sudo/su — the filter covers common paths; residual hits from web servers, CI agents, or app runtimes spawning root shells are high-fidelity
level: high
KQL — Microsoft Sentinel / Defender
// Hunt: Kernel exploitation behavior chain on Linux hosts ingested via Syslog/CEF
// Looks for module loads / eBPF activity followed by audit tampering within 30 minutes
let moduleActivity = Syslog
| where TimeGenerated > ago(7d)
| where SyslogMessage has_any ("insmod", "modprobe", "bpftool", "auditctl -D", "auditctl -e 0")
| project TimeGenerated, Computer, ProcessName, SyslogMessage;
moduleActivity
| summarize FirstSeen = min(TimeGenerated), LastSeen = max(TimeGenerated), EventCount = count(), SampleCommands = make_set(SyslogMessage, 5) by Computer
| where EventCount >= 2
| order by EventCount desc;
// Companion: identify hosts still running a vulnerable Oracle kernel
// (devices reporting via Defender for Endpoint or custom heartbeat)
DeviceInfo
| where TimeGenerated > ago(1d)
| where OSPlatform has "Linux"
| project DeviceName, OSPlatform, OSVersion
| distinct DeviceName, OSVersion
| order by DeviceName asc;
VQL — Velociraptor
-- Hunt for Linux hosts exposing exploitation-relevant artifacts:
-- recently loaded out-of-tree modules, eBPF programs, and unexpected setuid binaries
SELECT Pid, Name, Exe, Username, CommandLine
FROM pslist()
WHERE CommandLine =~ '(insmod|modprobe|bpftool|auditctl)'
   OR (Username = 'root' AND Name =~ '(bash|sh|dash|python|perl)')

-- Companion artifact: enumerate loaded kernel modules not from the
-- distribution's signed module tree (indicator of rootkit/LKM activity post-LPE)
SELECT FullPath, Mtime, Size
FROM glob(globs='/sys/module/*/holders')
ORDER BY Mtime DESC

Remediation

1. Patch — this is the fix

Kernel vulnerabilities of this class have no workaround short of patching. Apply the updated Oracle kernel packages and reboot:

Bash / Shell
# Identify your running kernel and confirm it is the Oracle variant
uname -r
apt list --installed 2>/dev/null | grep linux-image

# Apply the USN-9175-1 update
sudo apt-get update
sudo apt-get install --only-upgrade linux-image-oracle linux-headers-oracle
# Or run a full upgrade to catch dependent userspace fixes
sudo apt-get dist-upgrade -y

# Verify the new kernel version against the advisory's fixed-version list:
# https://ubuntu.com/security/notices/USN-9175-1

# Reboot is REQUIRED — a kernel patch does nothing until the new kernel is running
sudo reboot

# Post-reboot verification: confirm the running kernel matches the patched version
uname -r
# Confirm no pending reboot requirement remains
[ -f /var/run/reboot-required ] && cat /var/run/reboot-required.pkgs || echo "No reboot pending"

For large fleets, orchestrate this with your configuration management platform and use Canonical Livepatch to bridge the gap where an immediate reboot is operationally impossible — but treat Livepatch as a risk-reduction bridge, not a substitute for the reboot, since not all kernel fixes are livepatchable.

2. Reduce the attack surface now (defense-in-depth)

These hardening steps materially raise the bar for the exact subsystem classes fixed in this advisory:

Bash / Shell
# Restrict unprivileged eBPF — directly relevant to the XDP fixes
sudo sysctl -w kernel.unprivileged_bpf_disabled=1
echo 'kernel.unprivileged_bpf_disabled=1' | sudo tee /etc/sysctl.d/99-harden-bpf.conf

# Hide kernel pointers from unprivileged users (slows heap-grooming for races)
sudo sysctl -w kernel.kptr_restrict=2
echo 'kernel.kptr_restrict=2' | sudo tee -a /etc/sysctl.d/99-harden-kernel.conf

# Restrict dmesg access
sudo sysctl -w kernel.dmesg_restrict=1
echo 'kernel.dmesg_restrict=1' | sudo tee -a /etc/sysctl.d/99-harden-kernel.conf

# If ublk is not in use, block the module
echo 'blacklist ublk_drv' | sudo tee /etc/modprobe.d/blacklist-ublk.conf

# Apply and persist
sudo sysctl --system

3. Prioritize by exposure

EnvironmentPriorityRationale
Multi-tenant Kubernetes / container hosts on OCIP1 — within 48 hoursContainer escape + kernel LPE = full host and potentially cluster compromise
CI/CD runners executing untrusted buildsP1Arbitrary code execution by design
Shared dev/jump hostsP2 — within 1 weekMultiple local users
Single-purpose hardened appliancesP3 — next maintenance windowLower local-access surface

4. Verify coverage and close the loop

  • Query your asset inventory for all hosts running linux-oracle kernels and reconcile against patch completion — the KQL above includes a fleet-view starter query.
  • Subscribe to the Ubuntu Security Notices feed and route kernel USNs to your vulnerability management workflow with SLA tracking.
  • Review the upstream subsystem commits referenced from the USN page as individual CVE records populate; if any are later added to CISA KEV, re-baseline your remediation SLA to CISA's due dates.

Bottom Line

USN-9175-1 is a reminder that kernel patch hygiene remains one of the most consequential — and most frequently deferred — controls in Linux defense. The subsystem coverage here (XDP/eBPF, dma-buf, ublk, block layer) maps directly onto the primitives attackers prefer for local privilege escalation. Patch the Oracle kernels, reboot, disable unprivileged eBPF where you can, and hunt for the behavioral chain: module/bpf activity → privilege transition → audit tampering. That chain, not any single CVE number, is what will catch the exploitation attempt your patch window missed.

Related Resources

Security Arsenal Penetration Testing Services AlertMonitor Platform Book a SOC Assessment vulnerability-management Intel Hub

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.