Cloudanix VulnDB

CVE-2023-26083

Low · 3.3 CISA KEV

CVSS

3.3

EPSS

1.2%

Published

Apr 6, 2023

CISA due date

Apr 28, 2023

Cloudanix analysis

Act now SSVC: Attend

Some devices with affected Arm graphics drivers can let a malicious app read hidden system details that make a larger device takeover easier.

Active In The Wild Chainable
Exploitation
Local Privilege
Attacker needs
Low Privilege
Complexity
Low
Impact
Data Exposure
Component
OS kernel
Class
Information Disclosure
Remediation
Upgrade
Effort
Medium
Patch risk
Low

What it actually does

CVE-2023-26083 is a local information disclosure vulnerability in Arm Mali GPU kernel drivers. A non-privileged local user or application can submit otherwise valid GPU processing operations that trigger a memory leak and expose sensitive kernel metadata. The flaw does not directly provide code execution, privilege escalation, integrity impact, or denial of service by itself, but leaked kernel information can weaken kernel address space layout randomization or disclose metadata useful for chaining with other local kernel or sandbox-escape vulnerabilities.

Why it matters to the business

Systems using affected Arm Mali GPU drivers may allow a malicious local application to extract kernel metadata that helps bypass platform defenses. On mobile, embedded, Chromebook, Android, and other Arm-based devices, this can increase the likelihood that a seemingly low-privileged app becomes part of a larger compromise chain. The vulnerability is listed in CISA KEV with a remediation due date of 2023-04-28 and has been exploited in the wild, so organizations managing affected device fleets should treat it as a priority despite the low standalone CVSS score. There is no indication from KEV metadata that it is specifically associated with ransomware campaigns.

How an attack unfolds

An attacker first needs the ability to run code locally on a device with an affected Mali GPU kernel driver, commonly through a malicious application, compromised app process, browser or document exploit chain, or other local execution path. The attacker then opens the GPU driver interface and submits valid GPU workloads crafted to trigger the memory leak condition. Instead of receiving only intended GPU processing results, the attacker obtains sensitive kernel metadata leaked from driver or kernel memory. The attacker can use this information to defeat memory-layout randomization, identify kernel objects, or improve reliability of a second-stage exploit. In observed real-world exploitation patterns for mobile and embedded platforms, information disclosure bugs like this are often paired with separate memory corruption or privilege escalation flaws to escape a sandbox or gain elevated privileges.

How to fix it

Apply vendor-supplied security updates for the affected device platform, firmware, operating system image, or GPU driver package. The affected ranges include Midgard GPU Kernel Driver r6p0 through r32p0, Bifrost GPU Kernel Driver r0p0 through r42p0, Valhall GPU Kernel Driver r19p0 through r42p0, and Avalon GPU Kernel Driver r41p0 through r42p0. Fixed versions are not provided in the input, so use the relevant OEM, OS vendor, SoC vendor, or Arm security bulletin guidance to identify patched driver builds newer than the vulnerable ranges. For managed mobile and embedded fleets, prioritize OS/firmware updates that include Mali GPU driver fixes, block or quarantine devices that cannot receive updates where feasible, and restrict installation of untrusted applications. Mitigations such as application allowlisting, mobile threat defense, browser/app sandbox hardening, and reducing local code execution opportunities can reduce risk but should not be considered complete fixes.

How to confirm the fix worked

Confirm affected devices no longer run Mali GPU kernel driver versions in the vulnerable ranges. Validate through vendor patch level, firmware build identifiers, kernel module package metadata, Android security patch level where applicable, OEM advisory mapping, or direct driver version inspection if supported. For Linux-based systems, review loaded GPU kernel modules and package/firmware versions; for Android and embedded devices, verify the OEM security bulletin or build release notes explicitly include the Arm Mali GPU kernel driver fix. Where test tooling is available from the platform vendor, run a regression or proof-of-vulnerability check only in an authorized lab to confirm that crafted GPU operations no longer leak kernel metadata. Also verify fleet compliance reporting shows all affected device models have received the relevant security update or have been removed from production use.

Typical attack path

  1. Gain local code execution on a device with an affected Arm Mali GPU kernel driver, such as through a malicious app or exploit chain.
  2. Access the Mali GPU kernel driver interface from a non-privileged context.
  3. Submit valid but specially crafted GPU processing operations that trigger the memory leak.
  4. Collect leaked kernel metadata returned through the GPU operation path.
  5. Use the leaked metadata to bypass kernel hardening or improve reliability of a chained local privilege escalation or sandbox escape.
  6. Optionally deploy follow-on tooling, persistence, data collection, or surveillance capabilities after the broader exploit chain succeeds.

How to detect it

This is a local kernel driver information disclosure vulnerability, so exploitation typically leaves limited direct evidence and may not create distinctive network indicators. Focus detection on endpoint, mobile, and system telemetry showing suspicious local access to Mali GPU device nodes or kernel driver interfaces by unusual apps or processes, especially shortly before crashes, privilege changes, sandbox escapes, or installation of additional payloads. Useful log sources include Android logcat, kernel ring buffer logs, auditd or eBPF telemetry on Linux, mobile threat defense events, EDR process and file telemetry, app installation history, package manager logs, SELinux denial logs, and crash reports. Hunt for untrusted or newly installed applications repeatedly opening Mali device files, abnormal GPU ioctl activity, unexpected access to /dev/mali*, kernel warnings from Mali modules, or exploit-chain behavior such as a low-privileged app spawning privileged processes or writing to sensitive locations. Because the issue has been exploited in the wild and appears in CISA KEV, treat suspicious local exploitation artifacts around affected devices as high priority even if there is no standalone CVE-specific IOC.

  • /dev/mali
  • /dev/mali0
  • /dev/mali0 opened by untrusted application
  • mali_kbase
  • mali_kbase:
  • kbase
  • kbase_ioctl
  • GPU fault
  • Mali GPU Kernel Driver
  • avc: denied
  • ioctl

What your security team should do

  • Prioritize remediation for internet-managed, high-risk, mobile, embedded, or user-accessible devices because this vulnerability is in CISA KEV and has documented exploitation in the wild.
  • Inventory devices and OS images that use Arm Mali Midgard, Bifrost, Valhall, or Avalon GPU kernel drivers and map them to OEM security bulletins or firmware releases.
  • Apply available OS, firmware, or GPU driver updates; where fixed versions are unclear, upgrade to the latest vendor-supported build that postdates the relevant advisory.
  • Restrict installation and execution of untrusted applications, especially on Android, Chromebook, kiosk, and embedded platforms using affected Mali GPUs.
  • Hunt for signs of malicious local application activity and chained exploitation, rather than expecting a single reliable network IOC for this local information disclosure flaw.
  • Place unsupported or unpatchable devices behind compensating controls, reduce their privilege and data exposure, or plan accelerated replacement.
  • Track CISA KEV status and confirm remediation evidence because the KEV due date of 2023-04-28 has already passed.

The record

Memory leak vulnerability in Mali GPU Kernel Driver in Midgard GPU Kernel Driver all versions from r6p0 - r32p0, Bifrost GPU Kernel Driver all versions from r0p0 - r42p0, Valhall GPU Kernel Driver all versions from r19p0 - r42p0, and Avalon GPU Kernel Driver all versions from r41p0 - r42p0 allows a non-privileged user to make valid GPU processing operations that expose sensitive kernel metadata.

Affected packages

Ecosystem Package Introduced Fixed
Android :unknown: SoCVersion:0 SoCVersion:2023-07-05

Related vulnerabilities