CVE-2026-47569

A flaw was found in the NVIDIA GPU display driver. A local user can trigger a race condition during handle recycling in the kernel mode layer, resulting in a type confusion condition where memory of one type is misinterpreted as another. A successful exploit could allow an attacker to achieve privilege escalation, execute arbitrary code, access sensitive information, or cause a denial of service (DoS).

This vulnerability is rated as an Important severity issue because an authenticated local user can exploit kernel-level flaws to gain elevated root privileges or compromise system availability, rather than being remotely reachable. The affected proprietary driver modules are not enabled or installed by default in Red Hat Enterprise Linux, limiting exposure strictly to environments where NVIDIA GPU drivers and specialized hardware have been explicitly deployed for compute or graphical workloads. Multi-tenant systems and workstation deployments that permit unprivileged local user access to NVIDIA device nodes face the greatest operational risk.

The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.

To mitigate this vulnerability, prevent the NVIDIA kernel module from loading on systems where GPU acceleration is not strictly required:

1. Create a modprobe configuration file to disable the module:

echo "install nvidia /bin/true" > /etc/modprobe.d/disable-nvidia.conf

2. If the driver is currently loaded, unload it or reboot the system:

reboot

Alternatively, if the GPU is required for workload execution, restrict file permissions on /dev/nvidia* device nodes to authorized service accounts or groups only, preventing untrusted local users from interacting directly with the kernel interface.

Warning: Disabling the kernel module requires a system reboot and will disable hardware-accelerated graphics, CUDA, and GPU compute capabilities.