CVE-2026-47511

A flaw was found in the NVIDIA GPU display driver. An out-of-bounds write vulnerability in the kernel mode layer allows a local user with basic privileges to write data beyond allocated memory boundaries. Successful exploitation could allow an attacker to achieve arbitrary code execution, escalate privileges, access or tamper with sensitive information, or cause a Denial of Service (DoS).

This vulnerability is rated as Important rather than Moderate because an authenticated local user can escalate to kernel-level privileges and achieve complete host compromise, despite the attack vector being restricted to local access. On Red Hat Enterprise Linux systems, exposure is limited to installations where proprietary NVIDIA GPU display drivers are explicitly installed and active with compatible hardware. Default installations that do not utilize NVIDIA hardware or that rely on standard open-source drivers (such as Nouveau) do not load the affected kernel modules and remain unimpacted.

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.

Prevent the NVIDIA driver module from loading if GPU acceleration is not actively needed, or restrict access to the /dev/nvidia* device nodes to trusted administrative users.

To prevent the kernel module from loading automatically:

echo "blacklist nvidia" > /etc/modprobe.d/blacklist-nvidia.conf

If the driver is already loaded, unload it or restart the host:

modprobe -r nvidia

Warning: Blacklisting or unloading the module disables GPU hardware acceleration, display output on attached adapters, and compute frameworks such as CUDA. If the driver is currently in use by active processes or the display manager, unloading will fail and a system reboot will be required for the change to take effect.