CVE-2026-47560

A flaw was found in the NVIDIA GPU Display Driver for Linux. An unprivileged local user can trigger a use-after-free condition in the kernel mode layer, allowing memory to be accessed after it has been released. A successful exploit could permit an attacker to achieve arbitrary code execution, escalate privileges to gain full control of the system, cause a denial of service, or access and modify sensitive information.

This vulnerability is rated as Important severity because an authenticated local attacker can leverage kernel-mode memory corruption to elevate privileges to root or execute arbitrary code with kernel privileges. The issue specifically impacts Red Hat Enterprise Linux hosts and containers configured with NVIDIA hardware and the proprietary display driver stack. While remote exploitation is not possible, systems providing multi-tenant access or shell accounts to untrusted users face elevated risk whenever access to GPU device interfaces is available.

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 by configuring modprobe rules:

1. Create a configuration file to blacklist and disable loading of the driver:

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

2. If the module is currently active, reboot the system to unload it:

systemctl reboot

Caveats: Disabling the NVIDIA kernel driver will prevent hardware-accelerated rendering, GPU-driven display output, and CUDA/compute workloads from functioning on the host.

Warning: Modifying kernel module autoloading configurations requires a full system reboot to unload modules currently resident in kernel memory, which will interrupt running services.