CVE-2026-47589

A flaw was found in the NVIDIA GPU display driver. A local user with low privileges can trigger a use-after-free condition, where memory is improperly accessed after being released, by issuing a specific sequence of driver commands. Successful exploitation could allow the attacker to execute arbitrary code, escalate privileges, cause a denial of service (DoS), or disclose sensitive information.

This vulnerability is rated as an Important impact for Red Hat Enterprise Linux because an unprivileged local attacker could gain elevated kernel rights or execute arbitrary code on the underlying host. In standard Red Hat Enterprise Linux environments, proprietary NVIDIA driver packages are not deployed by default and are only utilized on hosts equipped with dedicated GPU hardware. Exposure is restricted to multi-user systems where untrusted accounts possess access to the driver's device nodes.

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.

On systems where GPU acceleration is not required, the driver kernel module can be prevented from loading.

1. Create a configuration file to blacklist 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 driver module is currently loaded, a system reboot is required for the configuration to take effect.

Caveat: Disabling the driver disables GPU hardware acceleration, graphical output managed by the GPU, and compute frameworks such as CUDA. Alternatively, restricting access permissions on /dev/nvidia* device nodes to trusted administrative accounts prevents unprivileged users from interacting with the driver.