CVE-2026-47503 in GPU Display Driver
Summary
by MITRE • 09/30/2026
NVIDIA GPU Display Driver for Linux contains a vulnerability in the Virtual GPU Manager (vGPU plugin), where a guest VM user may cause an out-of-bounds write by sending a crafted RPC message with invalid performance state list size parameters. A successful exploit of this vulnerability might lead to code execution, escalation of privileges, data tampering, denial of service, and information disclosure.
If you want to get the best quality for vulnerability data then you always have to consider VulDB.
Analysis
by VulDB Data Team • 09/30/2026
The NVIDIA GPU Display Driver for Linux contains a critical security flaw within its Virtual GPU Manager component, specifically affecting the vGPU plugin architecture used in virtualized environments. This vulnerability stems from an improper validation mechanism when processing Remote Procedure Call messages sent by guest virtual machines to the host driver. The core technical issue is an out-of-bounds write condition triggered when a malicious or compromised guest VM user submits a crafted RPC message containing invalid performance state list size parameters. In this context, the vGPU plugin fails to adequately verify that the specified size of the performance state list falls within acceptable bounds before attempting to process it. This lack of rigorous boundary checking allows an attacker with access to the virtual machine interface to write data beyond the allocated memory buffers associated with handling these performance states.
The operational impact of this vulnerability is severe due to its potential for arbitrary code execution and privilege escalation. Because the out-of-bounds write occurs within the kernel-space driver components that manage GPU resources, a successful exploit can allow an attacker to overwrite critical control structures or return addresses in memory. This capability effectively bypasses standard user-level isolation boundaries between guest VMs and the host system. Consequently, an adversary could achieve full code execution on the underlying Linux host with elevated privileges, potentially gaining root access. Beyond direct exploitation for remote code execution, this flaw also facilitates data tampering by allowing unauthorized modification of sensitive memory regions, denial of service through system crashes or kernel panics resulting from corrupted memory states, and information disclosure via leakage of confidential data stored in adjacent memory locations.
From a classification perspective, this vulnerability aligns with CWE-787: Out-of-bounds Write, which describes the act of writing to a memory location using an index that is outside the bounds of the intended buffer. Furthermore, it maps closely to MITRE ATT&CK technique T1055.009: Process Injection via Memory Allocation and Writing, as well as privilege escalation vectors such as T1068 Exploitation for Privilege Escalation within virtualized environments. The attack vector is classified as Local (T1072) because it requires the attacker to have access to a guest VM that shares the physical GPU resources managed by the vulnerable driver, making it an insider threat or lateral movement scenario rather than a remote network exploit.
Mitigation strategies must focus on immediate patching and architectural hardening. NVIDIA has released updated versions of the GPU Display Driver for Linux that include fixes for this validation logic in the vGPU plugin. Administrators should prioritize applying these security updates to all systems utilizing virtualized GPU resources, particularly those hosting multi-tenant workloads where guest VMs are not fully trusted. In addition to patching, organizations should enforce strict least-privilege principles regarding access to hypervisor management interfaces and ensure that only authorized users can create or modify vGPU configurations. Network segmentation between untrusted guest environments and critical host infrastructure further reduces the attack surface by limiting lateral movement opportunities if a single VM is compromised. Continuous monitoring for anomalous RPC traffic patterns involving GPU performance state requests can also aid in early detection of exploitation attempts before significant damage occurs.