GPU VulnDB

Database/NVIDIA / GPU stack

NVIDIA GPU driver: out-of-bounds writes in the kernel mode layer reachable by a privileged local user

CVSS 6.7CVE-2026-47509NVIDIA / GPU stack+3 more CVEscurated

Impact

Four out-of-bounds writes in the kernel mode layer of the GPU display driver, which NVIDIA split across CVE-2026-47509, CVE-2026-47529, CVE-2026-47532 and CVE-2026-47537 in bulletin 2026/5861 with the same 6.7 score, the same prerequisites and the same fix. Each lets an already-privileged local user corrupt kernel memory, with code execution, privilege escalation, information disclosure, data tampering and denial of service listed. Because PR:H is required, these are post-compromise depth-in-the-stack bugs rather than tenant escapes: they matter on nodes where root in a container or a privileged device plugin is not meant to equal host kernel control. Two of the four are Linux-only, two affect Windows and Linux; the operator action is identical for all.

Who can reach it

Local attacker who already holds high privileges on the node (root or equivalent, including a privileged container with the GPU devices mapped in). Not reachable by an ordinary unprivileged tenant.

What to do

Update the GPU display driver and guest driver to the fixed branch in NVIDIA bulletin 2026/5861. One driver roll covers all four ids. Drain the node and reboot to unload and replace the kernel modules.

Also covers 3 CVEs

The vendor assigned a separate id to each affected code path. They share this advisory, this score and this fix, so they are one entry here.

CVE-2026-47529CVE-2026-47532CVE-2026-47537

References

Related entries

All NVIDIA / GPU stack entries

This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.