In the Linux kernel’s AMDGPU DRM driver, a rare condition when the device EEPROM reports only invalid address entries leads to skipped data allocation. When the driver subsequently performs a bad page check, it dereferences a null pointer, causing a kernel Oops and crash. This produces a kernel panic, rendering the system unavailable until reboot and therefore constitutes a denial‑of‑service flaw.

The vulnerability resides in the AMDGPU driver that is part of the Linux kernel. Any Linux distribution that provides the amdgpu module and has not yet incorporated the patch is vulnerable. The included backtrace shows the issue on an Ubuntu 6.8.0‑38‑generic kernel; older kernels running the same driver before the fix are similarly at risk.

There is no CVSS score listed, and EPSS data is unavailable, but the impact is full kernel crash. The flaw can be triggered by reading the bad‑pages sysfs file, which is generally world‑readable, so a local user can cause the crash. The exact trigger is uncommon because it requires a GPU with only invalid EEPROM entries, but if such a condition exists it can be exploited for local denial of service. No remote exploitation path is documented.