A flaw in the Linux kernel DRM driver for Xe GPUs allows non-fault-mode scratch virtual machines to trigger page faults that are reported as successful without being resolved. When such a fault occurs on a valid virtual memory area, the handler fails to recover the fault, causing the GPU to retry the operation indefinitely. This results in the GPU being stuck in a retry loop and can lead to a system or application hang, representing a denial‑of‑service vulnerability. The weakness is a logic error in the ASID lookup that incorrectly accepts non‑fault‑mode VMs.

The vulnerability affects the generic Linux kernel (Linux:Linux) on systems that use the DRM/xe GPU driver. No specific kernel version is listed, so any installation that includes the affected driver and has been built before the patch that adds the ASID check will be vulnerable.

The EPSS score is below 1%, indicating a very low exploitation probability, and the vulnerability is not listed in CISA's KEV catalog, suggesting no known active exploits. The likely attack vector is local, requiring a privileged or kernel module that can create or manipulate scratch VMs. An attacker would need to trigger the fault condition on a page backed by a scratch VM to force the GPU into its infinite retry loop.