The vulnerability is a race condition in the Linux kernel’s mm/pagewalk code, where a concurrent split of a PUD entry can collide with a refault of the same PUD leaf. This race causes the kernel to walk a non‑existent PMD range and triggers an Oops, resulting in a kernel panic. The bug is not a direct code execution flaw, but it can destabilize the system, forcing restarts and disrupting services. The impact is therefore a local denial of service that could affect any process accessing memory mappings during high‑concurrency scenarios such as DMA set‑up. The weakness corresponds to a race condition and improper resource handling.

All Linux kernel releases that include the unfixed mm/pagewalk code are affected; no specific version range is listed in the advisory, so any kernel prior to the commit that introduced the fix (38ec58670a0c5fc1edabdeccd857e586b7b3f318) is vulnerable. The issue manifests when user space programs (e.g., VFIO‑PCI drivers issuing vfio_pin_pages_remote) interact with memory mappings while the kernel is concurrently splitting page tables.

The advisory does not provide a CVSS or EPSS score, and the vulnerability is not listed in CISA’s KEV catalog. Because the bug requires a race driven by concurrent VMAs and the use of privileged mechanisms (VFIO‑PCI), the likelihood of exploitation is low to moderate and likely limited to local users with the ability to invoke DMA operations or read process memory maps. Nonetheless, a successful race will cause an unhandled kernel fault, leading to a crash and potential data loss or denial of service. The risk is amplified on systems with high‑frequency DMA or NUMA workloads that trigger frequent page table splits.