The Linux kernel incorrectly calculates the compound page size during VTL0 memory registration via MSHV_ADD_VTL0_MEMORY. The vmemmap_shift value is derived from the number of trailing zero bits in the OR of start_pfn and last_pfn, and the value is not clamped to the system maximum order. When a sufficiently aligned address range produces a shift larger than MAX_FOLIO_ORDER, the memremap_pages function issues a warning and returns -EINVAL. This prevents the requested memory region from being mapped, potentially causing a failure to establish the hypervisor’s virtual‑to‑physical map and disrupting services that depend on that memory. The flaw results in a denial‑of‑service condition rather than remote code execution or privilege escalation. It exhibits an uncontrolled memory allocation flaw, classified as CWE-1285.

Affected software is the Linux kernel itself. All kernel releases prior to the patch that implements commit 646b67d57589 and clamps vmemmap_shift to MAX_FOLIO_ORDER are vulnerable. No specific downstream product line is listed, so all distributions that ship an unpatched kernel are at risk.

The CVSS score is not disclosed and the EPSS score is unavailable, indicating that public exploitation data is currently unknown. The vulnerability requires the ability to invoke MSHV_ADD_VTL0_MEMORY, which typically implies privileged or root access to the kernel, making it a local‑privilege problem. Because the defect leads to a predictable failure rather than a complex attack chain, the overall risk can be considered moderate at present. The lack of a KEV listing and the absence of known public exploits suggest that the likelihood of immediate exploitation is low, but organizations should still anticipate corrective action if custom hypervisor code interacts with this interface.