The Linux kernel memory management code incorrectly applies an alignment mask to hugetlbfs mappings on AMD F15h processors. By adding the generic va_align bits when a huge‑page memory mapping is requested, the resulting virtual address can fall outside the required huge‑page boundary. When the kernel later calls __unmap_hugepage_range() to tear the mapping down, the address check triggers a BUG_ON, causing the kernel to panic. The outcome is a system crash that results in a denial of service. This weakness is consistent with improper initialization or incorrect alignment handling (CWE‑665).

The flaw affects the Linux kernel on machines that use AMD Carrizo or FX‑8370E CPUs (both with the F15h microarchitecture). Any user who can create or access hugetlbfs file mappings on such hardware may be able to trigger the bug. The affected code path resides in the x86/mm subsystem and applies to all kernel versions released before the patch commit referenced in the advisory.

No CVSS score is published and EPSS data is not available, but the vulnerability can be exploited locally by mapping hugetlbfs pages. Because it relies on a privileged kernel data structure, a non‑privileged user needs elevated rights or an exploit that can access huge pages; therefore the availability of the attack depends on user permissions. The bug is not listed in the CISA KEV catalog, indicating no widespread exploitation evidence. Nevertheless, the potential for a kernel panic means affected systems should be updated promptly, and users should avoid creating hugetlbfs mappings on the affected CPUs until a patched kernel is released.