The vulnerability arises from an 8‑byte structure in the Linux kernel’s IPv4 stack that is accessed on ARM64 platforms with a 64‑bit load that demands natural alignment. When the kernel is compiled with Clang and link‑time optimization, the READ_ONCE macro evaluates the full struct, resulting in a 64‑bit ld_ar instruction that, if placed on a 4‑byte aligned address, triggers a strict Alignment Fault and panics the kernel. This fault does not leak or modify data dishonestly; it merely brings the system down, resulting in a loss of availability. The weakness corresponds to atomic read/write of a non‑aligned structure (CWE‑468).

The issue affects any Linux kernel built for ARM64 architecture that is compiled with Clang and enables LTO, before the patch that moves the READ_ONCE and WRITE_ONCE macros to operate on the individual u32 members. This includes versions such as 6.11 and pre‑release 7.0 RC kernels, as well as any custom builds that have not yet incorporated commit e35123d83ee3. All devices running such kernels—servers, embedded systems, mobile devices—are vulnerable if the ipv4 multipath hash seed sysctl is accessible.

The CVSS score of 5.5 indicates moderate severity, while the EPSS rate is below 1 %, suggesting a low likelihood of real‑world exploitation. The vulnerability is not listed in the CISA KEV catalog, so no active exploitation is documented. Based on the description, it is inferred that an attacker would need privileged operations to read or write the multipath hash seed sysctl, implying a local or elevated privilege attack vector. When triggered, the fault leads to an immediate kernel panic, causing denial of service.