In the Linux kernel, the dm_cache device’s smq policy lacked a necessary lock when invalidating cache blocks while operating in passthrough mode. The description shows that multiple workers can invoke the invalidate_mapping operation concurrently, which can corrupt the allocated blocks counter or trigger a use‑after‑free of internal data structures. These race conditions can cause a kernel panic or, if successfully exploited, permit an attacker running with privileged code to achieve arbitrary kernel execution. The impact is confined to the kernel memory management of the dm_cache subsystem and is not a straightforward user‑space vulnerability.

All Linux kernel installations that employ the dm_cache device with the smq policy are potentially affected. The advisory does not specify vulnerable kernel versions, so any distribution whose kernel has not incorporated the missing lock fix could be at risk. Systems that enable passthrough mode on dm_cache devices for direct I/O are the only ones that expose the race condition, as the fault occurs only in that mode.

The CVSS score of 7.8 indicates a high severity. The EPSS score of < 1% suggests a very low but non‑zero probability of exploitation in the wild. Because the vulnerability requires a precise race condition—multiple concurrent writes on a passthrough dm_cache device—the practical exploitation barrier is high. No public exploit is known and the vulnerability is not listed in CISA KEV, indicating limited real‑world usage. Nonetheless, the catastrophic kernel instability and potential for privilege escalation make it a risk that warrants immediate attention.