Fault‑around can map ready dropbehind folios without traversing the normal page‑cache lookup that clears dropbehind. The mapping represents a competing cached user, so the kernel retains the folio instead of forcibly unmapping it when writeback completes. For a mapped folio, folio_unmap_invalidate() may invoke unmap_mapping_folio(), which requires the i_mmap_rwsem and can sleep. By retaining mapped folios, the kernel avoids this path when folio_end_dropbehind() runs in a non‑preemptible context. This behavior can trigger a sleeping‑in‑atomic warning, signaling a potential race condition and improper use of synchronization (CWE‑821). The result is kernel instability that may manifest as performance degradation or occasional crashes, but no direct denial‑of‑service trigger is documented.

All Linux kernel builds prior to commit 2897601dffe576fcd87a4259102f41f7fb0cbfc8, which implements the fix, are affected. The vulnerability exists in every distribution shipping such a kernel until the patch is applied.

The EPSS score is less than 1%, and the vulnerability is not listed in CISA KEV, indicating a low probability of exploitation. The issue resides entirely within kernel internals; the description does not document a remote or privileged exploitation path. With a CVSS base score of 5.5, the vulnerability is considered moderate severity. When triggered, the sleeping‑in‑atomic warning is recorded in kernel logs, and repeated occurrences could impair system responsiveness, but no direct denial‑of‑service attack vector is referenced in the description.