CVE-2026-74365

Not enough EPSS history yet.

In the Linux kernel, the following vulnerability has been resolved: nvdimm/btt: Handle preemption in BTT lane acquisition BTT lanes serialize access to per-lane metadata and workspace state during BTT I/O. The btt-check unit test reports data mismatches during BTT writes due to a race in lane acquisition that can lead to silent data corruption. The existing lane model uses a spinlock together with a per-CPU recursion count. That recursion model stopped being valid after BTT lanes became preemptible: another task can run on the same CPU, observe a non-zero recursion count, bypass locking, and use the same lane concurrently. BTT lanes are also held across arena_write_bytes() calls. That path reaches nsio_rw_bytes(), which flushes writes with nvdimm_flush(). Some provider flush callbacks can sleep, making a spinlock the wrong primitive for the lane lifetime. Replace the spinlock-based recursion model with a dynamically allocated per-lane mutex array and take the lane lock unconditionally. Add might_sleep() to catch any future atomic-context caller. Found with the ndctl unit test btt-check.sh.

• git.kernel.orghttps://git.kernel.org/stable/c/417918783bcfe0be135019df16a267b3af442efd
• git.kernel.orghttps://git.kernel.org/stable/c/4eafa810b042d985ec6bbf5b514414e73cee6f6f
• git.kernel.orghttps://git.kernel.org/stable/c/5c53406098b599c420b031e6ec5ba8a2f3794c50
• git.kernel.orghttps://git.kernel.org/stable/c/73e35c1bdfa160b41fdbe204e02325f0687de506
• git.kernel.orghttps://git.kernel.org/stable/c/8d4b989d9c9afe5f185aa5853b666fc4617afe9e
• git.kernel.orghttps://git.kernel.org/stable/c/fd7a97b2514cfc4b4cc067a27dd39bde2a8b1735

No related CVEs by CWE or product.