The Linux kernel’s RAID implementations for raid1, raid10, and raid456 incorrectly handle write error scenarios when the REQ_NOWAIT flag is used. During a write, if one mirror succeeds and another returns -EAGAIN, the kernel cannot distinguish between temporary queue pressure and a genuine device failure. As a result, it neither records a bad block nor retries the write safely without REQ_NOWAIT, causing mirrors to drift apart and producing divergent data on the same logical volume. This flaw does not enable remote code execution but can lead to corruption of mirrored data, compromising the integrity of stored information.

The affected product is the Linux kernel in all distributions that include a kernel version in which raid1, raid10, and raid456 personalities advertise REQ_NOWAIT support. Kernel releases prior to the implementation change that removed this support remain vulnerable. The vulnerability is not limited to a specific vendor or distribution, as the kernel code is shared across all Linux‑using systems.

The CVSS score of 7.1 classifies this issue as high‑medium severity. The EPSS score of less than 1% indicates a very low likelihood of exploitation in the wild. The vulnerability is not listed in the CISA KEV catalog. Exploitation requires initiating kernel‑level write operations to a RAID volume that still advertises REQ_NOWAIT support, implying local or elevated privileges. No public exploit code is known. The primary risk remains the potential for data corruption on affected systems if the flaw is exercised while the kernel advertises REQ_NOWAIT support.