Executive brief
The Linux kernel's RAID1 and RAID10 storage subsystems contain a bug in how they track I/O operations (bio accounting) when failed storage requests are retried. This can result in incorrect accounting of storage I/O metrics, leading to potential performance monitoring inaccuracies or operational visibility issues in systems relying on RAID storage.
Technical details
The vulnerability is a logic error in the bio (block I/O) accounting mechanism for Linux kernel's RAID1 and RAID10 drivers. The vulnerable code uses unreliable flags (r1bio_existed in raid1 and io_accounting in raid10) to determine whether a bio should be accounted for. When a failed bio is split and resubmitted via bio_submit_split_bioset() on error recovery paths, this can cause either double accounting for md cloned bios or missing accounting for bios returned from bio_submit_split_bioset(). The fix replaces these unreliable checks with a call to md_cloned_bio() which properly detects md cloned bios and skips accounting accordingly. This is a data structure/accounting correctness issue, not a security vulnerability, and was patched in upstream and stable kernels starting May 2026.
Affected products
- Linux Linux kernel multiple versions (5.x through 7.x series)
Timeline
- 2026-08-15: disclosed
- 2026-05-01: patched