The kernel’s RAID1 subsystem has a flaw where the function raid1_run() registers a thread and, if a subsequent limit setting fails, does not deregister that thread. The unreleased md_thread structure and thread resource remain in memory, creating a leak that can grow with repeated errors and eventually exhaust system memory or thread tables. The vulnerability does not directly grant an attacker access or execution capability, but it can degrade availability by affecting system stability when trigger conditions occur.

All Linux kernel releases that include the unpatched md/raid1 code are affected. The issue was identified before the commit that fixed the leak, so any kernel built from the recommended source trees lacking the patch is vulnerable. No specific version list is supplied; rather, any kernel that contains the original raid1_run() implementation without the cleanup logic is impacted.

The CVSS score of 5.5 is available, but the EPSS score and KEV status are not available, and the vulnerability is not listed in CISA KEV. The risk assessment therefore relies on the nature of the flaw: a memory leak can lead to resource exhaustion, but exploitation requires conditions that cause raid1_set_limits() to fail repeatedly. This likely demands local or privileged access, and no public exploit is known. Consequently, the overall risk is moderate to low, pending further details about the failure trigger frequency in production workloads.