The Linux kernel audit subsystem had a data race condition involving concurrent access to the audit_queue length field via skb_queue_len(). Multiple readers accessed this field without holding the queue lock or using READ_ONCE(), while the writer updated it with WRITE_ONCE() protected by a spinlock. This mismatch caused data races detected by KCSAN. The race was fixed by switching to skb_queue_len_lockless(), which uses READ_ONCE() internally, ensuring proper synchronization with existing WRITE_ONCE() writes.

Data races can lead to inconsistent or corrupted audit queue length values, potentially causing incorrect audit behavior or kernel instability. However, no known exploits are reported in the wild. The impact is primarily on kernel reliability and correctness in audit message handling.

A fix is available that replaces unsafe queue length accesses with a lockless helper using proper memory access primitives. Users should update to a Linux kernel version that includes this fix. Since this is a kernel-level issue, applying the official kernel patch or upgrading to a fixed kernel version is the recommended remediation.