The vulnerable logic zeroed the entire cp_flags word in rds_conn_path_reset() using a plain write, while other parts of the kernel accessed individual bits with atomic operations. This unsynchronized write created a race condition where concurrent accesses could silently overwrite each other, leading to inconsistent or lost RDS connection state information, potentially causing premature or incomplete shutdown of RDS connections and unpredictable service behavior. This is a classic race condition vulnerability and concurrent modification issue (CWE-366).

The flaw existed in the Linux kernel’s implementation of the Reliable Datagram Sockets (RDS) protocol. Any kernel that included the unsynchronized reset logic before the patch was introduced was affected. The available data does not list specific kernel versions that are affected, so it is unknown which precise releases contain the issue. The fix is present in recent upstream releases and the corresponding Oracle UEK update that preserves essential connection state flags.

Based on the description, it is inferred that the vulnerability creates a data race where concurrent RDS operations can overwrite cp_flags bits. The likely attack vector is concurrent RDS traffic racing against the cp_flags reset, which would require precise timing to trigger dangerous state loss. The CVSS score of 5.5 indicates moderate severity, while the EPSS score is less than 1%, suggesting a low likelihood of exploitation. The vulnerability is not listed in the CISA KEV catalog. Practical exploitation is unlikely in typical workloads, but can cause intermittent RDS service disruptions, especially in high‑traffic deployments.