The Linux kernel memcg subsystem serializes file operations per open file but allows concurrent updates to the same memory.high or memory.max file via separate open files. Both update handlers store the new limit before synchronous reclaim but continue using the writer's local target during reclaim. If another writer raises or removes the limit, the first writer's reclaim loop may continue toward a stale target, causing an indefinite loop once reclaim retries are exhausted. This leads to the OOM path incorrectly returning true without killing processes, while the writer still records OOM events. The patch fixes this by verifying the current limit at each reclaim iteration and stopping if it differs from the writer's target. The issue was found via automated randomized testing and is not known to have been exploited in production.

The vulnerability can cause a reclaim operation in the Linux kernel memcg subsystem to loop indefinitely, potentially blocking the writer and causing repeated OOM event increments. This may affect system stability under specific memory limit update conditions but has not been reported in production workloads or exploited in the wild.

A fix is available that checks the current memory limit at the start of each reclaim iteration and stops reclaiming if the limit no longer matches the writer's target. Users should apply the official Linux kernel patch that resolves this issue. No additional mitigation is indicated.