The vulnerability arises from the Linux kernel failing to disable preemption between the scx_claim_exit() call and the subsequent kicking of a helper kthread, exposing a scheduling race condition (CWE-368). When a task is preempted in that narrow window, the BPF scheduler may fail to reschedule it, resulting in the helper work never being queued. Without the helper, bypass mode is never activated and tasks stop being dispatched, causing the system to wedge.

All Linux kernel installations that have not applied the patch which adds preemption disabling around scx_claim_exit() and the subsequent work kicking are vulnerable. The fix is applied to the upstream kernel via a commit that enforces preemption suppression across scx_claim_exit(), scx_disable(), and scx_vexit(). Use the vendor’s security advisories or kernel changelogs to verify whether your kernel version includes this commit.

The CVSS score is not provided and the EPSS score is < 1%, indicating a very low probability of exploitation. However, because the flaw leads to a full kernel lockup and requires a specific timing condition involving preemption and BPF scheduling, exploitation remains possible but unlikely, and the impact if successful would be catastrophic.