Executive brief
A technical issue was identified in the Linux kernel's extensible scheduler (sched_ext) where certain system configuration changes could trigger an unnecessary system warning. When managing system resources via cgroups, specifically when toggling CPU controls, the kernel might incorrectly flag a task migration as an error. This is a stability and logging issue rather than a security threat that would allow data theft or unauthorized access.
Technical details
A race condition or logic mismatch in `scx_cgroup_move_task()` caused a kernel warning (WARN) when `cgrp_moving_from` was NULL. This occurred because `scx_cgroup_can_attach()` tracks migrations based on cgroup identity, whereas the underlying migration mechanism drives moves based on CSS (Cgroup Subsystem State) identity. In scenarios where a CPU controller is toggled, a task may undergo a CSS-only migration without a change in its actual cgroup, leading to a NULL `cgrp_moving_from` pointer. The fix involves removing the incorrect warning and ensuring the function gracefully skips tasks where no cgroup change has occurred. This issue primarily affects Linux kernels version 6.12 and newer using the sched_ext scheduler.
Affected products
- Linux Linux Kernel 6.12+
Timeline
- 2026-06-01: disclosed: Initial report by Matt Fleming
- 2026-06-02: patched: Fix committed to mainline kernel by Tejun Heo
- 2026-07-01: advisory: CVE published in NVD dataset