Linux Kernel Daily: Race Condition Cleanup Across Cgroups, Workqueues, and EDAC
A wave of fixes landed correcting subtle race conditions and boundary-check bugs in cgroups, workqueues, and sched_ext, alongside a large EDAC maintenance pull that hardens memory error detection on Intel and Altera platforms. The common thread: locking and validation edge cases that only surface under specific timing or hardware conditions.
Duration: PT2M32S
Episode overview
This episode is a short developer briefing from Linux Kernel Daily.
It explains recent repository work in plain language.
- Show: Linux Kernel Daily
- Published: 2026-09-01T06:03:43Z
- Audio duration: PT2M32S
Transcript excerpt
This excerpt keeps the crawler page concise. Listen to the episode or use the RSS feed for the full update.
Good morning. It's September 1st, and today's kernel activity centers on one theme: closing timing and validation gaps that only bite in edge cases.
Start with cgroups. Commit 8e35992 from Etienne Perot fixes a real footgun: after writing to cgroup dot kill, any child process cloned into that cgroup with clone-into-cgroup was getting spuriously SIGKILLed, indefinitely, because a kill counter was snapshotted before the target cgroup was even resolved. That's now…
Workqueues saw a matching set of race and boundary fixes. Commit 068c35b from Jiacheng Xu rejects watchdog thresholds that would overflow jiffies during multiplication - a quiet but real integer overflow risk. Separately, an unbound worker pool use-after-free was fixed by unlinking it from the pending-activation…
Sched_ext picked up similar precision fixes: a dispatch queue ownership check was moved under the queue lock to stop spurious scheduler aborts when a task exits mid-move, and cgroup bandwidth callbacks can now be sleepable, matching how they actually run.
The EDAC pull, merged by Linus in 786262b, is more maintenance-scale but still notable: mpc85xx and ThunderX drivers are now orphaned for lack of hardware access,…
XF…
Nearby episodes from Linux Kernel Daily
- Weekly Recap - Late Fixes Sweep Ahead of the 7.3 Merge Window Close
- The Great Merge-Window Cleanup
- SCSI Layer Hardens Against Unsafe Locking
- Merge Window Hardening Wave
- Packet Scheduler Denial-of-Service Fixes
- The Late Merge Window Cleanup Wave
- The Merge Window Closes on Race Conditions and Resource Leaks
- The Merge Window Opens for 7.3