Hi Krystian, Update on the bisection tests. I ran mainline v6.8.12 (srv04) and v6.12 (srv08) in parallel.
**srv04 on v6.8.12:** Stable so far (~2 hours uptime, no panic). **srv08 on v6.12.0:** After about 36 minutes, this host started showing hung task warnings - a *different* symptom than the original panic: ``` INFO: task worker:8185 blocked for more than 122 seconds. Not tainted 6.12.0-061200-generic #20241220723 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. ``` The stack trace for the blocked worker goes through `xfs_file_write_iter` -> `do_iter_readv_writev` -> `vfs_writev` -> `__x64_sys_pwritev`, i.e. it's blocked on an XFS write I/O syscall, not on scheduler/stopper machinery like the original `multi_cpu_stop`/`cpu_stopper_thread` panic. Several worker threads (PIDs 8185, 8323, 8324, 8325, 8337, 14050) were reported blocked in this way, with rapidly increasing load average (14-27), though the host itself stayed up and all 8 guest VMs kept running - it didn't panic outright, just degraded significantly. Further hung task reports got suppressed after that. This doesn't look like the same failure mode we've been chasing, so I wanted to flag it separately rather than conflate it with the original bug. Not sure if it's worth investigating as a distinct issue or if you'd rather I just note it and move on with the bisection using a different host/kernel for the 6.12 test point. I'll let srv04 keep running on v6.8.12 to see if the original panic reproduces there. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2163642 Title: Kernel panic "Attempted to kill the idle task" on AMD Opteron multi- node NUMA under KVM (6.8.0-137) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163642/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
