Hi Viktor, Update from srv04 (HP DL385p Gen8, Opteron 6386 SE, mainline 6.8.12, no fixes):
Second baseline run WITHOUT boot params crashed again, after ~8d 18h (first baseline run: ~3d; run WITH processor.max_cstate=1 idle=nomwait: 20d 22h, no crash). Final panic again "Attempted to kill the idle task!" (console only, not in journal). New finding: the persistent journal shows a kernel oops ~6 days BEFORE the final panic: Oct 03 09:16:36 general protection fault, probably for non-canonical address 0xfffb77e9d05d8b4c CPU: 21 PID: 3581 Comm: CPU 0/KVM Not tainted 6.8.12-060812-generic RIP: 0010:kvm_replace_memslot+0xea/0x760 [kvm] Call Trace: kvm_read_guest_cached+0x13/0x30 [kvm] apic_sync_pv_eoi_from_guest+0x53/0x120 [kvm] kvm_lapic_sync_from_vapic+0xae/0xc0 [kvm] vcpu_enter_guest+0xa57/0xfe0 [kvm] vcpu_run / kvm_arch_vcpu_ioctl_run / kvm_vcpu_ioctl Faulting instruction is a list-node store with corrupted pointers (RDX non-canonical, RSI looks like x86 code bytes) - suggests memory corruption / use-after-free rather than a pure scheduler race. At the same second: hpilo 0000:02:00.2 AMD-Vi IO_PAGE_FAULT (address=0xbde0e000). The host kept running normally for 6 days after the oops (no stuck CPU in node-exporter metrics), journal is clean until ~4 min before the panic. Earlier crash consoles (srv04 6.8.12, srv06 6.14.0) also showed taint "D", so a prior oops may precede the migrate_swap/multi_cpu_stop panic in those cases too. Question: could the migrate_swap/multi_cpu_stop panic be a secondary effect of this earlier KVM oops? Current state: srv04 running 6.8.12 WITH processor.max_cstate=1 idle=nomwait (3rd run, started 2026-10-09 20:12 UTC). Full journal saved, can attach. ** Attachment added: "srv04-panic-2026-10-09.log" https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163642/+attachment/6007770/+files/srv04-panic-2026-10-09.log -- 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
