Created attachment 1130599 [details] nested vm libvirt log Description of problem: Nested VM is freezing during various activities like: installing fedora, booting fedora, `systemctl isolate graphical.target`, shutting down fedora ... My setup: Laptop - Fedora 22 virt-manager.noarch 1.2.1-4.fc22 qemu-kvm.x86_64 2:2.3.1-11.fc22 libvirt-daemon.x86_64 1.2.13.2-2.fc22 $ modinfo kvm filename: /lib/modules/4.3.4-200.fc22.x86_64/kernel/arch/x86/kvm/kvm.ko.xz license: GPL author: Qumranet depends: intree: Y vermagic: 4.3.4-200.fc22.x86_64 SMP mod_unload parm: allow_unsafe_assigned_interrupts:Enable device assignment on platforms without interrupt remapping support. (bool) parm: ignore_msrs:bool parm: min_timer_period_us:uint parm: kvmclock_periodic_sync:bool parm: tsc_tolerance_ppm:uint parm: lapic_timer_advance_ns:uint parm: halt_poll_ns:uint parm: halt_poll_ns_grow:int parm: halt_poll_ns_shrink:int laptop runs vm in virt-manager that serves as oVirt host (managed by vdsm) VDSM Host - Fedora 23 (VM): libvirt-daemon.x86_64 1.2.18.2-2.fc23 qemu-kvm.x86_64 2:2.4.1-7.fc23 $ modinfo kvm filename: /lib/modules/4.3.5-300.fc23.x86_64/kernel/arch/x86/kvm/kvm.ko.xz license: GPL author: Qumranet depends: intree: Y vermagic: 4.3.5-300.fc23.x86_64 SMP mod_unload parm: allow_unsafe_assigned_interrupts:Enable device assignment on platforms without interrupt remapping support. (bool) parm: ignore_msrs:bool parm: min_timer_period_us:uint parm: kvmclock_periodic_sync:bool parm: tsc_tolerance_ppm:uint parm: lapic_timer_advance_ns:uint parm: halt_poll_ns:uint parm: halt_poll_ns_grow:int parm: halt_poll_ns_shrink:int VDSM host started nested VM trying to install and run Fedora 23 `journalctl` if VDSM host shows: Feb 25 16:52:16 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x619 Feb 25 16:52:16 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x641 Feb 25 16:52:16 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x639 Feb 25 16:52:16 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x611 Feb 25 16:52:13 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x34 Feb 25 16:52:13 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x606 Feb 25 16:52:12 localhost kernel: kvm [1968]: vcpu1 unhandled rdmsr: 0x570 Feb 25 16:52:12 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x570 Feb 25 16:52:06 localhost libvirtd[1195]: Received unexpected event 3 Feb 25 16:52:06 localhost libvirtd[1195]: Received unexpected event 3 ... Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x619 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x641 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x639 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x611 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x619 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x641 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x639 Feb 25 16:33:54 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x611 Feb 25 16:33:49 localhost kernel: kvm [1968]: vcpu1 unhandled rdmsr: 0x606 Feb 25 16:33:46 localhost kernel: kvm [1968]: vcpu1 unhandled rdmsr: 0x570 Feb 25 16:33:46 localhost kernel: kvm [1968]: vcpu0 unhandled rdmsr: 0x570 Feb 25 16:33:33 localhost libvirtd[1195]: Received unexpected event 3 How reproducible: Approx. every third boot, hardly
Please provide /proc/cpuinfo Did this ever work reliably for you? As in, did this regress recently? Unfortunately unhandled msr warnings are pretty common and likely aren't indicative of whatever the root issue is.
Created attachment 1130642 [details] /proc/cpuinfo of laptop (root system)
Created attachment 1130643 [details] /proc/cpuinfo of vdsm host (vm)
Created attachment 1130644 [details] /proc/cpuinfo of nested vm
It is for the first time I run actual OS in nested VM so I can't tell.
You can try setting the ignore_msrs kvm module parameter... if things end up working that will indicate if it's an msr issue or not
(In reply to Cole Robinson from comment #6) > You can try setting the ignore_msrs kvm module parameter... if things end up > working that will indicate if it's an msr issue or not jniederm did this help at all?
I'm sorry. The VM got lost and I don't have capacity to recreate it. Closing. Thank you for your help.