[Bug 196161] New: Kernel freezes by starting a kvm vm.

[bugzilla-daemon@bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=196161

            Bug ID: 196161
           Summary: Kernel freezes by starting a kvm vm.
           Product: Virtualization
           Version: unspecified
    Kernel Version: 4.11.6
          Hardware: x86-64
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: high
          Priority: P1
         Component: kvm
          Assignee: virtualization_kvm@kernel-bugs.osdl.org
          Reporter: paulkek@protonmail.com
        Regression: No

Created attachment 257135
  --> https://bugzilla.kernel.org/attachment.cgi?id=257135&action=edit
dmesg

Hey,

I've been getting system freezes from time to time but I wasn't sure what
caused them because there were no logs or anything. A hard reboot was required
then.

Today I've actually found a way to reproduce this freeze and I was lucky enough
that I was able to get some logs about this freezes through a remote ssh
session which was very slow (you will see why.)

First of all my system:

uname -a: Linux 4.11.6-1-ck #1 SMP PREEMPT Thu Jun 22 02:06:23 CEST 2017 x86_64
GNU/Linux

CPU: Intel i7 6700k
RAM: 4 x 8GB DDR4 @ 2400MHz
GPU: NVIDIA GTX 1060 with NVIDIA's latest proprietary drivers
MB: Asus Z170-A with the latest BIOS

Note this this also happens with the vanilla kernel as I am also using a kernel
with the muqss scheduler (ck-patchset).

I can reproduce this issue by doing the following:

1.) Start applications like (chromium/pidgin/spotify)
2.) Start a windows 7 vm (libvirt/kvm/virt-manager) (the vm image is on an ssd)
3.) Start a windows 10 vm (the vm image is on an hdd)

This isn't a scientific reproduction and I am not even sure how exactly but
this worked every time I tried it.

I have observed the following:

- The system freezes / It's actually more like everything is going slow (The
audio was looping). Usually everything freezes and only a hard reboot was
possible. But I had luck that somehow it froze (audio was in a loop, everything
desktop wise was unresponsive) but I could actually move the mouse, so I
quickly changed to a different tty session and could login again but again
everything was very slow I couldn't even run a proper command.

But I guess this was helpful because I actually saw some proper kernel
warnings:

NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [CPU 1/KVM:2411]
NMI watchdog: BUG: soft lockup - CPU#6 stuck for 22s! [CPU 1/KVM:2629]
NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [CPU 1/KVM:2411]
NMI watchdog: BUG: soft lockup - CPU#6 stuck for 22s! [CPU 1/KVM:2629]

INFO: rcu_preempt self-detected stall on CPU
INFO: rcu_preempt self-detected stall on CPU
...

More warning which contains lots of stuff so I've simply taken a picture and
attached it here.

Finally I've added the dmesg from the ssh session, it is probably incomplete as
after some minutes it completely froze and the connection was broken.

The crash happens after the "kvm_get_msr_common: 6 callbacks suppressed" lines.
It basically starts with the messages about NVRM and followed by some ata4
error messages and more.

I've also included this in the attachments.

-- 
You are receiving this mail because:
You are watching the assignee of the bug.