FS#30679 - [qemu-kvm] KVM virtual machines core dump with virtio

Attached to Project: Arch Linux
Opened by Tobias Hunger (hunger) - Saturday, 14 July 2012, 20:24 GMT
Last edited by Tobias Powalowski (tpowa) - Wednesday, 18 July 2012, 05:50 GMT
Task Type Bug Report
Category Packages: Core
Status Closed
Assigned To Tobias Powalowski (tpowa)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:
All of my KVM virtual machines won't start anymore after upgrading my server to linux-3.4.4-3.

All VMs worked fine before I rebooted the server after the recent glibc/linux/kmod update (which went fine).
All machines start up fine, up to the point the network is brought up.

This is found in /var/log/messages of the host system:
Jul 14 21:57:24 jj kernel: [ 306.597037] device vnet0 entered promiscuous mode
Jul 14 21:57:24 jj kernel: [ 306.602359] br0: port 2(vnet0) entered forwarding state
Jul 14 21:57:24 jj kernel: [ 306.602406] br0: port 2(vnet0) entered forwarding state
Jul 14 21:57:39 jj kernel: [ 321.653322] br0: port 2(vnet0) entered forwarding state
Jul 14 21:58:12 jj kernel: [ 354.635423] qemu-kvm[1807]: segfault at 10 ip 00007fd3e4fbbd0e sp 00007fd3dc892c00 error 4 in qemu-system-x86_64[7fd3e4e8b000+308000]
Jul 14 21:58:12 jj kernel: [ 354.636773] br0: port 2(vnet0) entered disabled state
Jul 14 21:58:12 jj avahi-daemon[1086]: Withdrawing workstation service for vnet0.

I could not find anything enlightening in /var/log/libvirt/*: That basically reports the machine starting up and coming done right afterwards.

This is something in the host: All VMs are effected, even when booting from the CDs the VMs were originally installed from.

The issue vanishes when changing the NIC configuration from virtio to "hypervisor default".

Additional info:
* linux 3.4.4-3, qemu-kvm 1.1.0-1

Steps to reproduce:
* Set up a VM using virtio
* Boot from any linux CD that sets up networking
This task depends upon

Closed by  Tobias Powalowski (tpowa)
Wednesday, 18 July 2012, 05:50 GMT
Reason for closing:  Fixed
Additional comments about closing:  1.1.1-1
Comment by Tobias Powalowski (tpowa) - Monday, 16 July 2012, 08:19 GMT
Please report upstream, i don't think this is a packaging issue.
Comment by Tobias Powalowski (tpowa) - Tuesday, 17 July 2012, 14:54 GMT
Status on 1.1.1?
Comment by Tobias Hunger (hunger) - Tuesday, 17 July 2012, 21:57 GMT
Fixed in 1.1.1. Thanks for the update!

Loading...