[vfio-users] libvirt-3.0.0 from libvirt-2.4.0 breaks Windows VMs

sL1pKn07 SpinFlo sl1pkn07 at gmail.com
Thu Jan 26 18:05:47 UTC 2017


2017-01-26 19:05 GMT+01:00 sL1pKn07 SpinFlo <sl1pkn07 at gmail.com>:
> 2017-01-26 18:49 GMT+01:00 Anthony M de Beus <anthony.debeus at gmail.com>:
>> Haven't seen mention of this elsewhere yet, not sure if this is the best
>> place to post it.
>>
>> I just updated to libvirt-3.0.0-1 on Arch from libvirt-2.4.0-2 and my
>> Windows VMs no longer start. Reverting to libvirt-2.4.0-2 reverses the
>> problem.  The problem affects Windows 7 and XP, the XP machine doesn't
>> have GPU passthrough.  A CAELinux VM still runs.  journalctl shows,
>> under libvirtd 3.0.0
>>
>> Jan 26 06:46:35 bozo systemd[1]: Started Virtual Machine qemu-1-Windows7.
>> Jan 26 06:46:35 bozo virtlogd[30871]: 2017-01-26 13:46:35.878+0000:
>> 30871: info : libvirt version: 3.0.0
>> Jan 26 06:46:35 bozo virtlogd[30871]: 2017-01-26 13:46:35.878+0000:
>> 30871: info : hostname: bozo
>> Jan 26 06:46:35 bozo virtlogd[30871]: 2017-01-26 13:46:35.878+0000:
>> 30871: error : virNetSocketReadWire:1800 : End of file while reading
>> data: Input/output error
>> Jan 26 06:46:35 bozo virtlogd[30871]: 2017-01-26 13:46:35.880+0000:
>> 30871: error : virNetSocketReadWire:1800 : End of file while reading
>> data: Input/output error
>>
>> /var/log/libvirt/qemu/Windows7.log shows
>>
>> 2017-01-26 13:46:35.702+0000: starting up libvirt version: 3.0.0, qemu
>> version: 2.8.0, hostname: bozo
>> LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin
>> QEMU_AUDIO_DRV=spice /usr/sbin/qemu-system-x86_64 -name
>> guest=Windows7,debug-threads=on -S -object
>> secret,id=masterKey0,format=raw,file=/var/lib/libvirt/qemu/domain-1-Wind$
>> Domain id=1 is tainted: cdrom-passthrough
>> libvirt:  error : libvirtd quit during handshake: Input/output error
>> 2017-01-26 13:46:35.878+0000: shutting down, reason=failed
>>
>>
>> Whereas under 2.4.0, journalctl shows
>>
>> bozo systemd[1]: Started Virtual Machine qemu-3-Windows7.
>> Jan 26 10:28:07 bozo kernel: qemu-system-x86: sending ioctl 5326 to a
>> partition!
>> Jan 26 10:28:07 bozo dhcpcd[708]: vnet0: soliciting a DHCP lease
>> ......bunch of normal vm assignments....
>> Jan 26 10:28:11 bozo kernel: usb 4-1.8.4: reset high-speed USB device
>> number 8 using ehci-pci
>> Jan 26 10:28:11 bozo virtlogd[1512]: End of file while reading data:
>> Input/output error
>> Jan 26 10:28:12 bozo kernel: kvm: zapping shadow pages for mmio
>> generation wraparound
>> Jan 26 10:28:12 bozo kernel: kvm: zapping shadow pages for mmio
>> generation wraparound
>> Jan 26 10:28:14 bozo kernel: usb 4-1.2: reset low-speed USB device
>> number 4 using ehci-pci
>> ...bunch of normal vm assignments and synergy messages....
>> 017-01-26T10:29:33] NOTE: client "penguin" has connected
>> Jan 26 10:29:33 bozo synergys[1189]:
>> /build/synergy/src/synergy-1.8.7-stable/src/lib/server/Server.cpp,337
>> Jan 26 10:31:56 bozo synergys[1189]: [2017-01-26T10:31:56] NOTE: client
>> "penguin" has disconnected
>> Jan 26 10:31:56 bozo synergys[1189]:
>> /build/synergy/src/synergy-1.8.7-stable/src/lib/server/ClientProxy1_0.cpp,213
>>
>> and /var/log/libvirt/qemu/Windows7.log shows
>>
>> 2017-01-24 17:34:33.446+0000: starting up libvirt version: 2.4.0, qemu
>> version: 2.8.0, hostname: bozo
>> LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin
>> QEMU_AUDIO_DRV=spice /usr/sbin/qemu-system-x86_64 -name
>> guest=Windows7,debug-threads=on -S -object
>> secret,id=masterKey0,format=raw,file=/var/lib/libvirt/qemu/domain-1-Wind$
>> Domain id=1 is tainted: cdrom-passthrough
>> 2017-01-24T21:08:41.957097Z qemu-system-x86_64: terminating on signal 15
>> from pid 889 (/usr/bin/libvirtd)
>> 2017-01-24 21:08:43.686+0000: shutting down, reason=shutdown
>>
>>
>> In summary, looks like a handshake error with the network in
>> virNetSocketReadWire.  I found this
>> https://www.redhat.com/archives/libvir-list/2016-October/msg00745.html,
>> implementing a new libssh transport.
>>
>> Suggestions? Places to post this error other than here? I realize this
>> isn't probably a vfio related bug.
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> vfio-users mailing list
>> vfio-users at redhat.com
>> https://www.redhat.com/mailman/listinfo/vfio-users
>
>
> https://bugs.archlinux.org/task/52689

oops. bad paste:

https://bugzilla.redhat.com/show_bug.cgi?id=1413773




More information about the vfio-users mailing list