[ovirt-users] How to passthrough GPU to windows guest

Wed Mar 29 06:28:02 UTC 2017

Thank you very much.
I don't have quadro k2200, but I tried amd firepro s7150 again. There
is no code 43 error. But when I reboot the vm, the remote-viewer's
screen get displaying the "Starting Windows" splash screen, no matter
using spice or vnc protocol.
I tried to use nx protocl, I found the vm boot up all right  and gpu
passthrough is successful.
What is the right way to connect to the gpu passthrough's vm? Are
there any other cards working fine, for example cards in quadro
series.

Best Regards

2017-03-28 16:48 GMT+08:00 Martin Polednik <mpolednik at redhat.com>:
> On 28/03/17 16:22 +0800, yimao wrote:
>>
>> Thanks for your suggestions, I have tried that but problems still
>> exist. Yes, I used Nvidia GTX 750Ti.
>>
>> I found the qemu's arguments have changed.
>>
>>    7210 ?        SLl    0:31 /usr/libexec/qemu-kvm -name
>> guest=win7_1,debug-threads=on -S -object
>>
>> secret,id=masterKey0,format=raw,file=/var/lib/libvirt/qemu/domain-16-win7_1/master-key.aes
>> -machine pc-i440fx-rhel7.3.0,accel=kvm,usb=off -cpu
>>
>> Broadwell,+rtm,+hle,hv_time,hv_relaxed,hv_vapic,hv_spinlocks=0x1fff,hv_vendor_id=whatever,kvm=off
>> -m size=1048576k,slots=16,maxmem=4194304k -realtime mlock=off -smp
>> 1,maxcpus=16,sockets=16,cores=1,threads=1 -numa
>> node,nodeid=0,cpus=0,mem=1024 -uuid
>> 1fba3bd6-0659-4463-92ea-8efa43411082 -smbios
>> type=1,manufacturer=oVirt,product=oVirt
>>
>> Node,version=7-3.1611.el7.centos,serial=031B021C-040D-050E-D806-900700080009,uuid=1fba3bd6-0659-4463-92ea-8efa43411082
>> -no-user-config -nodefaults -chardev
>>
>> socket,id=charmonitor,path=/var/lib/libvirt/qemu/domain-16-win7_1/monitor.sock,server,nowait
>> -mon chardev=charmonitor,id=monitor,mode=control -rtc
>> base=2017-03-28T08:54:41,driftfix=slew -global
>> kvm-pit.lost_tick_policy=discard -no-hpet -no-shutdown -boot strict=on
>> -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -device
>> virtio-scsi-pci,id=scsi0,bus=pci.0,addr=0x8 -device
>> virtio-serial-pci,id=virtio-serial0,max_ports=16,bus=pci.0,addr=0x4
>> -drive if=none,id=drive-ide0-1-0,readonly=on -device
>> ide-cd,bus=ide.1,unit=0,drive=drive-ide0-1-0,id=ide0-1-0 -drive
>>
>> file=/rhev/data-center/00000001-0001-0001-0001-000000000311/c053a0c0-45ad-4f7a-b9ce-c1ac90c900d4/images/082d4d4c-3aae-4369-9cbb-e6a7b31c3204/63da97f3-7b1f-406b-905f-b12eeb96d55f,format=raw,if=none,id=drive-scsi0-0-0-0,serial=082d4d4c-3aae-4369-9cbb-e6a7b31c3204,cache=none,werror=stop,rerror=stop,aio=threads
>> -device
>> scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,bootindex=1
>> -netdev tap,fd=33,id=hostnet0,vhost=on,vhostfd=35 -device
>>
>> virtio-net-pci,netdev=hostnet0,id=net0,mac=00:1a:4a:16:01:51,bus=pci.0,addr=0x3
>> -chardev
>> socket,id=charchannel0,path=/var/lib/libvirt/qemu/channels/1fba3bd6-0659-4463-92ea-8efa43411082.com.redhat.rhevm.vdsm,server,nowait
>> -device
>> virtserialport,bus=virtio-serial0.0,nr=1,chardev=charchannel0,id=channel0,name=com.redhat.rhevm.vdsm
>> -chardev
>> socket,id=charchannel1,path=/var/lib/libvirt/qemu/channels/1fba3bd6-0659-4463-92ea-8efa43411082.org.qemu.guest_agent.0,server,nowait
>> -device
>> virtserialport,bus=virtio-serial0.0,nr=2,chardev=charchannel1,id=channel1,name=org.qemu.guest_agent.0
>> -device usb-tablet,id=input0,bus=usb.0,port=1 -vnc
>> 192.168.120.235:0,password -k en-us -device
>>
>> qxl-vga,id=video0,ram_size=67108864,vram_size=8388608,vram64_size_mb=0,vgamem_mb=16,bus=pci.0,addr=0x2
>> -device vfio-pci,host=01:00.0,id=hostdev0,bus=pci.0,addr=0x6 -device
>> vfio-pci,host=01:00.1,id=hostdev1,bus=pci.0,addr=0x7 -device
>> virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x5 -object
>> rng-random,id=objrng0,filename=/dev/urandom -device
>> virtio-rng-pci,rng=objrng0,id=rng0,bus=pci.0,addr=0x9 -msg
>> timestamp=on
>>
>> Does ovirt use vga passthrough or gpu passthrough? Should I install EDK2
>> OVMF?
>
>
> It uses secondary GPU passthrough; the VGA is still occupied by the
> spice QXL. No OVMF necessary at all.
>
> That being said, if you still have the problem, I'd try to maybe work
> with the hook to also disable hyperv and whatever you find around the
> internet. It looks like everyone's card needs something different. To
> be on the safe side, quadro k2200 seem to work fine.
>
>
>> Thanks in advance.
>>
>> 2017-03-28 15:17 GMT+08:00 Martin Polednik <mpolednik at redhat.com>:
>>>
>>> On 28/03/17 02:12 +0800, yimao wrote:
>>>>
>>>>
>>>> Hi,
>>>>
>>>> I'm trying to passthrough gpu card to windows 7 guest, but I got the
>>>> error code 43.
>>>> I use ovirt node: ovirt-node-ng-installer-ovirt-4.1-2017032304.iso and
>>>> ovirt-engine-appliance-4.1-20170322.1.el7.centos.noarch.rpm
>>>> I have added  options "intel_iommu=on rdblacklist=nouveau
>>>> pci-stub.ids=10de:1380,10de:0fbc" to the grub.conf and attached these
>>>> devices to the guest. After I have installed the gpu's driver, I got
>>>> the error code 43.
>>>> Did I miss something? Any pointers or suggestions would be greatly
>>>> appreciated.
>>>
>>>
>>>
>>> You are most likely using consumer grade GPU that errors out in
>>> Windows if it's running under virtualized environment. You could try
>>> using the following hook (make sure it's in
>>> /usr/libexec/vdsm/hooks/before_vm_start and it's executable bit is set
>>> (chmod +x ...)).
>>>
>>> Just copy paste the following block to a bash and try to start the VM
>>> again.
>>>
>>> cd /usr/libexec/vdsm/hooks/before_vm_start
>>> cat << EOF >> 99_mask_kvm
>>> #!/usr/bin/python2
>>>
>>> import hooking
>>> domxml = hooking.read_domxml()
>>>
>>> hyperv = domxml.getElementsByTagName('hyperv')[0]
>>> smm = domxml.createElement('vendor_id')
>>> smm.setAttribute('state', 'on')
>>> smm.setAttribute('value', 'whatever')
>>> hyperv.appendChild(smm)
>>>
>>> features = domxml.getElementsByTagName('features')[0]
>>> kvm = domxml.createElement('kvm')
>>> hidden = domxml.createElement('hidden')
>>> hidden.setAttribute('state', 'on')
>>> kvm.appendChild(hidden)
>>> features.appendChild(kvm)
>>>
>>> hooking.write_domxml(domxml)
>>> EOF
>>> chmod +x 99_mask_kvm
>>>
>>>
>>>
>>>> Thanks in advance.
>>>> Yiimao Yang.