Arch Linux

Duelist

Member

Registered: 2014-09-22

Posts: 358

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

@ghormoon:
And what about the video output? Switchable graphics always was a problem, even on windows machines.
Since you're having a separate(logitech G15) keyboard attached, i guess you have your video output routed via external display connector to the separate screen, and the notebook is actually... closed? Y' trying to assemble something only from the available parts and the PC has a broken screen or something? Why you aren't doing this thing with a normal PC?..

If host will be headless - i'd suggest you chopping the said head off before doing anything with the VM.
Don't load the radeon. Don't start Kernel ModeSetting(KMS). Disable all video output as we'll need VGA free from anything.
And drop all the unnecessary devices like the SD controller just for testing the GPU part. Then, after everything's set, you might attach all the other devices needed. Stuff like SD controllers may be not the separate device, but a part of something integrated, having their ROM and power controls hidden in depths of proprietary stuff and closed firmwares.

---

The forum rules prohibit requesting support for distributions other than arch.
I gave up. It was too late.
What I was trying to do.
The reference about VFIO and KVM VGA passthrough.

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

as for video output, I use the internal display. the reason I'm not doing this on desktop PC is that desktop PC is too heavy to carry daily in backpack
as for the HW, there is not any switchable graphics. the integrated one in the i7 cpu is not wired in at all. there's no way to turn off the radeon if you wanna have some output. not like the laptops which allow you to switch between igp and radeon.
as for the g15 usb keyboard - when I go headless, I can't use the internal one for input, since it's connected as ps/2 one.  That's next level, which involves patching qemu. so that's the last one

I'll retry with blacklisted radeon, kms off, also tell grub not to change modes, let's see what it does.
current state is that linux VM can both show boot messages and actually work on the passed in GPU, windows goes black after showing logo. likely at the point where the radeon driver kicked in.

Duelist

Member

Registered: 2014-09-22

Posts: 358

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I'll retry with blacklisted radeon, kms off, also tell grub not to change modes, let's see what it does.
current state is that linux VM can both show boot messages and actually work on the passed in GPU, windows goes black after showing logo. likely at the point where the radeon driver kicked in.

Try attaching a VGA(or hdmi or whatever you have there) screen to the GPU. Maybe it gets in trouble with the integrated screen, i'm not sure how it's attached to the GPU. I guess you'd be disappointed to have a burden of a separate screen.
If you're saying that intel's GPU isn't wired anyhow at all - that looks promising, maybe you have a real, full-featured AMD GPU inside, that fact will simplify some things.

So, yeah, do the clean install with no unnecessary hardware attached and see how it goes.

If this will work - it'll be the first actual notebook we've observed passthrough running.

---

The forum rules prohibit requesting support for distributions other than arch.
I gave up. It was too late.
What I was trying to do.
The reference about VFIO and KVM VGA passthrough.

aw

Member

Registered: 2013-10-04

Posts: 921

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

@ghormoon

Are you letting the host radeon driver claim the GPU that you're assigning?  Linux GPU drivers are really not that good at releasing devices.  What happens if you force it to be bound to pci-stub?  Secondary laptop graphics are pretty much uncharted territory, but you've got a lot of errors throughout that dmesg log, which all seem to point to radeon.

---

http://vfio.blogspot.com
Looking for a more open forum to discuss vfio related uses?  Try https://www.redhat.com/mailman/listinfo/vfio-users

mutiny

Member

Registered: 2015-02-15

Posts: 13

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I've gotten everything working nicely with a Radeon card passed through to a Windows guest and using Intel integrated graphics for host. However, I've found that I am starting to use more USB devices in the Windows guest VM (game controllers, wheels, etc), and find it inconvenient to have to start and stop the VM each time I want to use a new/different USB device. From aw blog, it is mentioned that entire USB controlled could be passed to VM using same process (vfio-pci). But... from lsusb I find that the USB bus that a plugged in device resides on is always the same 3.0 controller bus no matter what physical USB port I plug into (motherboard has 2 USB 2.0 ports, 6 3.0 ports, and an additional 1x 2.0 and 1x 3.0 port are available through the case). lspci shows 2 EHCI 2.0 controllers and 1 XHCI 3.0 controllers, so it doesn't make sense that all USB ports would be handled by a single controller, especially the 2.0 ports. Is there something I am not understanding here? Would a separate pci-e USB card dedicated to the guest be a better option? What would be best way to get hot-pluggability of USB devices to guest?

Physical layout of motherboard for reference:

lspci | grep USB:

00:14.0 USB controller: Intel Corporation 9 Series Chipset Family USB xHCI Controller
00:1a.0 USB controller: Intel Corporation 9 Series Chipset Family USB EHCI Controller #2
00:1d.0 USB controller: Intel Corporation 9 Series Chipset Family USB EHCI Controller #1

lsusb:

Bus 004 Device 002: ID 8087:8001 Intel Corp. 
Bus 004 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 003 Device 002: ID 8087:8009 Intel Corp. 
Bus 003 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 002 Device 002: ID 045b:0210 Hitachi, Ltd 
Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 007: ID 09da:9066 A4Tech Co., Ltd. F3 V-Track Gaming Mouse
Bus 001 Device 006: ID 05c7:2012 Qtronix Corp. 
Bus 001 Device 002: ID 045b:0209 Hitachi, Ltd 
Bus 001 Device 021: ID 045e:028e Microsoft Corp. Xbox360 Controller
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

A4Tech and Qtronix are plugged into motherboard 2.0 port, Xbox360 controller is plugged into 3.0 port, but everything gets put on Bus 001...

$ readlink /sys/bus/usb/devices/usb1
../../../devices/pci0000:00/0000:00:14.0/usb1
$ readlink /sys/bus/usb/devices/usb2
../../../devices/pci0000:00/0000:00:14.0/usb2
$ readlink /sys/bus/usb/devices/usb3
../../../devices/pci0000:00/0000:00:1a.0/usb3
$ readlink /sys/bus/usb/devices/usb4
../../../devices/pci0000:00/0000:00:1d.0/usb4

Shouldn't the 2.0 ports at least go to Bus 003 or Bus 004?

Last edited by mutiny (2015-05-25 10:42:27)

slis

Member

Registered: 2014-06-02

Posts: 127

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Usb 2.0 ports are probably ones on motherboard, usb headers for front usb...

mutiny

Member

Registered: 2015-02-15

Posts: 13

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Usb 2.0 ports are probably ones on motherboard, usb headers for front usb...

The motherboard has both 2.0 and 3.0 ports, as well as headers for front/external 2.0 and 3.0, and even if I plug into either of the front USB ports they still end up on same bus/controller?

Last edited by mutiny (2015-05-25 10:41:45)

slis

Member

Registered: 2014-06-02

Posts: 127

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

For me on Z77 intel board, usb 2.0 are accessible through 7 pin (or 9) headers on board, and on some revisions also 2 top ports on rear.

They work with vfio passing 00:1d.0, 00:1a.0 or 00:1f.0 (depends on revision)

Last edited by slis (2015-05-25 10:47:14)

fld

Member

Registered: 2015-02-06

Posts: 9

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I've gotten everything working nicely with a Radeon card passed through to a Windows guest and using Intel integrated graphics for host. However, I've found that I am starting to use more USB devices in the Windows guest VM (game controllers, wheels, etc), and find it inconvenient to have to start and stop the VM each time I want to use a new/different USB device. From aw blog, it is mentioned that entire USB controlled could be passed to VM using same process (vfio-pci). But... from lsusb I find that the USB bus that a plugged in device resides on is always the same 3.0 controller bus no matter what physical USB port I plug into (motherboard has 2 USB 2.0 ports, 6 3.0 ports, and an additional 1x 2.0 and 1x 3.0 port are available through the case). lspci shows 2 EHCI 2.0 controllers and 1 XHCI 3.0 controllers, so it doesn't make sense that all USB ports would be handled by a single controller, especially the 2.0 ports. Is there something I am not understanding here? Would a separate pci-e USB card dedicated to the guest be a better option? What would be best way to get hot-pluggability of USB devices to guest?

I had this exact same problem with my ASRock Z87 Extreme 6. No matter what USB port I used (external or internal), the device would always end up behind the XHCI controller. This is how the situation looks like in lshw:

        *-usb:0
             description: USB controller
             product: 8 Series/C220 Series Chipset Family USB xHCI
             vendor: Intel Corporation
             physical id: 14
             bus info: pci@0000:00:14.0
             version: 05
             width: 64 bits
             clock: 33MHz
             capabilities: pm msi xhci bus_master cap_list
             configuration: driver=xhci_hcd latency=0
             resources: irq:34 memory:f0c20000-f0c2ffff
           *-usbhost:0
                product: xHCI Host Controller
                vendor: Linux 4.0.0-1-amd64 xhci-hcd
                physical id: 0
                bus info: usb@4
                logical name: usb4
                version: 4.00
                capabilities: usb-3.00
                configuration: driver=hub slots=6 speed=5000Mbit/s
              *-usb
                   description: USB hub
                   product: AS2107
                   vendor: ASMedia
                   physical id: 3
                   bus info: usb@4:3
                   version: 0.01
                   serial: USB2.0 Hub
                   capabilities: usb-3.00
                   configuration: driver=hub maxpower=8mA slots=4 speed=5000Mbit/s
           *-usbhost:1
                product: xHCI Host Controller
                vendor: Linux 4.0.0-1-amd64 xhci-hcd
                physical id: 1
                bus info: usb@3
                logical name: usb3
                version: 4.00
                capabilities: usb-2.00
                configuration: driver=hub slots=14 speed=480Mbit/s
              *-usb:0
                   description: USB hub
                   product: AS2107
                   vendor: ASMedia
                   physical id: 3
                   bus info: usb@3:3
                   version: 0.01
                   serial: USB2.0 Hub
                   capabilities: usb-2.10
                   configuration: driver=hub maxpower=100mA slots=4 speed=480Mbit/s
              *-usb:1
                   description: Mouse
                   product: Powerful Receiver
                   vendor: Itron
                   physical id: 4
                   bus info: usb@3:4
                   version: 1.00
                   capabilities: usb-2.00
                   configuration: driver=usbhid maxpower=40mA speed=2Mbit/s
              *-usb:2
                   description: Audio device
                   product: C-Media USB Headphone Set
                   vendor: C-Media Electronics, Inc.
                   physical id: d
                   bus info: usb@3:d
                   version: 1.00
                   capabilities: usb-1.10 audio-control
                   configuration: driver=usbhid maxpower=100mA speed=12Mbit/s


        *-usb:1
             description: USB controller
             product: 8 Series/C220 Series Chipset Family USB EHCI #2
             vendor: Intel Corporation
             physical id: 1a
             bus info: pci@0000:00:1a.0
             version: 05
             width: 32 bits
             clock: 33MHz
             capabilities: pm debug ehci bus_master cap_list
             configuration: driver=ehci-pci latency=0
             resources: irq:16 memory:f0c3c000-f0c3c3ff
           *-usbhost
                product: EHCI Host Controller
                vendor: Linux 4.0.0-1-amd64 ehci_hcd
                physical id: 1
                bus info: usb@1
                logical name: usb1
                version: 4.00
                capabilities: usb-2.00
                configuration: driver=hub slots=2 speed=480Mbit/s
              *-usb
                   description: USB hub
                   vendor: Intel Corp.
                   physical id: 1
                   bus info: usb@1:1
                   version: 0.05
                   capabilities: usb-2.00
                   configuration: driver=hub slots=6 speed=480Mbit/s


        *-usb:2
             description: USB controller
             product: 8 Series/C220 Series Chipset Family USB EHCI #
             vendor: Intel Corporation
             physical id: 1d
             bus info: pci@0000:00:1d.0
             version: 05
             width: 32 bits
             clock: 33MHz
             capabilities: pm debug ehci bus_master cap_list
             configuration: driver=ehci-pci latency=0
             resources: irq:23 memory:f0c3b000-f0c3b3ff
           *-usbhost
                product: EHCI Host Controller
                vendor: Linux 4.0.0-1-amd64 ehci_hcd
                physical id: 1
                bus info: usb@2
                logical name: usb2
                version: 4.00
                capabilities: usb-2.00
                configuration: driver=hub slots=2 speed=480Mbit/s
              *-usb
                   description: USB hub
                   vendor: Intel Corp.
                   physical id: 1
                   bus info: usb@2:1
                   version: 0.05
                   capabilities: usb-2.00
                   configuration: driver=hub slots=8 speed=480Mbit/s

Why would manufacturers implement the USB 2.0 ports behind USB 3.0 controllers (and even throw in an additional ASMedia AS2107 hub), is quite strange. How can it possibly be cheaper to do it that way, than to wire the USB 2.0 ports from the chipset?

Anyways, I solved the problem by buying (reluctantly) a cheap uPD720201 based PCI-e USB 3.0 controller off ebay and passed that trough to win7. Works great. But it really sucks having to use an extension card, when the motherboard physically already has multiple USB controllers and plenty of ports :\

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

well, I've retried linux guest without radeon on the host (added to grub defaults "intel_iommu=on nofb nomodeset modprobe.blacklist=radeon quiet" and turned off graphical output in grub)
the result is that the dmesg is not *that* bad, but still not ok.

on VM start I get: (00:1d.0 is usb controller where I have keyboard and mouse, they work in the vm)
[  125.866335] dmar: DRHD: handling fault status reg 3
[  125.866400] dmar: DMAR:[DMA Read] Request device [00:1d.0] fault addr ef000
               DMAR:[fault reason 06] PTE Read access is not set
[  125.992123] kvm: zapping shadow pages for mmio generation wraparound
[  132.713767] br0: port 1(vnet0) entered forwarding state
[  134.660405] kvm [585]: vcpu1 unhandled rdmsr: 0x606
[  139.884802] kvm [585]: vcpu0 unhandled rdmsr: 0x611
[  139.884885] kvm [585]: vcpu0 unhandled rdmsr: 0x639
[  139.884959] kvm [585]: vcpu0 unhandled rdmsr: 0x641
[  139.885026] kvm [585]: vcpu0 unhandled rdmsr: 0x619

on vm shutdown, nothing happens, virsh list says it's "in shutdown" (forever )
if I do virsh destroy vmname it hangs.
then if I do virsh list, it hangs.
I need to shoot down the qemu process manually, then virsh list works again.
any ideas what's wrong?

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

update: I've tried windows again, from a snapshot without amd drivers.
I can boot it and have the gpu working (as generic VGA adapter), it even recognized correctly max resolution of the screen if I tried to change it (defualt was 800x600).  so far so good.
When I tried to install the driver, during "detecting hardware", I get BSOD 0xCA (nothing new in host dmesg at that time, boot time it gives the same errors as linux)

Last edited by ghormoon (2015-05-26 03:33:59)

aw

Member

Registered: 2013-10-04

Posts: 921

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

update: I've tried windows again, from a snapshot without amd drivers.
I can boot it and have the gpu working (as generic VGA adapter), it even recognized correctly max resolution of the screen if I tried to change it (defualt was 800x600).  so far so good.
When I tried to install the driver, during "detecting hardware", I get BSOD 0xCA

Use the driver specific to your hardware, don't go the detect driver path, even for updates via the catalyst control panel.

---

http://vfio.blogspot.com
Looking for a more open forum to discuss vfio related uses?  Try https://www.redhat.com/mailman/listinfo/vfio-users

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I'm not sure what do you mean. When I select the driver on the website manually, it still gives me the catalyst installer, that goes through the "detecting graphics hardware" step and there it crashes. I can't select anything in the installer manually. should I extract only the driver and don't install the rest or what?

Punkbob

Member

Registered: 2014-10-02

Posts: 14

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Just a heads up it looks like any Nvidia drivers past 347.88 are disabling themselves again, and spitting back code 43 errors.

toxster

Member

Registered: 2015-04-28

Posts: 5

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Hey guys

Introducing My next project, Virtualized OSX Yosemite (10.10.3) host with passthrough Nvidia GTX 970 running at 4K 60Hz..

Got everything working except USB passthrough of mouse (only apple keyboard works for some apparent reason)

Heaven ~68fps 720p with Extreme profile, 1080p ~40fps will benchmark with my win8.1 host later.

looks promising

Modo: Gigabyte X99-Gaming 5
CPU: Intel 5960x
Passthrough GPU: Asus Strix NVidia 970 GTX

Last edited by toxster (2015-05-26 10:33:54)

aw

Member

Registered: 2013-10-04

Posts: 921

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Just a heads up it looks like any Nvidia drivers past 347.88 are disabling themselves again, and spitting back code 43 errors.

Running 352.86 here, no Code 43 problems.  The NVIDIA Control Panel stopped working, but I see others reporting that in non-VM applications.  It worked on 350.12.  Code 43 is a pretty generic "something is wrong" error, hypervisor detection isn't the only reason for it to get thrown.  I recently discovered that failure to plugin aux power to cards also generates a Code 43.

---

http://vfio.blogspot.com
Looking for a more open forum to discuss vfio related uses?  Try https://www.redhat.com/mailman/listinfo/vfio-users

aw

Member

Registered: 2013-10-04

Posts: 921

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I'm not sure what do you mean. When I select the driver on the website manually, it still gives me the catalyst installer, that goes through the "detecting graphics hardware" step and there it crashes. I can't select anything in the installer manually. should I extract only the driver and don't install the rest or what?

Just trying to say that the "Automatically detect and install driver" code is known to cause BSODs, if you're doing the "Manually select your driver" path, then you're getting something else.

---

http://vfio.blogspot.com
Looking for a more open forum to discuss vfio related uses?  Try https://www.redhat.com/mailman/listinfo/vfio-users

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I don't get where you maen to do that. on the web site, I do manually select (notebook gpu > HD series > 6xxx> win 7 x64) and it gives me 288MB catalyst. that one only gives me default/custom selection and field for path. on next (either default or custom), it goes through few steps before allowing you to select things (at least on custom, default won't let you) and it crashes before that selection just after saying "detecting graphics hardware" in the progress bar. I can't find any way to get more specific driver than by the series 6xxx. What am I missing?

Duelist

Member

Registered: 2014-09-22

Posts: 358

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I don't get where you maen to do that. on the web site, I do manually select (notebook gpu > HD series > 6xxx> win 7 x64) and it gives me 288MB catalyst. that one only gives me default/custom selection and field for path. on next (either default or custom), it goes through few steps before allowing you to select things (at least on custom, default won't let you) and it crashes before that selection just after saying "detecting graphics hardware" in the progress bar. I can't find any way to get more specific driver than by the series 6xxx. What am I missing?

You're doing everything right, it just doesn't work somewhere else. Does dmesg say anything when the bsod happens?
And the bsod in question.. 0x000000CA you say? What are the other parameters? That's an interesting error actually. And try chopping off that 1d controller and plug in USB devices the other way, i'm not sure if it isn't causing some other weird things to happen.

Last edited by Duelist (2015-05-26 15:43:11)

---

The forum rules prohibit requesting support for distributions other than arch.
I gave up. It was too late.
What I was trying to do.
The reference about VFIO and KVM VGA passthrough.

ghormoon

Member

Registered: 2015-05-20

Posts: 17

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

not a single line in dmesg.
when I'll get home, I'll retry and get the missing bsod parameters.
currently 1d is only controller I can use, maybe I can not passthrough it and pass the usb devices one by one, that should be possible somehow, though I've not tried it yet
or maybe I can chop them off and try to install the driver over rdp? let's see

lordleto

Member

Registered: 2014-11-14

Posts: 6

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Mind sharing your xml/commandline?

Hey guys

Introducing My next project, Virtualized OSX Yosemite (10.10.3) host with passthrough Nvidia GTX 970 running at 4K 60Hz..

Got everything working except USB passthrough of mouse (only apple keyboard works for some apparent reason)

Heaven ~68fps 720p with Extreme profile, 1080p ~40fps will benchmark with my win8.1 host later.

looks promising

Modo: Gigabyte X99-Gaming 5
CPU: Intel 5960x
Passthrough GPU: Asus Strix NVidia 970 GTX

Punkbob

Member

Registered: 2014-10-02

Posts: 14

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Just a heads up it looks like any Nvidia drivers past 347.88 are disabling themselves again, and spitting back code 43 errors.

Running 352.86 here, no Code 43 problems.  The NVIDIA Control Panel stopped working, but I see others reporting that in non-VM applications.  It worked on 350.12.  Code 43 is a pretty generic "something is wrong" error, hypervisor detection isn't the only reason for it to get thrown.  I recently discovered that failure to plugin aux power to cards also generates a Code 43.

Hmm, I will look into it more then. Btw, are you running Windows 10 or 8.1, cause it might be an issue on my end as I am running my VMs with Windows 10.

aw

Member

Registered: 2013-10-04

Posts: 921

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Just a heads up it looks like any Nvidia drivers past 347.88 are disabling themselves again, and spitting back code 43 errors.

Running 352.86 here, no Code 43 problems.  The NVIDIA Control Panel stopped working, but I see others reporting that in non-VM applications.  It worked on 350.12.  Code 43 is a pretty generic "something is wrong" error, hypervisor detection isn't the only reason for it to get thrown.  I recently discovered that failure to plugin aux power to cards also generates a Code 43.

Hmm, I will look into it more then. Btw, are you running Windows 10 or 8.1, cause it might be an issue on my end as I am running my VMs with Windows 10.

8.1, AIUI 10 has numerous issues running as a VM

---

http://vfio.blogspot.com
Looking for a more open forum to discuss vfio related uses?  Try https://www.redhat.com/mailman/listinfo/vfio-users

toxster

Member

Registered: 2015-04-28

Posts: 5

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

Mind sharing your xml/commandline?

Sure, havent gotten around in attempting to forward the entire USB controller, supposedly doing so should work better.

Some comments to the XML below:

yoyo.iso is yosemite installer created with unibeast, end result is an installation that don't require any "hacked" kexts.
the direkt kernel /root/vm/boot is chameleon r2700 boot

<domain type='kvm' id='8' xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'>
  <name>osx</name>
  <uuid>ddb18084-f1d7-4041-b7be-9ab0437470c6</uuid>
  <memory unit='KiB'>8286208</memory>
  <currentMemory unit='KiB'>8286208</currentMemory>
  <vcpu placement='static'>4</vcpu>
  <resource>
    <partition>/machine</partition>
  </resource>
  <os>
    <type arch='x86_64' machine='pc-q35-2.3'>hvm</type>
    <kernel>/root/vm/boot</kernel>
    <boot dev='hd'/>
  </os>
  <features>
    <acpi/>
    <apic/>
    <pae/>
    <hyperv>
    </hyperv>
    <kvm>
      <hidden state='on'/>
    </kvm>
  </features>
  <cpu mode='custom' match='exact'>
    <model fallback='allow'>core2duo</model>
    <topology sockets='1' cores='4' threads='1'/>
  </cpu>
  <clock offset='utc'/>
  <on_poweroff>destroy</on_poweroff>
  <on_reboot>restart</on_reboot>
  <on_crash>restart</on_crash>
  <pm>
    <suspend-to-mem enabled='no'/>
    <suspend-to-disk enabled='no'/>
  </pm>
  <devices>
    <emulator>/usr/sbin/qemu-system-x86_64</emulator>
    <disk type='file' device='disk'>
      <driver name='qemu' type='qcow2'/>
      <source file='/cam/mac_hdd.img'/>
      <backingStore/>
      <target dev='sda' bus='sata'/>
      <alias name='sata0-0-0'/>
      <address type='drive' controller='0' bus='0' target='0' unit='0'/>
    </disk>
    <controller type='sata' index='0'>
      <alias name='sata0'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x1f' function='0x2'/>
    </controller>
    <controller type='pci' index='0' model='pcie-root'>
      <alias name='pcie.0'/>
    </controller>
    <controller type='pci' index='1' model='dmi-to-pci-bridge'>
      <alias name='pci.1'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x1e' function='0x0'/>
    </controller>
    <controller type='pci' index='2' model='pci-bridge'>
      <alias name='pci.2'/>
      <address type='pci' domain='0x0000' bus='0x01' slot='0x01' function='0x0'/>
    </controller>
    <controller type='usb' index='0'>
      <alias name='usb0'/>
      <address type='pci' domain='0x0000' bus='0x02' slot='0x01' function='0x0'/>
    </controller>
    <interface type='network'>
      <mac address='c8:85:50:7f:9c:bf'/>
      <source network='default' bridge='virbr0'/>
      <target dev='vnet0'/>
      <model type='e1000-82545em'/>
      <alias name='net0'/>
      <address type='pci' domain='0x0000' bus='0x02' slot='0x02' function='0x0'/>
    </interface>
    <hostdev mode='subsystem' type='usb' managed='yes'>
      <source>
        <vendor id='0x05ac'/> <!-- apple keyboard -->
        <product id='0x0250'/>
        <address bus='1' device='9'/>
      </source>
      <alias name='hostdev0'/>
    </hostdev>
    <memballoon model='none'>
      <alias name='balloon0'/>
    </memballoon>
  </devices>
  <seclabel type='none' model='none'/>
  <qemu:commandline>
    <qemu:arg value='-device'/>
    <qemu:arg value='isa-applesmc,osk=googleit!'/> <!-- read out from your mac or google it. -->
    <qemu:arg value='-smbios'/>
    <qemu:arg value='type=2'/>
    <qemu:arg value='-device'/>
    <qemu:arg value='ide-drive,bus=ide.1,drive=MacDVD'/>
    <qemu:arg value='-drive'/>
    <qemu:arg value='id=MacDVD,if=none,snapshot=on,file=/root/vm/yoyo.iso'/>
    <qemu:arg value='-device'/>
    <qemu:arg value='vfio-pci,host=03:00.0,x-vga=on'/>
    <qemu:arg value='-device'/>
    <qemu:arg value='vfio-pci,host=03:00.1'/>
    <qemu:arg value='-vga'/>
    <qemu:arg value='none'/>
  </qemu:commandline>
</domain>

Nesousx

Member

Registered: 2012-03-27

Posts: 46

Website

Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9

I don't get where you maen to do that. on the web site, I do manually select (notebook gpu > HD series > 6xxx> win 7 x64) and it gives me 288MB catalyst. that one only gives me default/custom selection and field for path. on next (either default or custom), it goes through few steps before allowing you to select things (at least on custom, default won't let you) and it crashes before that selection just after saying "detecting graphics hardware" in the progress bar. I can't find any way to get more specific driver than by the series 6xxx. What am I missing?

Have you tried manually installing the driver? Once downloaded, you only extract it somewhere on your disk but do not use wizard. Then from Windows Device Manager, you right clic on the graphic card (probably with the yellow warning sign), and install driver from here by pointing to the path were the drivers has been extracted.

I used to do it this way with my old setup (Xen + AMD cards), in order to avoid BSOD.