Re: [vfio-users] Low power graphics card that plays nice with vfio?

[Nick S]

How low do you want to go? I had absolutely no problems with AMD R7 240
cards. You can find some fanless options for it.

2018-01-10 10:33 GMT-08:00 taii...@gmx.com :

> I am looking for a low power fanless/low price graphics card that can be
> assigned to a VM without trouble.
>
> I am also wondering as to if there are any normal graphics cards with FLR
> (normal as in not the ones intended for RDP that have only a single display
> connector)
>
> ___
> vfio-users mailing list
> vfio-users@redhat.com
> https://www.redhat.com/mailman/listinfo/vfio-users
>
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

Re: [vfio-users] AMD reset bug on rx 480

[Nick S]

I think I was able to mitigate the issue by ejecting the GPU and switching
to qemu emulated graphics before the shutdown and then reverting this
process after a restart. I am fairly sure it all started with 17.12.02
version of Radeon drivers. 17.10.02 seems to be working fine. Did we get
the reset bug back via a software upgraded and even on the cards where it
did not happen before?

2017-12-26 14:06 GMT-08:00 Nick S <nick.kv...@gmail.com>:

> I have several RX 480 cards that I use to run virtual machines and one of
> them is getting bugged intermittently after several virtual machine
> restarts. When it happens, the screen either turns off or turns light grey
> and Windows 7 or 10 just BSOD repeatedly when starting. The only thing that
> helps consistently is restarting the entire PC. Once I was able to fix the
> card by doing the following:
>
> 1) Pull and reinsert the PCIE 6 pin power cable from it
> 2) Unbind card from vfio driver
> 3) echo "1" > /sys/bus/pci/devices/MY:CARD:NUMBER/remove
> 4) echo "1" > /sys/bus/pci/rescan
>
> Just steps 2 through 4 do not seem to help and I am a bit hesitant messing
> with cables on a running PC.
>
> Question: is there a way to reset the card without a restart? I get this
> situation fairly infrequently, but still it is a bit annoying. Also, would
> it make sense to add my card to pci-quirks.c where it handles older cards?
> There is no polaris cards there as far as I can see.
>
>
>
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

[vfio-users] AMD reset bug on rx 480

[Nick S]

I have several RX 480 cards that I use to run virtual machines and one of
them is getting bugged intermittently after several virtual machine
restarts. When it happens, the screen either turns off or turns light grey
and Windows 7 or 10 just BSOD repeatedly when starting. The only thing that
helps consistently is restarting the entire PC. Once I was able to fix the
card by doing the following:

1) Pull and reinsert the PCIE 6 pin power cable from it
2) Unbind card from vfio driver
3) echo "1" > /sys/bus/pci/devices/MY:CARD:NUMBER/remove
4) echo "1" > /sys/bus/pci/rescan

Just steps 2 through 4 do not seem to help and I am a bit hesitant messing
with cables on a running PC.

Question: is there a way to reset the card without a restart? I get this
situation fairly infrequently, but still it is a bit annoying. Also, would
it make sense to add my card to pci-quirks.c where it handles older cards?
There is no polaris cards there as far as I can see.
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

Re: [vfio-users] Using a physical disk in QEMU

[Nick S]

Can you at least boot via this UEFI screen if you select the necessary
device? or does it complain? I don't remember doing anything special for
virsh and for command line I had to provide two OVMF files as separate
devices:
-drive if=pflash,format=raw,readonly,file=$folder/OVMF_CODE-pure-efi.fd \
-drive if=pflash,format=raw,file=$folder/OVMF_VARS-pure-efi.fd \


On Sat, Oct 21, 2017 at 3:36 PM, Patrick O'Callaghan <p...@usb.ve> wrote:

> On Sat, 2017-10-21 at 09:52 -0700, Nick S wrote:
> > If that helps, I am using a USB SSD for Win10 and I installed a USB
> > bootable version so that I can boot it both under QEMU and directly.
> > Really, the only difference is that I put the device alias from
> > /dev/disk/by-id/ instead of the file path in the quemu command line (or
> in
> > virsh as the source path).
> >
>
> OK, I did that and the system dual-boots correctly. However the KVM
> boot now lands me in the UEFI screen. I've seen this before and got it
> to work but it was a while ago and I've lost the reference to the magic
> incantation. A hint would be appreciated.
>
> poc
>
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

Re: [vfio-users] Using a physical disk in QEMU

[Nick S]

If that helps, I am using a USB SSD for Win10 and I installed a USB
bootable version so that I can boot it both under QEMU and directly.
Really, the only difference is that I put the device alias from
/dev/disk/by-id/ instead of the file path in the quemu command line (or in
virsh as the source path).

On Sat, Oct 21, 2017 at 3:07 AM, Patrick O'Callaghan  wrote:

> I'm about to install a SATA drive for my Win10 VM, and relocate the
> Windows system from the current Linux raw file to the new drive. I'd
> like to do this in such a way as to leave open the option of dual-
> booting Windows from the new drive. Is this possible?
>
> poc
>
> ___
> vfio-users mailing list
> vfio-users@redhat.com
> https://www.redhat.com/mailman/listinfo/vfio-users
>
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

[vfio-users] was the issue with eventual freeze when assigning GPU audio to guests fixed

[Nick S]

Hi Alex, all,

I have a question related to this issue:
https://www.redhat.com/archives/vfio-users/2016-February/msg00061.html

I wonder whether it is still present in Kernel 4.10. I think I am
experiencing now something similar to this but on a much newer kernel than
what is mentioned in that thread.

Thank you!
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

Re: [vfio-users] RX 480 vga passthrough 3D mode host hangs

[Nick S]

I can confirm that it works with no issues with UEFI. I recently replaced
R9 270 by RX 480 and it even worked with not issues after I forgot to
update/remove the old ROM file for R9. Here is my command line. It uses a
few variables for multiple GPU but you should get the base picture. I have
both Windows 7 and WIndows 10 VMs and all of them work fine.

qemu-system-x86_64 -enable-kvm -cpu host -m $memoryarg -smp
sockets=1,cores=$corearg,threads=2 \
-serial none \
-ctrl-grab -no-quit \
-parallel none \
$soundarg \
-name passthrough-$1 \
-rtc base=localtime \
-drive if=pflash,format=raw,readonly,file=$folder/OVMF_CODE-pure-efi.fd \
-drive if=pflash,format=raw,file=$folder/OVMF_VARS-pure-efi.fd \
-netdev bridge,id=hn0 -device e1000,netdev=hn0,id=nic1,mac=$macaddr \
-drive file=$folder/W7_UEFI.qcow2,id=d0,if=none -device
ide-hd,drive=d0,serial=$hddserial \
-device vfio-pci,host=$vgapci \
-device vfio-pci,host=$audiopci


On Tue, Aug 22, 2017 at 8:22 AM, Quentin Deldycke  wrote:

> Hi,
>
> Ovmf makes the process more easy. I don't remember exactly why, but before
> i was using a AMD HD 5870. Using old style bioses as it was not uefi. When
> i switched to the r9 290, i had to switch to uefi so ovmf.
>
> I experienced issues with the integrated sound card of my r9 (was all time
> default on windows...). I have the sound device linked to vfio-pci but
> never pass it to the virtual machine, you can possibly try like this also.
>
> For a more easy process i never disable the intergrated intel gpu of my
> cpu. It used plain old vga for bios post and cie.
>
>
>
> --
> Deldycke Quentin
>
>
> On 22 August 2017 at 16:57, Jiri Novak  wrote:
>
>> Hi,
>>
>> yes, not using OVMF, maybe that will fix it, not sure, I can try.
>> I don't have any R9 card, I actually did have one before, but now I have
>> R7 240 and RX 480 at hand. R9 did work fine before, like the R7 does.
>> I pass in two (three) devices, one is the card (plus the integrated
>> soundcard, you can't pass it without, right? it's the same iommu group
>> always) and a USB controller to be able to connect kb+mouse.
>> for the RX, the romfile has no effect obviously, though for the R7 it
>> won't start without it (R9 didn't care either).
>> no isolation so far, just taskset so it won't roam across numa nodes.
>> might add later.
>>
>> my boot card is nvidia, so I have radeon and amdgpu blacklisted. also
>> tried disabling "boot option rom" on the pcie ports so bios of the host
>> doesn't try to initialize them. doesn't have effect, except that if it is
>> allowed and the order of the cards is that amd is sooner, boot display will
>> go to it. if disabled, it will go to the nvidia (I keep that one enabled at
>> all cases, if I disable all, I'm doomed - bios reset needeed :))
>>
>> I think my only chance now is ovmf, I have no other idea.
>>
>> lspci for reference:
>>
>> 00:00.0 Host bridge: Intel Corporation Xeon E5/Core i7 DMI2 (rev 07)
>> 00:01.0 PCI bridge: Intel Corporation Xeon E5/Core i7 IIO PCI Express
>> Root Port 1a (rev 07)
>> 00:02.0 PCI bridge: Intel Corporation Xeon E5/Core i7 IIO PCI Express
>> Root Port 2a (rev 07)
>> 00:03.0 PCI bridge: Intel Corporation Xeon E5/Core i7 IIO PCI Express
>> Root Port 3a in PCI Express Mode (rev 07)
>> 00:04.0 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 0 (rev 07)
>> 00:04.1 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 1 (rev 07)
>> 00:04.2 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 2 (rev 07)
>> 00:04.3 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 3 (rev 07)
>> 00:04.4 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 4 (rev 07)
>> 00:04.5 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 5 (rev 07)
>> 00:04.6 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 6 (rev 07)
>> 00:04.7 System peripheral: Intel Corporation Xeon E5/Core i7 DMA Channel
>> 7 (rev 07)
>> 00:05.0 System peripheral: Intel Corporation Xeon E5/Core i7 Address Map,
>> VTd_Misc, System Management (rev 07)
>> 00:05.2 System peripheral: Intel Corporation Xeon E5/Core i7 Control
>> Status and Global Errors (rev 07)
>> 00:05.4 PIC: Intel Corporation Xeon E5/Core i7 I/O APIC (rev 07)
>> 00:11.0 PCI bridge: Intel Corporation C600/X79 series chipset PCI Express
>> Virtual Root Port (rev 06)
>> 00:16.0 Communication controller: Intel Corporation C600/X79 series
>> chipset MEI Controller #1 (rev 05)
>> 00:16.1 Communication controller: Intel Corporation C600/X79 series
>> chipset MEI Controller #2 (rev 05)
>> 00:1a.0 USB controller: Intel Corporation C600/X79 series chipset USB2
>> Enhanced Host Controller #2 (rev 06)
>> 00:1b.0 Audio device: Intel Corporation C600/X79 series chipset High
>> Definition Audio Controller (rev 06)
>> 00:1c.0 PCI bridge: Intel Corporation C600/X79 series chipset PCI Express
>> Root Port 1 (rev b6)
>> 00:1c.5 PCI bridge: Intel 

[vfio-users] How would you estimate my probability of a successful passthrough on this hardware?

[Nick S]

Hi vfio-users,

I am looking for expert opinions whether I will be able to successfully
passthrough a GPU on the following hardware:

1) CPU: i7-3770 (not K)
2) Motherboard: ga-b75m-d3h Gigabyte that claims to support VT-x and VT-d
but I was not able to find any success stories
3) Using Intel Graphics for host and either AMD or Nvidia GPU for guest. I
have plenty of options here, aiming for OVMF setup may be with Nvidia
1060gtx.

I already have a setup based on a Haswell-E CPU and trying to set up a
second PC on some older hardware I have. As I understand, this setup will
require ACS override that I did not have to apply for my Haswell-E
configuration. So, a couple of questions if it looks feasible:

1) What is the best place to find the ACS override patch for 4.10 kernel?
2) Should I expect any other complications compared to a relatively
straight forward Haswell-E setup where I also have OVMF guests and BIOS
host?

Thank you,
Nick
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

Re: [vfio-users] Looking for recommendations for screen-switching

[Nick S]

Monitor source selection menu is the cheapest option and also the least
convenient. The most convenient option is a KVM switch but then you would
have to deal with passing one of its USB ports to the VM while keeping the
other one on host and a good KVM will cost you several hundred bucks. The
way I am using it:
1) Cheap HDMI switcher (i.e. something like this:
https://www.amazon.com/TBS%C2%AE2083-Intelligent-Switcher-Adapter-Splitter/product-reviews/B00M6I8K2E).
It allows me to switch the cards without fumbling with the monitor menu.
2) I keep the emulated VGA as a second adapter on my virtual machines. For
Windows 7 it keeps a window on the desktop that I can use to grab or
release focus for keyboard and mouse. This way I don't have to deal with
any USB passthrough - only the GPU is passed to the VM.

On Sun, Mar 19, 2017 at 7:14 AM, Patrick O'Callaghan  wrote:

> I have VFIO passthrough working with a Windows 10 VM, running on Fedora
> 25. Currently I have two cables running to my single monitor (one HDMI
> and one DVI), so switching back and forth means toggling a monitor
> setting. This does work, but is a bit clunky.
>
> BTW the mouse and keyboard also switch back and forth (this slightly
> surprised me but I'm not going to question it :-)
>
> I thought I could make this simpler by exchanging the DVI cable for
> another HDMI cable, both cables going into an HDMI switch and hence to
> the HDMI port on the monitor, with the idea of toggling back and forth.
> However this doesn't entirely work. I suspect that the switch powers
> off the input it's not using and this confuses Windows.
>
> Most of the discussion I've seen recommends Synergy, a software
> KB/mouse switcher, however this explicitly does not support video
> switching, i.e. it assumes you have a separate monitor for each machine
> (real or virtual), which is not the case here.
>
> The other option seems to be a physical KVM switch, but before shelling
> out more money I'd like to hear people's thoughts on this. What do you
> do to support a single monitor?
>
> poc
>
> ___
> vfio-users mailing list
> vfio-users@redhat.com
> https://www.redhat.com/mailman/listinfo/vfio-users
>
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

[vfio-users] GPU passtrough for R9 290x2, is it not possible?

[Nick S]

I remember seing somewhere that dual GPU cards are not working with the
passthrough. Is it stilll the case or did somebody manage to get it working?

Thanks!
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users

[vfio-users] I am confused about Intel GPU status

[Nick S]

Hi guys,

I remember reading that to use Intel GPU on your host you need to patch
your kernel. Is it still required? The reason I am asking is that I watched
the "30k PC" video from Alex blog and it looks like they are using
motherboard GPU for host there with no issues. Also, is it required for
OVMF virtual machines? I am trying to chose another motherboard for my
virtualization PC and I would like to maximize the PCIE slots available for
passthrough.

Thank you,
Nick
___
vfio-users mailing list
vfio-users@redhat.com
https://www.redhat.com/mailman/listinfo/vfio-users