(SOLVED) KVM passthrough Geforce GT 750


chvb

Recommended Posts

Hello together,

 

i've got a problem with my GPU passthrough to my Windows 10 VM.

My Hardware is a HP Proliant ML310e. My Graphics Card is a Nvidia 750 with 2048MB RAM.

I can use only one PCIe Slot for this Card.

 

My Bios is the Version J04 11/09/2013. (newest)

I've read so much in this Forum and tested so much with PCIe ACS Override and the syslinux.cfg

 

This is the error Code when i try to start the VM with the Nvidia 750 GPU:

 

internal error: process exited while connecting to monitor: 2016-04-25T09:04:53.998337Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x4: vfio: error, group 1 is not viable, please ensure all devices within the iommu_group are bound to their vfio bus driver.

2016-04-25T09:04:53.998362Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x4: vfio: failed to get group 1

2016-04-25T09:04:53.998369Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x4: Device initialization failed

 

 

I'm using this in the syslinux.cfg:

append intel_iommu=on vfio_iommu_type1.allow_unsafe_interrupts=1 pcie_acs_override=downstream initrd=/bzroot

 

Also tested PCIe ACS Override enabled/disabled.

My onboard Graphic Card is enabled. So the Console from Unraid goes to this Adapter.

 

The GPU is shown under devices.

05:00.0 VGA compatible controller [0300]: NVIDIA Corporation GM107 [GeForce GTX 750] [10de:1381] (rev a2)

05:00.1 Audio device [0403]: NVIDIA Corporation Device [10de:0fbc] (rev a1)

 

no other devices are using the Group of 05:00.0 and 05:00.1 when i use PCIe ACS Override enabled.

 

What can i try?

Many thanks.

Link to comment

This might be the showstopper:

 

Apr 25 11:48:31 NAS kernel: pci 0000:05:00.1: Device is ineligible for IOMMU domain attach due to platform RMRR requirement.  Contact your platform vendor.

 

Seems there might be some problems with RMRR (memory reporting regions).

Is it possible for you to try to Disable HP Shared Memory features in system configuration and under the nvidia 750?

 

You could also try setting iommu=nopt in the syslinux.cfg, but not sure it would help.

You could try to remove the audio part of the nvidia card and see if it allows you to pass through only the GPU.

Link to comment

Hello,

 

i've got wrong. there is no option in the BIOS of the ML310e with HP Shared Memory features.

The only option with memory is this: No-Execute Memory Protection (it is enabled)

 

 

i have set the setting iommu=nopt in the syslinux.cfg with no success.

 

Any other ideas?

This is frustraded. I want to use the Nvidia Card for GPU encoding.

 

 

This is the current message, when i try to start the VM:

 

internal error: process exited while connecting to monitor: 2016-04-26T10:51:38.882123Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to set iommu for container: Operation not permitted

2016-04-26T10:51:38.882149Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to setup container for group 12

2016-04-26T10:51:38.882154Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to get group 12

2016-04-26T10:51:38.882163Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: Device initialization failed

 

 

Just for information. there is no Monitor connected. But i think, this isn't the problem?

As i told, i would like to use the nvidie for hardware encoding.

Link to comment

Not sure what else to suggest as you have done all the things I know about that should fix the problem you have.

Could you try to see if it works with unraid 6.1.9? You don't need to format your current USB, just use a new USB stick and use a trial to check.

Link to comment

i also use the docker function.

i have deactivate the emby docker.

when i try to start the VM the following error is shown:

 

 

internal error: early end of file from monitor, possible problem: 2016-04-26T17:32:34.961905Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to set iommu for container: Operation not permitted

2016-04-26T17:32:34.961934Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to setup container for group 12

2016-04-26T17:32:34.961939Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: vfio: failed to get group 12

2016-04-26T17:32:34.961947Z qemu-system-x86_64: -device vfio-pci,host=05:00.0,id=hostdev0,x-vga=on,bus=pci.2,addr=0x5: Device initialization failed

Link to comment

I wanted to chime in on this thread because this is definitely an oddball and the first time I've seen an RMRR message on a GTX 7xx GPU.  That said, RMRR errors like this aren't a good sign, though I do have one more thing for you to try before we have to give up hope.  First, if you want some light-reading on the subject, this RedHat article is pretty comprehensive and specifically covers assigning PCI devices to VMs in this situation.

 

Now with the boring stuff out of the way, the last thing I would ask you to try is a change the PCIe ACS Override option in the syslinux.cfg.  The way this normally looks is as such:

 

append pcie_acs_override=downstream initrd=/bzroot

 

I want you to change that to this:

 

append pcie_acs_override=id:10de:1381,10de:0fbc initrd=/bzroot

 

Don't forget to reboot your system after applying this change.

 

This is different than the vfio-pci.ids thing you tried before.  Please report back if this changes anything for you.

Link to comment

Hello,

 

thanks for the info.

 

I've tested the changes in the sysconfig with reboot.

But also no success.

 

I have also tested the card in another HP Server (ML310e V2)

There i'am using the Unraid Version 6.1.9.

Same Problems here.

 

in the Bios settings it is possible to change the irq of the Card.

But when i change the number, some other components would be changed automatically.

Other Components are in the same group of the GPU.

 

Maybe it is a Problem of HP. As i know, there is a big Problem with pci passthrough since version 5.1 U2.

The latest Version without issues was the 5.1U1.

 

Thanks for any more ideas.

Link to comment

Hello together,

 

now i tested another Graphic Card in the HP Server.

I've testet with a Radeon HD5000.

Same errors here. So it is a HP issue.

 

 

Next step:

 

Install ESX Server. Passthrough the LSI Controller and  TV Tuners to UNRAID.

Create a Win10 Machine and Passthrough the Nvidia Card.

I hope this solution will work for me.

 

Link to comment

Hello together,

 

now I tested with the ESXi 6 U2 Server.

I habe passthrough my Digital Devices Devices, the onboard Controller and the IT flashed LSI 9211 (Dell perc H310) to the UNRAID VM.

I've got a problem with the Passthrough of the LSI Controller.

The Controller is show up, but there are no drives.

 

With the onboard Intel Controller the other devices are shown up.

Now i Also passthrough my Nvidia Card to a Windows VM which works with ESXi.

 

Does anybody know, why there are no disks show up with the passthrough LSI Controller?

When i start Unraid without ESXi, there are no Problems. All disks showed up.

 

Thanks

Link to comment

Hello together,

 

finally, i got UNRAID working under the ESXi 6 Server with my HP ML310e Server.

 

the Solution:

 

Add in the syslinux.cfg

 

mpt3sas.msix_disable=1

 

 

so all my Disks on the LSI controller are visible.

I got passthrough my 2 Digital Devices Devices and the LSI Controller.

 

Finally, i created a Windows VM with a passthrough of my Geforce GTX 750 for Hardware encoding.

 

Now i will test if it works stable.

 

 

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.