[vfio-users] Finding a suitable replacement mainboard for GPU passthrough

[Martin Schrodt]

On 09/28/2016 09:40 PM, Alex Williamson wrote:

>     2) It won't suspend correctly if I leave it's USB-3.1 controller to
>     Linux (will wake up immediately, complaining that the USB controller
>     prevented proper suspend) (ASMedia Technology Inc. ASM1142)
> 
> A poorly behaving ASMedia device, shocking </sarcasm> 

I bound it to vfio-pci, which lets me suspend the host properly.
The Windows VM likes the device too, so I'll leave it there.

>     3) The first and the second PCIe-slot are in the same IOMMU-group,
>     preventing me to use a second GPU for the host in slot 2 without
>     ACS-override. (*1)
> 
> Of course, you need a certain class of processors to have ACS, Z170
> doesn't support any of those processors.  See
> http://vfio.blogspot.com/2015/10/intel-processors-with-acs-support.html
>  You need an X99 motherboard for that.

Oh, thanks for the hint, I did not know this before. I did not want to
ACS-override anyway, and thought there would be mainboard layouts that
would separate those 2 slots even on Z170. Is this not the case?

>     5) The VM is dead after I suspend/resume the host.
> 
> Hmm, if you're thinking a VM with assigned devices should survive a host
> suspend/resume, I'm not sure that's a reasonable expectation right now.

Yea, I suspected as such. But I found this guy on github

https://github.com/qdel/scripts/tree/master/vfio/systemd

who claims he can do it by using systemd-hooks to suspend / resume the
VM before going to sleep / after waking up.

His scripts simply issue a

virsh dompmsuspend --domain win10 --target mem

and a

sudo virsh dompmwakeup win10

If I try to replicate this, it'll go to sleep but not wake up on the
second command. Maybe that's also because of the passed through ASMedia
Device, I didn't test yet.

And by the way: Thanks for all your work on all topics vfio, much
appreciated!