[vfio-users] No available IOMMU models

[Mark Nipper]

	Ran into a weird problem with my vfio / gpu passthrough
last night.  I was playing a game under a Windows 10 Pro guest
when suddenly the game seemed to freeze, I started to get the
blue spinning wheel pointer from Windows indicating that the
application wasn't responding suddenly, and then everything went
dark.  I tried switching back to my Linux host and couldn't get
it to come up either.  After investigating further on the
network, the host seemed like it was down completely.

	So after a physical reset of the machine, everything came
up under Linux and I tried starting up my guest again, and now
I'm getting the following:
---
qemu-system-x86_64: -device vfio-pci,host=02:00.0,bus=root,addr=00.0,multifunction=on,x-vga=on: vfio: No available IOMMU models
qemu-system-x86_64: -device vfio-pci,host=02:00.0,bus=root,addr=00.0,multifunction=on,x-vga=on: vfio: failed to setup container for group 18
qemu-system-x86_64: -device vfio-pci,host=02:00.0,bus=root,addr=00.0,multifunction=on,x-vga=on: vfio: failed to get group 18
qemu-system-x86_64: -device vfio-pci,host=02:00.0,bus=root,addr=00.0,multifunction=on,x-vga=on: Device initialization failed
qemu-system-x86_64: -device vfio-pci,host=02:00.0,bus=root,addr=00.0,multifunction=on,x-vga=on: Device 'vfio-pci' could not be initialized

I'm following the Debian testing branch at the moment, so I
started checking logs to see if my kernel or qemu version had
changed recently via a package update, and didn't see anything.
I also couldn't find anything useful in any logs to see what
might have gone wrong in the first place.  So I pulled the
primary video card, just to make sure the secondary was still
working correctly, and the Linux host booted up on the previously
secondary / now primary card with no problem.  However, after
putting the old primary back in, I'm still getting the above
errors.

	So I'm not entirely sure what has gone wrong at this
point.  I still see vfio-pci bound to the secondary card:
---
02:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK104 [GeForce GTX 670] [10de:1189] (rev a1)
        Subsystem: NVIDIA Corporation Device [10de:097a]
        Kernel driver in use: vfio-pci
02:00.1 Audio device [0403]: NVIDIA Corporation GK104 HDMI Audio Controller [10de:0e0a] (rev a1)
        Subsystem: NVIDIA Corporation Device [10de:097a]
        Kernel driver in use: vfio-pci

and I still see the secondary card as the only member of its
IOMMU group:
---
/sys/kernel/iommu_groups/18/devices/0000:02:00.0
/sys/kernel/iommu_groups/18/devices/0000:02:00.1

And like I said, nothing has really changed otherwise.  So what
exactly has gone wrong here?  I upgraded to 4.2-rc8 just to try
something different and I'm still getting the errors.

	Any ideas?

-- 
Mark Nipper
nipsy at bitgnome.net (XMPP)
+1 979 575 3193
-
"Computers are like Old Testament gods: lots of rules and no mercy."
 -- Joseph Campbell, _The Power of Myth_, 1988