I actually used that tutorial to set it up and I can't find anything that I 
might have missed. I have unfortunately returned the card by now while I still 
had time so I can't test it further. It would be really nice if we could 
somehow force vfio to do a reset so that we would know if it is even possible 
for a card. Is there any cheap (perferably AMD) card that I can get which is 
known to downright work properly with vfio reset?

________________________________
From: Nicolas Roy-Renaud <nicolas.roy-renau...@ens.etsmtl.ca>
Sent: 05 July 2016 07:20 PM
To: Gerhard de Clercq
Cc: vfio-users
Subject: Re: [vfio-users] No reset on GT730


Hmm... It's hard to tell from here if this is due to a misconfiguration or from 
an actual qemu issue. Would you mind checking sections 1, 2 and 3 of this 
tutorial<https://wiki.archlinux.org/index.php/PCI_passthrough_via_OVMF> to make 
sure that your system is properly configured? Section 4 is geared more towards 
libvirt, while you appear to be using qemu so you can skip that one. I get that 
you have no issues with your VM as of now aside from that reboot problem, but 
for all we know, it might just be some driver override issue.


- Nicolas

On 2016-07-05 02:05, Gerhard de Clercq wrote:

Ok thanks for the info. There is nothing else suspicious in dmesg and Qemu 
returned no error messages. I do believe the error stream is working though 
because I did get error when I forgot to attach my new card vfio-pci.

________________________________
From: Nicolas Roy-Renaud 
<nicolas.roy-renau...@ens.etsmtl.ca><mailto:nicolas.roy-renau...@ens.etsmtl.ca>
Sent: 04 July 2016 11:49:36 PM
To: Gerhard de Clercq
Cc: vfio-users
Subject: Re: [vfio-users] No reset on GT730


You can't really reset a card using SeaBIOS because it uses VGA arbitration, 
which has little to do with how an actual PCI passthrough would work IIRC. What 
makes you so sure that the issue comes from the gpu not resetting, is dmesg 
returning anything suspicious or is your qemu command completing successfully 
after you shutdown your VM? If you're not using libvirt, qemu is supposed to 
log error messages in the standard error stream instead, does anything abnormal 
come up when the VM shuts down?


- Nicolas

On 2016-07-04 15:46, Gerhard de Clercq wrote:

I was actually running though command line so no libvirt log and I didn't see 
anything in journalctl though I'm not entirely sure what to look for. I do 
indeed pass some other devices. I also saw in dmesg dat vfio-pci was warning 
about the ROM being invalid when I tried to start a VM the second or later time 
but passing a vgabios explicitly didn't make any difference. Here is my command:


sudo qemu-system-x86_64 -enable-kvm -M q35 -m 4G -cpu 
core2duo,hv_time,hv_relaxed,hv_vapic,hv_spinlocks=0x1fff,kvm=off,hv_vendor_id=badapple
  -smp 4,sockets=1,cores=2,threads=2  -bios OVMF.fd -device 
ioh3420,bus=pcie.0,addr=1c.0,multifunction=on,port=1,chassis=1,id=root.1 
-device i82801b11-bridge,id=pci.1,bus=pcie.0,addr=0x1e -device 
pci-bridge,chassis_nr=2,id=pci.2,bus=pci.1,addr=0x1 -device 
piix3-usb-uhci,id=usb,bus=pci.2,addr=0x1 -device 
ahci,id=sata0,bus=pci.2,addr=0x5 -drive 
file=./Untitled.iso,if=none,id=drive-sata0-0-2,format=raw -device 
ide-hd,bus=ide.2,drive=drive-sata0-0-2,id=sata0-0-2,bootindex=1 -drive 
file=./nomachine.qcow2,if=none,media=disk,id=drive-sata0-0-0,format=qcow2 
-device ide-hd,bus=sata0.0,drive=drive-sata0-0-0,id=sata0-0-0 -smbios type=2 
-serial stdio -device vfio-pci,host=05:00.0,bus=pcie.0,multifunction=on -device 
vfio-pci,host=05:00.1,bus=pcie.0 -netdev 
tap,id=net0,ifname=tap0,script=no,downscript=no -device 
e1000-82545em,netdev=net0,id=net0,mac=52:54:00:c9:18:27 -usb -device 
usb-host,hostbus=3,hostaddr=3 -vga none

Also, I was wondering, is it definitely impossible to reset a card when using 
vfio-vga and seabios?
________________________________
From: Nicolas Roy-Renaud 
<nicolas.roy-renau...@ens.etsmtl.ca><mailto:nicolas.roy-renau...@ens.etsmtl.ca>
Sent: 04 July 2016 07:18:12 PM
To: Gerhard de Clercq
Cc: vfio-users
Subject: Re: [vfio-users] No reset on GT730


Are you passing through any other PCI devices, like USB controllers or 
soundcards? I'm guessing you've already checked your vm logs in 
/var/log/libvirt/qemu and the daemon logs in journalctl?


- Nicolas

On 2016-07-03 12:33, Gerhard de Clercq wrote:

I am trying to set up PCI passthrough of my Nvidia GT730 GPU on a fully up to 
date Fedora 24 system. I have managed to get it working but the problem is that 
it only works once after booting the host and to run again I need to reboot the 
host. I have determined that this is most likely caused by the fact that the 
GPU is not being reset. I bought this GPU because it has UEFI support after I 
could not get another one working with BIOS. It appeared as if though only AMD 
GPUs had a reset problem. Is this a known problem with this GPU and is there 
any way that I can try to force a reset?



_______________________________________________
vfio-users mailing list
vfio-users@redhat.com<mailto: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

Reply via email to