IGD Passthrough doesn't work after upgrade

coyote

New Member
Dec 4, 2018
29
0
1
36
Ok, will try that in the afternoon and also i will update Bios. Hope it works with an older Kernel (or Bios Update and latest Kernel).
It drives me crazy o_O

And again, thanks for your patience;)
 

lckmt

New Member
Dec 5, 2018
8
0
1
Code:
romfile=undefined
is always set by newest proxmox gui, maybe this should be fixed if no romfile is defined.

this cannot work, please remove the 'romfile=undefined' from your config (except you put a romfile in /usr/share/kvm/undefined )
also you should add 'x-vga=on' so that the qemu gpu gets deactivated and the intel will be the primary

what you can do further is to boot an older kernel and see if it fixes that, or upgrade your bios/firmware of the nuc
 
Last edited:

coyote

New Member
Dec 4, 2018
29
0
1
36
Yesterday i spended a few hours to do some tests.
These are my findings, i hope i can explain, what i found out:

- Update Bios to the latest Version -> same result, no luck
- booted the oldest proxmox Kernel installed (14.3.16-2) -> same result, no luck

Next test, booted new Mint VM without installing Mint on Harddisk.
Booted just the live Version with config:

Code:
hostpci0: 00:02.0,x-vga=1,rombar=0
Now Mint booted and i see the desktop on my HDMI Monitor. But it booted not
correctly, because there is no start menu. Just mouse pointer, Firefox, Terminal,
File Explorer icon and the desktop background.
I stopped the VM and boot it again. Same result.
Again, stop and boot it again. VM comes up with correct Desktop.
Menu, Icons and Dektop are there. Stop again and start it.
Now, it stucks on boot up.

Ok, i try it again with config:

Code:
hostpci0: 00:02.0
to only passthrough the gpu to mint VM.
Open console, i see mint bootlogo. Than a black screen with a blinking cursor and it
stucks on this screen.

I tried this again with my old Mint VM and it also stucks on this screen

photo_2018-12-07_06-23-13.jpg

If i delete the pci device, VM booted up fine.
I tried the old Mint VM with config:

Code:
hostpci0: 00:02.0,x-vga=1,rombar=0
(same config as the test VM)

No boot logo, no Desktop. I think it stucks on boot, because teamviewer is not showing the VM.
Tried again with:
Code:
hostpci0: 00:02.0
VM stucks on boot, only if i choose Mint VM in recovery mode, it booted up.


Is it possible, that something from the old guide doing this?
https://forum.proxmox.com/threads/guide-intel-intergrated-graphic-passthrough.30451/

I bought my NUC in August 2018, i think i have installed proxmox iso 5.2.1 as first install.
And there it works fine so far
 

dcsapak

Proxmox Staff Member
Staff member
Feb 1, 2016
3,798
345
83
31
Vienna
is always set by newest proxmox gui, maybe this should be fixed if no romfile is defined.
yes this is a bug, for which i already sent a patch to the list

@coyote

from your post, it seems that it works in general (with a new vm) but not with the old vm
i guess the reason for that is that the pci addresses changed since the last version (see one of my previous posts where i mentioned that)

so i guess you have 2 options:

try to repair you existing mint installation to use the correct pci paths ( e.g. they could be in /etc/X11/xorg.conf if it uses X; no idea for wayland though)
or you install a new instance of mint and migrate your data there
 

coyote

New Member
Dec 4, 2018
29
0
1
36
No, it doesn't work with a new installed VM. Only if i boot the live image, i see the Desktop on the display, but not correctly. There isn't a start menu.
If i install it to the harddisk it doesn't boot up.....and i don't have a signal on hdmi.
Something doesn't work with the passing through.

I have tried 3 different, fresh installed VM's. VM doesn't boot....

It would not be a problem if a new VM were running, I just need the web browser and the usb touchscreen for my home automation
 

dcsapak

Proxmox Staff Member
Staff member
Feb 1, 2016
3,798
345
83
31
Vienna
Again, stop and boot it again. VM comes up with correct Desktop.
it has worked at least once in this configuration
maybe your iso is corrupt ?

does anything relevant show in the syslog ? (/var/log/syslog or journalctl )
 

coyote

New Member
Dec 4, 2018
29
0
1
36
it has worked at least once in this configuration
Only one shot and that's exactly what I don't understand...
I tried many, many times to get this to work yesterday...

maybe your iso is corrupt ?
I don't think so, i tried 3 different iso files. Mint19Mate, Mint19.1Cinnamon and Debian Desktop.

journalctl after starting VM:

Code:
Dec 07 10:27:02 pve pvedaemon[6053]: <root@pam> update VM 106: -hostpci0 00:02.0,x-vga=1,romfile=undefined
Dec 07 10:27:30 pve pvedaemon[14253]: <root@pam> starting task UPID:pve:000063E6:00511605:5C0A3D02:qmstart:106:root@pam:
Dec 07 10:27:30 pve pvedaemon[25574]: start VM 106: UPID:pve:000063E6:00511605:5C0A3D02:qmstart:106:root@pam:
Dec 07 10:27:30 pve systemd[1]: Started 106.scope.
Dec 07 10:27:30 pve systemd-udevd[25588]: Could not generate persistent MAC address for tap106i0: No such file or directory
Dec 07 10:27:31 pve kernel: device tap106i0 entered promiscuous mode
Dec 07 10:27:31 pve kernel: vmbr0: port 8(tap106i0) entered blocking state
Dec 07 10:27:31 pve kernel: vmbr0: port 8(tap106i0) entered disabled state
Dec 07 10:27:31 pve kernel: vmbr0: port 8(tap106i0) entered blocking state
Dec 07 10:27:31 pve kernel: vmbr0: port 8(tap106i0) entered forwarding state
Dec 07 10:27:37 pve kernel: vfio-pci 0000:00:02.0: enabling device (0400 -> 0403)
Dec 07 10:27:37 pve kernel: vfio_ecap_init: 0000:00:02.0 hiding ecap 0x1b@0x100
Dec 07 10:27:40 pve pvedaemon[14253]: <root@pam> end task UPID:pve:000063E6:00511605:5C0A3D02:qmstart:106:root@pam: OK
Dec 07 10:28:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec 07 10:28:01 pve systemd[1]: Started Proxmox VE replication runner.
Dec 07 10:29:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec 07 10:29:01 pve systemd[1]: Started Proxmox VE replication runner.
Dec 07 10:30:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec 07 10:30:01 pve systemd[1]: Started Proxmox VE replication runner.
Dec 07 10:30:03 pve pvedaemon[14253]: <root@pam> successful auth for user 'root@pam'
Dec 07 10:31:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec 07 10:31:01 pve systemd[1]: Started Proxmox VE replication runner.
lines 8032-8082/8082 (END)

syslog after starting VM:

Code:
Dec  7 10:33:58 pve pvedaemon[27813]: start VM 106: UPID:pve:00006CA5:0051ADAE:5C0A3E86:qmstart:106:root@pam:
Dec  7 10:33:58 pve pvedaemon[27289]: <root@pam> starting task UPID:pve:00006CA5:0051ADAE:5C0A3E86:qmstart:106:root@pam:
Dec  7 10:33:59 pve systemd[1]: Started 106.scope.
Dec  7 10:33:59 pve systemd-udevd[27827]: Could not generate persistent MAC address for tap106i0: No such file or directory
Dec  7 10:33:59 pve kernel: [53528.985198] vmbr0: port 8(tap106i0) entered blocking state
Dec  7 10:33:59 pve kernel: [53528.985298] vmbr0: port 8(tap106i0) entered blocking state
Dec  7 10:34:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec  7 10:34:01 pve systemd[1]: Started Proxmox VE replication runner.
Dec  7 10:34:01 pve pvedaemon[27289]: <root@pam> end task UPID:pve:00006CA5:0051ADAE:5C0A3E86:qmstart:106:root@pam: OK
Dec  7 10:35:00 pve systemd[1]: Starting Proxmox VE replication runner...
Dec  7 10:35:01 pve systemd[1]: Started Proxmox VE replication runner.
 

coyote

New Member
Dec 4, 2018
29
0
1
36
It looks like we are not getting much further here.:(
Logfile looks ok, right?

What is the easiest way to test an older proxmox version?
Can I install an older iso and restore my LXC / VM backups (tar / vma.lzo) from my NAS?
 

dcsapak

Proxmox Staff Member
Staff member
Feb 1, 2016
3,798
345
83
31
Vienna

dcsapak

Proxmox Staff Member
Staff member
Feb 1, 2016
3,798
345
83
31
Vienna
additionally, i maybe get some access to a nuc in the near future, i can then test your setup and see if i find something
 

coyote

New Member
Dec 4, 2018
29
0
1
36
Is it possible to install proxmox on a USB Flashdrive/Stick? Just to test it, not for daily use
 

coyote

New Member
Dec 4, 2018
29
0
1
36
I did some more tests this weekend. New Proxmox install with Version 5.1-41
I have tested 3 different VM's, once on the TV and once on the touchscreen monitor

- my "old" MintVM
- new, fresh installed MintVM
- live boot MintVM (not installed)

and i tried 3 different configs:

First one:
Code:
hostpci0: 00:02.0,x-vga=1,rombar=0
without defined machine type

Second one:
Code:
hostpci0: 00:02.0,x-vga=1,rombar=0
machine: pc-i440fx-2.2
Third one:
Code:
args: -device vfio-pci,host=00:02.0,addr=0x02
machine: pc-i440fx-2.2
vga: none
Here are the results:

- Live boot MintVM with first and second config (TV and Touchscreen) -> VM boots up either to the desktop without menu or console output
photo_2018-12-10_11-09-25.jpg photo_2018-12-10_11-09-29.jpg photo_2018-12-10_11-09-220.jpg

- Live boot with third config -> no signal, doesn't boot (TV and Touchscreen)
---------------------
- "old" MintVM with first and second config (TV and Touchscreen) -> no signal, doesn't boot (TV and Touchscreen)
- "old" MintVM with third config (TV and Touchscreen) -> works fine
photo_2018-12-10_11-23-54.jpg
---------------------
- fresh installed MintVM with first and second config (TV and Touchscreen) -> no signal, doesn't boot (TV and Touchscreen)
- fresh installed MintVM with third config (TV and Touchscreen) -> works fine
---------------------

Only with "args line" and "vga: none" it works fine and i don't know why.
I know that this is no longer supported or no longer works in the new versions. But this is the only way at the moment how it works.

Maybe you still have an idea?!

Another question, is there e command to upgrade to 5.2 not 5.3?
 

Giara

New Member
Jul 29, 2018
1
0
1
23
I have the same problem, before the last upgrade my integrated video card (Intel HD 4000 on i7-3770) work without problem on Windows 7 VM.
Before upgrade i used these configurations:
Code:
machine:pc-i440fx-2.2
args: -device vfio-pci,host=00:02.0,addr=0x02
vga: none
But after upgrade cause me this error on vm start
Code:
kvm: -device VGA,id=vga,bus=pci.0,addr=0x2: PCI: slot 2 function 0 not available for VGA, in use by vfio-pci
Reading this thread i have tried some configuration like this:
Code:
machine:pc-i440fx-2.2
hostpci0: 00:02,x-vga=1,rombar=0
and at the first run windows 7 found the Video card and install the driver, but after reboot
it went into the blue screen :(

I also tried install a new Windows 10 virtual machine but after reboot only black screen appears on my TeamViewer display
 

dcsapak

Proxmox Staff Member
Staff member
Feb 1, 2016
3,798
345
83
31
Vienna
so i write it once again for all to read:

'vga: none' has not ever worked -> it had no effect whatsoever on the qemu-kvm call, so please leave it out

second:

if you try to recreate your 'args' line with the 'hostpci' use the same defaults
if you do not have set 'x-vga' do not set it on the hostpci line
if you did not set rombar=0, leave it on (this is the default)

for the detailed explanation of the options please read the documentation: https://pve.proxmox.com/pve-docs/chapter-qm.html#qm_pci_passthrough

third:
why do you all try with the ancient '2.2' machine type? we are currently on 2.12 and it should bring better compatibility
 

coyote

New Member
Dec 4, 2018
29
0
1
36
'vga: none' has not ever worked -> it had no effect whatsoever on the qemu-kvm call, so please leave it out
Then I wonder why only this configuration works and not only with me. I have exactly this configuration here from the forum.

second:....
I've already tried everything, in every constellation. I documented everything in the last post and I tried even more, without success.

third:...
Quite simply, because I'm back to an older version of Proxmox. Machine Type: 2.12 doesn't work on version 5.1.
But even with the latest version (5.3), I have tried this type, without success.

Here is my current proxmox version:

Code:
root@pve:~# pveversion -v
proxmox-ve: 5.1-32 (running kernel: 4.13.13-2-pve)
pve-manager: 5.1-41 (running version: 5.1-41/0b958203)
pve-kernel-4.13.13-2-pve: 4.13.13-32
libpve-http-server-perl: 2.0-8
lvm2: 2.02.168-pve6
corosync: 2.4.2-pve3
libqb0: 1.0.1-1
pve-cluster: 5.0-19
qemu-server: 5.0-18
pve-firmware: 2.0-3
libpve-common-perl: 5.0-25
libpve-guest-common-perl: 2.0-14
libpve-access-control: 5.0-7
libpve-storage-perl: 5.0-17
pve-libspice-server1: 0.12.8-3
vncterm: 1.5-3
pve-docs: 5.1-12
pve-qemu-kvm: 2.9.1-5
pve-container: 2.0-18
pve-firewall: 3.0-5
pve-ha-manager: 2.0-4
ksm-control-daemon: 1.2-2
glusterfs-client: 3.8.8-1
lxc-pve: 2.1.1-2
lxcfs: 2.0.8-1
criu: 2.11.1-1~bpo90
novnc-pve: 0.6-4
smartmontools: 6.5+svn4324-1
zfsutils-linux: 0.7.3-pve1~bpo9
Sorry, but this is the only config that works, even if you don't want to believe or hear it.
HDMI touchscreen Display shows me the Desktop, Touchscreen works fine and the gpu is passed through

Look:

Unbenannt.PNG
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE and Proxmox Mail Gateway. We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get your own in 60 seconds.

Buy now!