PCIe passthrough does not work

spirit said:
Code: 
Hi guy,ovmf support has been added to last proxmox 4.X in pve-no-subscription repostory.just add   ovmf:1 in vmid.conf.(use virtio-scsi to boot).Your guest os need to be installed to boot on uefi.
Click to expand...
Code: 
Having a hard time getting this to work.Updated Test machine via apt-get update && apt-get upgrade to latestthen edited vmid.confi tried:ovmf:1 --->   omvf unable to parse value of 'ovmf' - unknown setting 'ovmf'ovmf1:ovmf:ovmfI even tried the above combo with and without "smbios1: uuid=e4c83fdf-aa6c-46f5-b9e6-a5397a3aafdf"You sure it is in yet ?
 
Last edited:
Q-wulf said:
Code: 
Having a hard time getting this to work.Updated Test machine via apt-get update && apt-get upgrade to latestthen edited vmid.confi tried:ovmf:1 --->   omvf unable to parse value of 'ovmf' - unknown setting 'ovmf'ovmf1:ovmf:ovmfI even tried the above combo with and without "smbios1: uuid=e4c83fdf-aa6c-46f5-b9e6-a5397a3aafdf"You sure it is in yet ?
Click to expand...


Hi, yes, sorry, I thinked my patch was already commited. It's be ready in coming days.
I'll make an update in this post.
 
Guess its this one:
qemu-server (4.0-41) unstable; urgency=medium * add new option to select BIOS (--bios ) * fix bug #841: replace get_used_paths with is_volume_in_use -- Proxmox Support Team Thu, 10 Dec 2015 11:15:23 +0100
Click to expand...
will report back results as i have em. note:- Can confirm that the Option shows up.- Do i need to change the SMBIO settings (type1) uui= values ?
 
using the following steps:

https://pve.proxmox.com/wiki/USB_physical_port_mapping#Add_the_device_to_the_VM

lsusb -t
lsusb

Bus 005 Device 004: ID 046d:c50d Logitech, Inc. Cordless Mouse
Bus 005 Device 003: ID 0566:3107 Monterey International Corp. Keyboard


nano /etc/pve/nodes/<hostname>/qemu-server/<vmnumber>.conf
usb0: host=046d:c50d
usb1: host=0566:3107


https://forum.proxmox.com/threads/24362-PCIe-passthrough-does-not-work/page2




install windows 10 from Microsoft ISO:
Select SCSI Bus
Select Virtio Controller
Add windows 10 Iso as First DVD
Add Virtio-win-0.1.110 as secondary DVD
add omvf based bios via “options” tab.

Formart disk to GPT:
In windows Setup hit “shift+F10”
Type Diskpart
List Disk
—
Select Disk <DiskNumber>
clean
convert GPT
repeat for all disks to be converted
—

On Install Select the “load Driver” option - install VioSCSI Driver (64 Bit)
Install Windows
Click to expand...

Once the Windows install is finished, the VM shuts down. When i Start it i get the following error:

Boot Failed. EFI SCSI Device
Click to expand...

after a while the following pops up in noVNC:
https://www.dropbox.com/s/t7vkikngocxkhat/Screen Shot 2015-12-10 at 15.24.33.png

Side-Note
apparently i cannot upload any attachments anymore and the WYSIWYG Editor is broken :()

edit:
vmid.conf
#e4c83fdf-aa6c-46f5-b9e6-a5397a3aafdf
#ovmf
balloon: 1024
bios: ovmf
bootdisk: scsi0
cores: 6
machine: q35
memory: 4096
name: redacted
net0: virtio=32:62:37:32:33:61,bridge=vmbr0
numa: 0
ostype: win8
sata2: none,media=cdrom
sata3: none,media=cdrom
scsi0: local:120/vm-120-disk-1.qcow2,cache=writeback,size=20G
scsi1: P1__HDD_OSD_EC_F2:vm-120-disk-1,cache=writeback,size=500G
scsihw: virtio-scsi-pci
smbios1: uuid=e4c83fdf-aa6c-46f5-b9e6-a5397a3aafdf
sockets: 1
usb0: host=046d:c50d
usb1: host=0566:3107
Click to expand...
 
Last edited:
spirit said:
packages have been pushed to pve-no-subscription today !

small config difference :

bios: ovmf

(always available in gui, in vm options)
Click to expand...


Hey spirit, so i've finally got some time to test this and am running into the following issue:

When i try to pass the GPU through (working windows 10 installed via noVNC)

errors during startup of KVM:
Code: 
Use of uninitialized value $kvmver in pattern match (m//) at /usr/share/perl5/PVE/QemuServer.pm line 6378.
Use of uninitialized value $current_major in numeric ge (>=) at /usr/share/perl5/PVE/QemuServer.pm line 6384.
GPU does not get passed through, noVNC displays "This Vm has no graphic display device.". guest had been installed via noVNC and uefi support (and works without passed through GPU).


using the following vmid.conf
Code: 
balloon: 1024
bios: ovmf
boot: cd
bootdisk: scsi0
cores: 10
hostpci0: 02:00,x-vga=on,pcie=1
hostpci1: 00:1b.0
ide0: local:iso/virtio-win-0.1.110.iso,media=cdrom,size=55260K
ide2: local:iso/Win10_1511_EnglishInternational_x64.iso,media=cdrom,size=3895734K
keyboard: de
machine: q35
memory: 8192
name: win10.datamile.lan
numa: 0
ostype: win8
scsi0: P2__HDD_OSD_R_F3:vm-1051-disk-1,cache=writeback,size=50G
scsi1: P1__HDD_OSD_EC_F2:vm-1051-disk-1,cache=writeback,size=200G
scsihw: virtio-scsi-pci
smbios1: uuid=d5c400b0-2f7e-47a0-8f20-5aca126c08c1
sockets: 1
usb0: host=046d:c50d
usb1: host=0566:3107

hostpci0: 02:00,x-vga=on,pcie=1 <<- AMD R9 290X 8GB
hostpci1: 00:1b.0 <<- Onboard soundcard

Not sure where the issue is. Any ideas ?

edit: the error seems to be tied to teh following entry:
Code: 
machine: q35
If i uncomment it,
Code: 
#machine: q35

The error message does not get displayed, but it complaints that Q35 is missing. If i then uncomment the hostpci lines then there is no error message.

edit2: fully updated proxmox (17/12/15@07:15) with pve-no-subscription repository
 
Last edited:
I also have a problem with GPU passthrough, I figured the problem is caused by the iommu grouping:
/sys/kernel/iommu_groups/1/devices/0000:00:01.0 <- PCI bridge - 8086:0c01
/sys/kernel/iommu_groups/1/devices/0000:00:01.1 <- PCI bridge - 8086:0c05
/sys/kernel/iommu_groups/1/devices/0000:01:00.0 <- AMD Radeon HD5450 - 1002:68f9
/sys/kernel/iommu_groups/1/devices/0000:01:00.1 <- AMD Radeon Audio - 1002:aa68
/sys/kernel/iommu_groups/1/devices/0000:02:00.0 <- LSI Logic / Symbios Logic SAS2308 - 1000:0086


Since my onboard SAS controller already has been passed through to another guest, I can't pass through the GPU, starting the guest with the GPU results in:
kvm: -device vfio-pci,host=01:00.0,id=hostpci0.0,bus=ich9-pcie-port-1,addr=0x0.0,x-vga=on,multifunction=on: vfio: error opening /dev/vfio/1: Device or resource busy

Does anyone have a fix for the iommu grouping issue? I'm on
proxmox-ve: 4.1-26 (running kernel: 4.2.6-1-pve)
 
Last edited:
Q-wulf said:
Hey spirit, so i've finally got some time to test this and am running into the following issue:

When i try to pass the GPU through (working windows 10 installed via noVNC)

errors during startup of KVM:
Code: 
Use of uninitialized value $kvmver in pattern match (m//) at /usr/share/perl5/PVE/QemuServer.pm line 6378.
Use of uninitialized value $current_major in numeric ge (>=) at /usr/share/perl5/PVE/QemuServer.pm line 6384.
GPU does not get passed through, noVNC displays "This Vm has no graphic display device.". guest had been installed via noVNC and uefi support (and works without passed through GPU).


using the following vmid.conf
Code: 
balloon: 1024
bios: ovmf
boot: cd
bootdisk: scsi0
cores: 10
hostpci0: 02:00,x-vga=on,pcie=1
hostpci1: 00:1b.0
ide0: local:iso/virtio-win-0.1.110.iso,media=cdrom,size=55260K
ide2: local:iso/Win10_1511_EnglishInternational_x64.iso,media=cdrom,size=3895734K
keyboard: de
machine: q35
memory: 8192
name: win10.datamile.lan
numa: 0
ostype: win8
scsi0: P2__HDD_OSD_R_F3:vm-1051-disk-1,cache=writeback,size=50G
scsi1: P1__HDD_OSD_EC_F2:vm-1051-disk-1,cache=writeback,size=200G
scsihw: virtio-scsi-pci
smbios1: uuid=d5c400b0-2f7e-47a0-8f20-5aca126c08c1
sockets: 1
usb0: host=046d:c50d
usb1: host=0566:3107

hostpci0: 02:00,x-vga=on,pcie=1 <<- AMD R9 290X 8GB
hostpci1: 00:1b.0 <<- Onboard soundcard

Not sure where the issue is. Any ideas ?

edit: the error seems to be tied to teh following entry:
Code: 
machine: q35
If i uncomment it,
Code: 
#machine: q35

The error message does not get displayed, but it complaints that Q35 is missing. If i then uncomment the hostpci lines then there is no error message.

edit2: fully updated proxmox (17/12/15@07:15) with pve-no-subscription repository
Click to expand...

Update on this:

I did some more testing and that "error message" also happens when i switch from omvf to seabios . But here is the kicker. NOW GPU passthrough works with Seabios.

So my guess is that with the patch introduced here:
spirit said:
packages have been pushed to pve-no-subscription today !
[...]
Click to expand...

The original issue was fixed.

However with omvf it still is not working :P
 
Q-wulf said:
GPU does not get passed through, noVNC displays "This Vm has no graphic display device.". guest had been installed via noVNC and uefi support (and works without passed through GPU).
Click to expand...

AFAIK,This is normal, when you passthrough video card, you don't have emulated vga card, so novnc can't work.

you don't have any display on your physicard video card output ?
 
spirit said:
AFAIK,This is normal, when you passthrough video card, you don't have emulated vga card, so novnc can't work.

you don't have any display on your physicard video card output ?
Click to expand...

Yes, i would not get any Output on a monitor attached via DVI to the GPU i passed through.

Since with omvf you already installl windows as UEFI, does the GPU need to have a UEFI Bios ?


In the meantime i moved to testing Windows 10 + Gpu passthrough using Seabios (after my initial issue there seemed to have been fixed with one of the last updates). Running into a issue there as well, documented (in detail) here:
https://forum.proxmox.com/threads/amd-gpu-passthrough-issue-windows-10-steps-included.25427/

In any case, i do still have the OMVF Vm, so i can continue doing tests on that.
 
Yeah I am passing through a small AMD Radeon HD5450 to a guest running OpenELEC. (Supermicro X10SL7-F + Intel Xeon E3-1230v3)
But I haven't verified if I even have a working display output and sound, since I'm not at home at the moment, I just started the VM remotely.

One problem was, I used SeaBIOS at first, which caused a complete freeze of the host when I rebooted the guest, after switching to OVMF the problem was gone.
So I have to test this for the next 3-4 weeks before I can say it runs smooth and stable, but if I don't get any problems with this setup, I might replace my HTPC with this. :)

But I'm not really sure, since the Xeon E3 doesn't support real ACS and I could run into problems.. But until now I've been passing through the onboard LSI SAS controller to a guest running OpenMediaVault with ZFS on Linux and a Tevii S471 TV card to a Debian Jessie guest with VDR and I haven't had any problems since November 2015.
 
Last edited:
shawly said:
Yeah I am passing through a small AMD Radeon HD5450 to a guest running OpenELEC. (Supermicro X10SL7-F + Intel Xeon E3-1230v3)
But I haven't verified if I even have a working display output and sound, since I'm not at home at the moment, I just started the VM remotely.

One problem was, I used SeaBIOS at first, which caused a complete freeze of the host when I rebooted the guest, after switching to OVMF the problem was gone.
So I have to test this for the next 3-4 weeks before I can say it runs smooth and stable, but if I don't get any problems with this setup, I might replace my HTPC with this. :)

But I'm not really sure, since the Xeon E3 doesn't support real ACS and I could run into problems.. But until now I've been passing through the onboard LSI SAS controller to a guest running OpenMediaVault with ZFS on Linux and a Tevii S471 TV card to a Debian Jessie guest with VDR and I haven't had any problems since November 2015.
Click to expand...

Thanks for the report ! Glad to see that ovmf is working fine.
 
Have any of you gotten "This device cannot find enough free resources that it can use. (Code 12)" when trying to install nvidia drivers?
 
Soulgriever said:
Have any of you gotten "This device cannot find enough free resources that it can use. (Code 12)" when trying to install nvidia drivers?
Click to expand...

are you using last proxmox 4.1 with all updates ?

This error code from detection of kvm and hyper-v features by the nvidia driver, (since 344.11) and we mask them in last proxmox updates.
 
spirit said:
are you using last proxmox 4.1 with all updates ?

This error code from detection of kvm and hyper-v features by the nvidia driver, (since 344.11) and we mask them in last proxmox updates.
Click to expand...
Code: 
proxmox-ve: 4.1-26 (running kernel: 4.2.6-1-pve)
pve-manager: 4.1-1 (running version: 4.1-1/2f9650d4)
pve-kernel-4.2.6-1-pve: 4.2.6-33
lvm2: 2.02.116-pve2
corosync-pve: 2.3.5-2
libqb0: 0.17.2-1
pve-cluster: 4.0-29
qemu-server: 4.0-41
pve-firmware: 1.1-7
libpve-common-perl: 4.0-41
libpve-access-control: 4.0-10
libpve-storage-perl: 4.0-38
pve-libspice-server1: 0.12.5-2
vncterm: 1.2-1
pve-qemu-kvm: 2.4-17
pve-container: 1.0-32
pve-firewall: 2.0-14
pve-ha-manager: 1.0-14
ksm-control-daemon: 1.2-1
glusterfs-client: 3.5.2-2+deb8u1
lxc-pve: 1.1.5-5
lxcfs: 0.13-pve1
cgmanager: 0.39-pve1
criu: 1.6.0-1
zfsutils: 0.6.5-pve6~jessie
root@GameMachine:~# nano /etc/pve/qemu-server/100.conf
root@GameMachine:~# pveversion -v
proxmox-ve: 4.1-26 (running kernel: 4.2.6-1-pve)
pve-manager: 4.1-1 (running version: 4.1-1/2f9650d4)
pve-kernel-4.2.6-1-pve: 4.2.6-33
lvm2: 2.02.116-pve2
corosync-pve: 2.3.5-2
libqb0: 0.17.2-1
pve-cluster: 4.0-29
qemu-server: 4.0-41
pve-firmware: 1.1-7
libpve-common-perl: 4.0-41
libpve-access-control: 4.0-10
libpve-storage-perl: 4.0-38
pve-libspice-server1: 0.12.5-2
vncterm: 1.2-1
pve-qemu-kvm: 2.4-17
pve-container: 1.0-32
pve-firewall: 2.0-14
pve-ha-manager: 1.0-14
ksm-control-daemon: 1.2-1
glusterfs-client: 3.5.2-2+deb8u1
lxc-pve: 1.1.5-5
lxcfs: 0.13-pve1
cgmanager: 0.39-pve1
criu: 1.6.0-1
zfsutils: 0.6.5-pve6~jessie

GTX 970 Nvidia 361.75 Do I need to update something? also I'm running windows 7 ultimate on the ovmf bios if that makes a difference.
 
Soulgriever said:
Code: 
proxmox-ve: 4.1-26 (running kernel: 4.2.6-1-pve)
pve-manager: 4.1-1 (running version: 4.1-1/2f9650d4)
pve-kernel-4.2.6-1-pve: 4.2.6-33
lvm2: 2.02.116-pve2
corosync-pve: 2.3.5-2
libqb0: 0.17.2-1
pve-cluster: 4.0-29
qemu-server: 4.0-41
pve-firmware: 1.1-7
libpve-common-perl: 4.0-41
libpve-access-control: 4.0-10
libpve-storage-perl: 4.0-38
pve-libspice-server1: 0.12.5-2
vncterm: 1.2-1
pve-qemu-kvm: 2.4-17
pve-container: 1.0-32
pve-firewall: 2.0-14
pve-ha-manager: 1.0-14
ksm-control-daemon: 1.2-1
glusterfs-client: 3.5.2-2+deb8u1
lxc-pve: 1.1.5-5
lxcfs: 0.13-pve1
cgmanager: 0.39-pve1
criu: 1.6.0-1
zfsutils: 0.6.5-pve6~jessie
root@GameMachine:~# nano /etc/pve/qemu-server/100.conf
root@GameMachine:~# pveversion -v
proxmox-ve: 4.1-26 (running kernel: 4.2.6-1-pve)
pve-manager: 4.1-1 (running version: 4.1-1/2f9650d4)
pve-kernel-4.2.6-1-pve: 4.2.6-33
lvm2: 2.02.116-pve2
corosync-pve: 2.3.5-2
libqb0: 0.17.2-1
pve-cluster: 4.0-29
qemu-server: 4.0-41
pve-firmware: 1.1-7
libpve-common-perl: 4.0-41
libpve-access-control: 4.0-10
libpve-storage-perl: 4.0-38
pve-libspice-server1: 0.12.5-2
vncterm: 1.2-1
pve-qemu-kvm: 2.4-17
pve-container: 1.0-32
pve-firewall: 2.0-14
pve-ha-manager: 1.0-14
ksm-control-daemon: 1.2-1
glusterfs-client: 3.5.2-2+deb8u1
lxc-pve: 1.1.5-5
lxcfs: 0.13-pve1
cgmanager: 0.39-pve1
criu: 1.6.0-1
zfsutils: 0.6.5-pve6~jessie

GTX 970 Nvidia 361.75 Do I need to update something? also I'm running windows 7 ultimate on the ovmf bios if that makes a difference.
Click to expand...
 
You must log in or register to reply here.

About

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

Quick Navigation

Home Get Subscription Wiki Downloads Proxmox Customer Portal About

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 yours easily in our online shop.

Buy now!
Community platform by XenForo® © 2010-2024 XenForo Ltd.
Top Bottom