Updated to 8.2 - DMA error
- Thread starter JayTee75
- Start date
In my case I had to downgrade to kernel version 6.5.13-5-pve, and seems it works!
Thanks!
Dear Staff,
this error still occurs with the lastest kernal 6.8.8-2-pve.
After system boot, root login appears and around 5 seconds later i get 17 lines "... DMAR: ERROR: PTE for vPFN ..." like in the screenshot from inital post here.
System is a HP Microserver Gen 8, 16 GB Ram, PCIe Slot has a Dual Intel 10GBe Card, latest J06 Bios and BMC v2.82 iLO Firmware.
System had run fine with PVE 8.1 / Kernal 6.5.x.
Solution until kernal 6.8.8-2-pve had been is disabling vt-d in bios.
System and VMs had been a little bit slower, but worked for me.
Now wtih 6.8.8-2-pve + disabled vt-d i need to switch off KVM virtualisation and can not select host-cpu anymore,
i need to switch to "CPU Typ= Ivy Bridge or something like that", wich causes the VM runnig really slow.
Not good.
PVE System Log for the suspecting part:
this error still occurs with the lastest kernal 6.8.8-2-pve.
After system boot, root login appears and around 5 seconds later i get 17 lines "... DMAR: ERROR: PTE for vPFN ..." like in the screenshot from inital post here.
System is a HP Microserver Gen 8, 16 GB Ram, PCIe Slot has a Dual Intel 10GBe Card, latest J06 Bios and BMC v2.82 iLO Firmware.
System had run fine with PVE 8.1 / Kernal 6.5.x.
Solution until kernal 6.8.8-2-pve had been is disabling vt-d in bios.
System and VMs had been a little bit slower, but worked for me.
Now wtih 6.8.8-2-pve + disabled vt-d i need to switch off KVM virtualisation and can not select host-cpu anymore,
i need to switch to "CPU Typ= Ivy Bridge or something like that", wich causes the VM runnig really slow.
Not good.
PVE System Log for the suspecting part:
Hi,
did you already try disabling the
intel_iommu setting on the kernel commandline? The default for the setting changed in kernel 6.8 and your backtrace mentions IOMMU: https://pve.proxmox.com/wiki/Roadmap#8.2-known-issuesSo far, 2 worarounds possible:
1) disabling the intel_iommu setting on the GRUB kernel commandline
2) pin kernel to 6.5
With #1, System is slower, with #2 there may be incomaptibilities.
Still hoping for a working 6.8+ kernal, because the HP Gen8 microserver is still a very good piece of hardware.
1) disabling the intel_iommu setting on the GRUB kernel commandline
2) pin kernel to 6.5
With #1, System is slower, with #2 there may be incomaptibilities.
Still hoping for a working 6.8+ kernal, because the HP Gen8 microserver is still a very good piece of hardware.
Hello!
Just noticed this error too, but contrary to others, my Gen8 microserver with 2 VM's is working just fine. Running 8.2.4 with all the latest updates. No kernel downgrades and no features disabled in Bios.
Remote console seems to be stuck, but Proxmox Gui shell and SSH is working.
TBH I rarely access the host console, so I guess it's been like this for months. No RAID, just AHCI.
Interesting!
Just noticed this error too, but contrary to others, my Gen8 microserver with 2 VM's is working just fine. Running 8.2.4 with all the latest updates. No kernel downgrades and no features disabled in Bios.
Remote console seems to be stuck, but Proxmox Gui shell and SSH is working.
TBH I rarely access the host console, so I guess it's been like this for months. No RAID, just AHCI.
Interesting!
Having just completed a clean install of Proxmox 8.3.1 I'm getting the same DMAR: ERROR: DMA PTE for vPFN 0xf1xxx already set error in the console.
But neither of the boot configurations (followed by a reboot) resolved this.
or
It was necessary to disable VT-d in the BIOS configuration to resolve the errors.
Boot > F9 Bios > System > Processor Options > VT-d = disabled
Environment:
But neither of the boot configurations (followed by a reboot) resolved this.
echo "INTEL_IOMMU=OFF" >> /etc/default/grubupdate-grubor
echo "INTEL_IOMMU=OFF" >> /etc/kernel/cmdlineproxmox-boot-tool refreshIt was necessary to disable VT-d in the BIOS configuration to resolve the errors.
Boot > F9 Bios > System > Processor Options > VT-d = disabled
Environment:
| Hardware | HP Proliant Microserver G8 https://support.hpe.com/connect/s/product?language=en_US&kmpmoid=5379860&tab=driversAndSoftware |
| CPU | Intel(R) Xeon(R) CPU E3-1265L V2 @ 2.50GHz Supports VT-x and VT-d https://ark.intel.com/content/www/u...-processor-e3-1265l-v2-8m-cache-2-50-ghz.html |
| Proxmox | 8.3.1 Installed from proxmox-ve_8.3-1.iso All updates applied (apt update && apt full-upgrade |
| Kernel | Linux pve 6.8.12-5-pve #1 SMP PREEMPT_DYNAMIC PMX 6.8.12-5 (2024-12-03T10:26Z) x86_64 GNU/Linux |
AFAIK, this is case-sensitive. Please try with
intel_iommu=off. Also please see https://pve.proxmox.com/pve-docs/chapter-sysadmin.html#sysboot_edit_kernel_cmdline for how to properly edit the commandline.
Last edited:
Thanks Fiona for correctly pointing out I'm not following the instructions https://pve.proxmox.com/pve-docs/chapter-sysadmin.html#sysboot_edit_kernel_cmdline
Running the following resolves the errors:
i.e. configuring
And further testing confirms you are correct that intel_iommu=off must be in lower case.
Thanks for pointing me in the right direction
Running the following resolves the errors:
Code:
sed -i s/"quiet"/"quiet intel_iommu=off"/g /etc/default/grub
update-grub
rebooti.e. configuring
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=off"And further testing confirms you are correct that intel_iommu=off must be in lower case.
Thanks for pointing me in the right direction
This also affects the installer. Guess what, I am unable to type an = character via the servers remote kvm.
So my solution was to grab a keyboard drive to the location where the server is located just to press = on the physical keyboard, so I could do the rest remotely again. Good luck if You're trying to install proxmox on a random rented root server.
I propose for that reason to add an extra boot entry to the installer with intel_iommu=off allready present in linux commandline.
So my solution was to grab a keyboard drive to the location where the server is located just to press = on the physical keyboard, so I could do the rest remotely again. Good luck if You're trying to install proxmox on a random rented root server.
I propose for that reason to add an extra boot entry to the installer with intel_iommu=off allready present in linux commandline.