Updated to 8.2 - DMA error

I had a similar error dmar error dma pte for vpfn 0xcf4de already set,
Saw another article on https://forum.openmediavault.org/ by searching just that it was the 3rd hit on Google.
Saw about disabling virtualization. I can't believe that actually worked.
Nothing in that error would have caused me to believe that virtualization was the issue.
If anyone else has insight on that, please enlighten me.
 
I had a similar error when installing Proxmox Backup Server on a Dell PowerEdge r410
dmar error dma pte for vpfn 0xcf4de already set,
Saw another article on https://forum.openmediavault.org/ by searching just that and it was the 3rd hit on Google.
Said about disabling virtualization. I can't believe that actually worked.
Nothing in that error would have caused me to believe that virtualization was the issue.
If anyone else has insight on that, please enlighten me.
 
fiona said:
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-issues
Click to expand...
I had the exact same issue (DMAR ERROR) installing 8.3.1 on a 12 year old Dell R710.
Adding "nomodeset intel_iommu=off" to the installer kernel command line solved it for me.
https://pve.proxmox.com/wiki/Installation#nomodeset_kernel_param
Thanks for the hints. Now it is running like a charm :)
 
  • Like
Reactions: Gavino
I'm installing on an old Dell R710 as well. Thanks for the info! I selected "Install Proxmox Backup Server (Console)" and hit "e" to edit. At the line starting with "linux" I appended "nomodeset intel_iommu=off" to the end. Then Ctrl-x and I'm good. :) When editing the GRUB file, it's best to go to the line below the one that starts with "linux" and step backwards - otherwise it takes forever to step through the whole line.
 
  • Like
Reactions: Stefan Radman
This made it possible to install on my T410's without issue until I can replace them. So thankyou very much!
Gavino said:
I'm installing on an old Dell R710 as well. Thanks for the info! I selected "Install Proxmox Backup Server (Console)" and hit "e" to edit. At the line starting with "linux" I appended "nomodeset intel_iommu=off" to the end. Then Ctrl-x and I'm good. :) When editing the GRUB file, it's best to go to the line below the one that starts with "linux" and step backwards - otherwise it takes forever to step through the whole line.
Click to expand...
FYI, you can press the END key on your keyboard to jump directly to the end of the line instead of moving char by char
 
  • Like
Reactions: Gavino
JayTee75 said:
I just upgrded HPE Microserver gen8 to new version (8.2), and got this after while with system stuck:

View attachment 67053


Any ideas?
Click to expand...
I wanted to share a fix for the Proxmox Backup Server error encountered above

I’m running Proxmox Backup Server on a Dell PowerEdge R710, and I managed to resolve the issue by doing the following:

Add kernel parameters to GRUB
1. Open terminal and run:
nano /etc/default/grub

2. Find the line:
GRUB_CMDLINE_LINUX_DEFAULT="quiet"

3. Change it to:
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on iommu=pt intremap=off"

4. Save and exit

5. Run:
update-grub

6. Reboot:
reboot
 
Last edited:
  • Like
Reactions: kalobok and Gavino
I also have an HPE ProLiant MicroServer Gen8 (with E3-1220L V2): it's not running Proxmox (yet) but I've been testing this hardware using vanilla Debian 12.11 (kernel v6.1.140-1).

OS is installed on a dedicated HDD connected to the B120i internal card with extra SFF-8087 cables, the four data HDDs are connected to a PCI Express SAS card: INSPUR 9211-8i (Avago BIOS v7.39.00.00).

In idling, the System was stable for days but the moment I've created an mdadm RAID1 over the 4 HDDs I've got multiple errors (full log attached):

kernel: DMAR: ERROR: DMA PTE for vPFN 0xf1f80 already set (to f1f80003 not 120d5c001)
Click to expand...
Maybe it's the CPU operation (system was clearly idling before this RAID test), maybe it's the disk I/O.

I'm not attempting to highjack this thread, but I wanted to share that kernel 6.1 also has the issue on this hardware.

I'll try stressing the CPU without RAID to see if it triggers the issue or not.

I'll also try Debian 13 with kernel 6.12 and PVE 9 with kernel 6.14 to see how it performs.
 

Attachments

  • Like
Reactions: Kurgan
I am running the gen 8 with HP P420 HBA, with proxmox 9 (Debian 13 Trixie). I just came across this because i was getting same error. I did the grub changes suggested above, and have not seen the issue so far within 1 hour and over 3 terabytes of data transferred. Prior to disabling iommu, i had a lot of trouble setting up zfs pool, and could not reliably transfer even small amounts of data.
 
  • Like
Reactions: Kurgan
Kurgan said:
This is interesting indeed, please keep us informed. Did it run smoothly on Debian 11?
Click to expand...
Sorry, I did not tried Debian 11: this is a new setup initiated with Debian 12.

jrl_1644 said:
I am running the gen 8 with HP P420 HBA, with proxmox 9 (Debian 13 Trixie). I just came across this because i was getting same error. I did the grub changes suggested above, and have not seen the issue so far within 1 hour and over 3 terabytes of data transferred. Prior to disabling iommu, i had a lot of trouble setting up zfs pool, and could not reliably transfer even small amounts of data.
Click to expand...
Interesting, it (sadly) looks like I'm gonna have the same issue without disabling iommu :'(
 
Hello everyone,

I had an issue when trying to install Proxmox VE 8.4.1 on a Dell PowerEdge R610 (with PERC 6/i RAID controller).
During the installer boot, the screen was flooded with: "DMAR: ERROR: DMA PTE for vPFN ..."

The system still loaded, but these errors are related to Intel VT-d (IOMMU).
The R610 is quite old, and its BIOS/firmware implementation of VT-d is incomplete/buggy, which causes problems with modern Linux kernels.

Fix that worked for me​

  1. Boot the Proxmox installer from USB.
  2. On the first boot screen, highlight “Install Proxmox VE”.
  3. Press e to edit.
  4. On the line starting with: "linux /boot/linux26 root=... quiet" I add "intel_iommu=off"
  5. Press F10 (or Ctrl+X) to boot
This allowed me to install Proxmox without the DMAR errors.

Make it permanent after installation​

Edit the GRUB configuration "nano /etc/default/grub" and change the line: "GRUB_CMDLINE_LINUX_DEFAULT="quiet"" to "GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=off""

Then update GRUB:
update-grub or
proxmox-boot-tool refresh
 
Last edited:
  • Like
Reactions: Supaman
Same console error in my Microserver Gen8 - Xeon 1265L - Proxmox version 9.0.3 - No hw controller - AHCI passthr. and ZFS.

:confused:
 
This problem still seems to exist in proxmox 9.1. I've installed 9.1 on a couple of old Dell R620s and R720s, but attempted to load it on a R510 with an H310 card in IT mode and got those errors during install. The installer was unable to complete.
 
Gruic said:
This problem still seems to exist in proxmox 9.1. I've installed 9.1 on a couple of old Dell R620s and R720s, but attempted to load it on a R510 with an H310 card in IT mode and got those errors during install. The installer was unable to complete.
Click to expand...
I just did an install of 9.1 on a T610. To do the install, I had to press e when "Install Proxmox (Graphical)" was selected, then add intel_iommu=off to the end of the "linux /boot/linux26" line, and press F10 to get it booting using that setting.

I didn't need to edit GRUB during the reboot as it already had intel_iommu=off added (the installer may have added it itself).
 
  • Like
Reactions: Kurgan
Yes, I can confirm the issue is still there (and will never go away, it's an issue with older hardware) and that if you add the intel_iommu=off during installation, then the installer adds it to the installed GRUB so no need to re-add it.
 
I seem to have a similar issue. I recently updated my system from 8.4.11 to 8.4.19 on my HP ProLiant DL380p Gen8 server with P420i RAID controller

It booted several VM and after 5 to 10 minutes just crashed.

Partial log below

If I interrupt the startup and run kernel /boot/vmlinuz-6.8.12-4-pve, the system seems stable

but if I run the default /boot/vmlinuz-6.8.12-29-pve it crashes

I am confused by the beginning of this post were someone is setting

intel_iommu=on

in their command line

But the Proxmox admin is saying this should be set to
intel_iommu=off

Could anyone suggest a solution for my issue? I cannot seem to get an updated bios as HP does not seem to give these out unless you pay for a service agreement. If I am wrong, can someone suggest where to downlaod a newer bios for this machine?

The log where I think the error is:
Jun 04 10:54:35 pve kernel: DMAR: ERROR: DMA PTE for vPFN 0xbdf84 already set (to bdf84003 not 1028ecf003)
Jun 04 10:54:35 pve kernel: ------------[ cut here ]------------
Jun 04 10:54:35 pve kernel: WARNING: CPU: 21 PID: 1485 at drivers/iommu/intel/iommu.c:2231 __domain_mapping+0x30c/0x510
Jun 04 10:54:35 pve kernel: Modules linked in: cfg80211 cmac nls_utf8 cifs cifs_arc4 nls_ucs2_utils rdma_cm iw_cm ib_cm ib_core cifs_md4 netfs veth ebtable_filter ebtables ip_set ip6table_raw iptable_raw ip6table_filter ip6_tables iptable_filter nf_tables bonding tls softdog sunrpc nfnetlink_log nfnetlink binfmt_misc intel_rapl_msr intel_rapl_common sb_edac x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel ipmi_ssif kvm irqbypass crct10dif_pclmul polyval_clmulni polyval_generic ghash_clmulni_intel sha256_ssse3 sha1_ssse3 aesni_intel crypto_simd cryptd rapl acpi_power_meter ipmi_si mgag200 acpi_ipmi ipmi_devintf intel_cstate vga16fb input_leds i2c_algo_bit ioatdma ipmi_msghandler vgastate serio_raw hpilo mac_hid pcspkr dca zfs(PO) spl(O) vhost_net vhost vhost_iotlb tap efi_pstore dmi_sysfs ip_tables x_tables autofs4 btrfs blake2b_generic xor raid6_pq hid_generic usbmouse usbkbd usbhid hid dm_thin_pool dm_persistent_data dm_bio_prison dm_bufio libcrc32c crc32_pclmul psmouse uhci_hcd ehci_pci hpsa ehci_hcd lpc_ich pata_acpi tg3
Jun 04 10:54:35 pve kernel: scsi_transport_sas
Jun 04 10:54:35 pve kernel: CPU: 21 PID: 1485 Comm: kvm Tainted: P IO 6.8.12-29-pve #1
Jun 04 10:54:35 pve kernel: Hardware name: HP ProLiant DL380p Gen8, BIOS P70 03/01/2013
Jun 04 10:54:35 pve kernel: RIP: 0010:__domain_mapping+0x30c/0x510
Jun 04 10:54:35 pve kernel: Code: 48 89 c2 4c 89 5d b0 48 c7 c7 80 38 c4 95 e8 eb 31 6c ff 8b 05 89 ad 9c 01 4c 8b 5d b0 85 c0 74 09 83 e8 01 89 05 78 ad 9c 01 <0f> 0b e9 f3 fe ff ff 41 80 e2 7f e9 d1 fe ff ff 41 83 c8 01 49 63
Jun 04 10:54:35 pve kernel: RSP: 0018:ffffd0a0b251efc0 EFLAGS: 00010202
Jun 04 10:54:35 pve kernel: RAX: 0000000000000004 RBX: 0000000000000001 RCX: 0000000000000000
Jun 04 10:54:35 pve kernel: RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
Jun 04 10:54:35 pve kernel: RBP: ffffd0a0b251f058 R08: 0000000000000000 R09: 0000000000000000
Jun 04 10:54:35 pve kernel: R10: 0000000000000000 R11: 0000000000000002 R12: ffff8cfc8826ca00
Jun 04 10:54:35 pve kernel: R13: ffff8cfc8a861c20 R14: 0000001028ecf003 R15: ffff8cfc8a861c20
Jun 04 10:54:35 pve kernel: FS: 00007b2dca4b76c0(0000) GS:ffff8d0d77480000(0000) knlGS:0000000000000000
Jun 04 10:54:35 pve kernel: CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jun 04 10:54:35 pve kernel: CR2: 00007b2ae57ff000 CR3: 0000001266f9e001 CR4: 00000000000626f0
 
Last edited:
On is the default config, so setting it to on makes no difference. Off seems to work for quite a lot of old hardware with newer kernels. Set it to off.
 
slammers67 said:
...

I am confused by the beginning of this post were someone is setting

intel_iommu=on

in their command line
...
Click to expand...
If you look closely, the original posting indeed said "intel_iommu = on" (mind the spaces, inbetween the setting and its value!). So, if the quote was actually what that person had set, then it is quite likely, that "intel_iommu" never was set to "on", as the syntax was simply wrong.

And I would listen to what the Proxmox Admin tells you, rather than listening to some stranger.
 
celemine1gig said:
If you look closely, the original posting indeed said "intel_iommu = on" (mind the spaces, inbetween the setting and its value!). So, if the quote was actually what that person had set, then it is quite likely, that "intel_iommu" never was set to "on", as the syntax was simply wrong.

And I would listen to what the Proxmox Admin tells you, rather than listening to some stranger.
Click to expand...
Agreed, I am just not 100% sure this is my issue. From reading the threads it sounds like this change was made in kernel 6.8. I have tested all the way upto 6.8.12-18 and it I have no issues. But if I try 6.8.12-29, it crashes after some time. I actualy was able to update my bios to the newest version. It ran much longer but its still crashed.

I am interested in the difference between 18 and 29.
 
You must log in or register to reply here.
Top Bottom
Back