KVM virtualisation broken with kernel 5.15 Proxmox 7 / Hp DL380p Gen8

[AurelienLoron]

Hi,
I've recently update proxmox to V7.
Everythings was fine until i do a reboot, i need to do it for a kernel update.

Since this last update, i've message on the boot :

ACPI: SPCR: Unexpected SPCR access Width. Defaulting to byte size
DMAR: [fimware bug] : No firmware reserved region can cover this RMRR [ .....
[fimware bug] : the bios has corrupted hw-PMU ressources (MSR 38d is 330)
x86/cpu: VMX (outside TXT) disabled by bios
DMAR: DRHD: handling fault status reg 2
DMAR: [INTR-REMAP] request device [01:00.0] fault index 0x32 [fault reason 0x26] blocked an interrupt request due to source-id verification failure

The consequence is that i can't run my VM using host as processeur.
I need to put "KVM hardware virtualisation" to NO and choose kvm64 as processeur to launch the VM. I loose lot of performance !

My machine is an HPe DL380p Gen8 with 2 Xeon E5 2630
Kernel version : Linux 5.15.108-1-pve #1 SMP PVE 5.15.108-2
PVE version : pve-manager/7.4-16/0f39f621
VT is enabled in bios, i did an update for the bios.

I've open the same ticket on proxmox formum : https://forum.proxmox.com/threads/kvm-virtualisation-broken-with-kernel-5-15.132137/

[Suman_1978]

Hi,

May we know the following details:

• What is the Linux flavour?  RHEL/SuSE/Ubuntu/etc
• What is the BIOS version you have updated to?
• What is the DIMM capacity you have in this server?
• Have you installed any PCI cards inside this server, is so, what are they?

Thank You!
I work with HPE but opinions expressed here are mine.
HPE Tech Tips videos on How To and Troubleshooting topics

[AurelienLoron]

Hi,

it is Proxmox VE 7.4, it is based on Debian 11.6 (“Bullseye”)

proxmox-ve: 7.4-1 (running kernel: 5.15.108-1-pve)
pve-manager: 7.4-16 (running version: 7.4-16/0f39f621)
pve-kernel-5.15: 7.4-4

i've update the BIOS in : P70 05/24/2019 (firmware-system-p70-2019.05.24-1.1.i386.rpm)

[Suman_1978]

Hi,

Thankyou for sharing this information.

Have you checked if Debian and the Kernel is actually supported on this server model?

According to the error, it looks like a compatibility issue between Debian and server bios.  There are some articles but they do not directly point to your issue.
https://techlibrary.hpe.com/us/en/enterprise/servers/supportmatrix/
https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=c03265132
https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=a00019143en_us
https://support.hpe.com/hpesc/public/docDisplay?docId=c00191707

Thank You!
I work with HPE but opinions expressed here are mine.
HPE Tech Tips videos on How To and Troubleshooting topics

[AurelienLoron]

@Suman_1978 

Thank you for your help, even if I still haven't found a solution.

I read here : https://wiki.debian.org/HP/ProLiant that Debian Bookworm have been install on a dl380 Gen8.

I read here : 2.5 Proxmox VE 7.2

• intel_iommu defaulted to on in kernels 5.15 based kernels up to pve-kernel-5.15.35-3-pve with version 5.15.35-6. An enabled IOMMU can cause problems with older hardware - issues were reported with, for example, HP DL380 g8 servers, and Dell R610 servers (both older than 10 years). The issue can be fixed by explicitly disabling intel_iommu on the kernel commandline (intel_iommu=off) following the reference documentation - https://pve.proxmox.com/pve-docs/chapter-sysadmin.html#sysboot_edit_kernel_cmdline As the setting has been reverted in newer pve-kernel-5.15 packages the issue is now mostly relevant during installation from the ISO.

Disabling IOMMU has no effect in my case.