Arrow Lake-HX Boot Loop with Qubes OS - APIC ID Mismatch and Xen Issues

User Support Hardware Issues
1

I’m attempting to install Qubes OS on a system with the following specifications:

Hardware
CPU: Intel Arrow Lake-HX (8P+16E cores)
Chipset: Intel 800 Series (WM880) [8086:7d1c]
GPU: Intel Arrow Lake-S Graphics [8086:7d67] & NVIDIA GeForce RTX PRO 5070Ti [10de:2db8]

What I’ve tried
Initially installed Qubes R4.2.4. Installation went fine, but didn’t want to boot becuase sys-usb fails and on the sys-net the WiFi PCI gives problems. After removing the PCI from sys-net and setting sys-usb to not autostart the system boots but won’t start X (I never even see the LightDM screen). This is because the nouveau doesn’t recognize my videocard. Tried different solutions couldn’t get one to work.
Tried updating dom0 to the latest version; that will result in a bootloop.

Then went on to install R4.3-rc3 that also goes into a bootloop. So far the only kernel that boots fine is the 6.6.* that is default to 4.2.4.

As that also failed I’ve tried building my own ISO; with some minor tweaks and fixes I got that working and updated the kernel to the latest version that qubes supports. But that also fails…

Identified errors
After booting a custom build R4.3 version I’ve added params to GRUB to halt the boot process and dump me into a console so that I could see the dmesg output clearly. From there I could identify some errors:

Firmware/APIC Issues

[Firmware Bug]: CPU 23: APIC ID mismatch. CPUID: 0x0017 APIC: 0x0058
[Firmware Bug]: CPU 23: APIC ID mismatch: Firmware: 0x004e APIC 0x0058

Nouveau Driver Failure

nouveau 0000:02:00.0: unknown chipset (1b5000a1)

USB HID Error

usbhid 3-7-1.0: couldn't find an input interrupt endpoint

tmpfs Parameter Issue

tmpfs: Unsupported parameter huge

Multiple PCI Bridge Window Errors
Extensive PCI resource allocation errors for 64-bit prefetchable memory ranges across multiple bridges.

ACPI Table Issues
Multiple ACPI-related warnings and errors during boot, though the system appears to parse ACPI tables.

I’ve also tried to upgrade xen to a newer version but that seems to be a bridge to far for my current Qubes knowledge.

Working Intel Core Ultra
I’ve found this HCL that seems to be able to get somethings working with the same family of processor.

Edit:
Fedora workstation 43 works flawlessly on this hardware

Any guidance would be greatly appreciated! I understand Qubes may need updates to support it fully. I’m willing to test patches or specific configurations if the development team, or anyone else, has suggestions.

Kind regards, Bloged

1 Like
2

Little update:

After investigating what is going on with 4.3 when installing I’ve booted it with console=vga noreboot multiboot and removed the quiet part from the kernel line. That shows me the following error:

BUG: unable to handle page fault for address: ffffe90d4c1f0083
RIP: supervisor read access in kernel mode
intel_pmc_ssram_telemetry_probe:0xa7/0xe0 [intel_pmc_ssram_telemetry]
Kernel panic - not syncing: Fatal exception

That is strange as it seems that the kernel that is being used in 4.3 does not include CONFIG_INTEL_PMC_SSRAM_TELEMETRY; when looking at the source code. I’ve added that to kernel-latest as module in qubes-builderv2 and upgrade the kernel to 6.17.11.

After booting that ISO I can reach the installer with GRUB kernel parameters: nomodeset i915.modeset=0 nouveau.modeset=0 intel_iommu=off intel_pmc_core.disable=1 initcall_blacklist=intel_pmc_ssram_telemetry_driver_init module_blacklist=intel_pmc_ssram_telemetry debug loglevel=7. That brings me to the installer.

The installer that gives the error:

X startup failed, aborting installation
Failed to adjust memory pressure threshold, ignoring: Device or resource busy

Adding inst.text to my kernel parameters shows me an installer that won’t install as the HD shows the exclamation mark; and I’m not able to continue.

Any help or pointers would be welcome!

Kind regards, Bloged

1 Like
3

Another update:

Just using intel_pmc_core.disable=1 initcall_blacklist=intel_pmc_ssram_telemetry_driver_init module_blacklist=intel_pmc_ssram_telemetry debug loglevel=7 on a custom build r4.3 with kernel 6.17.11 get’s me to the installer. Installed just fine… just boots me in an empty dom0.

Will test r4.3-rc4 with these params.

1 Like
4

Probably final update:

I can install r4.3-rc4 with intel_pmc_core.disable=1 initcall_blacklist=intel_pmc_ssram_telemetry_driver_init module_blacklist=intel_pmc_ssram_telemetry.

This version of the kernel solves the nouveau problem I had before not recognizing my graphics card.

Now back to one of the original problems I had with 4.2.4.

Qubes sys-net failed to start. Requested operation is not valid: PCI device 0000:80.14.3 is not assignable.

The log file at /var/log/libvirt/libxl/libxl-driver.log is empty.

Same goes for sys-usb with a different PCI ID 80:14.0.

Let me check the rest of the system and then create a HCL report for this laptop.

1 Like