Problem during installation of Qubes 4.0.2

[Fabian]

Hello,

I recently bought new hardware and tried to install Qubes on it, sadly without any success at all.

I tried it with UEFI only and with CSM (Compatibility Support Module) only, but neither had worked.

The media I used is the recent Qubes 4.0.2 ISO, I had the same issues 2 weeks ago with the Qubes 4.0.2 RC3 image.

I copied it to the USB Stick with Rufus (In both MBR/CSM Mode and GPT/UEFI Mode) as well as just copying it over to the USB drive and then extract it.

First of all, the Hardware:

CPU: AMD Ryzen 9 3950X

GPU: AMD Radeon RX 5700XT

Mainboard: ASUS TUF Gaming X570-Plus WiFi

I made the mainboard decision because of the price and the good VRMs on the board - most other "cheap" x570 mainboards are having big problems with VRM temperatures.

Other parts shouldn't play a significant role here, because I didn't come far enough.

UEFI Problem:

When I boot the USB Stick in UEFI Mode, the screen simply stays blank after a couple lines of output appeared on the screen.

I then tried what's written on the "UEFI Troubleshooting" site: https://www.qubes-os.org/doc/uefi-troubleshooting/

The output which appears then is visible in the attachment "UEFI.jpg". I also saved the logfile, the last lines are the following:

[    9.681987] localhost kernel: usb 3-3.3: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[    9.682372] localhost kernel: usb 3-3.3: Product: USB2.0 Hub            
[    9.682754] localhost kernel: usb 3-3.3: Manufacturer: VIA Labs, Inc.        
[    9.724854] localhost kernel: hub 3-3.3:1.0: USB hub found
[    9.725585] localhost kernel: hub 3-3.3:1.0: 4 ports detected
[    9.792658] localhost kernel: usb 3-6: new full-speed USB device number 6 using xhci_hcd
[    9.933606] localhost kernel: usb 3-6: config 1 has an invalid interface number: 2 but max is 1
[    9.933988] localhost kernel: usb 3-6: config 1 has no interface number 1
[    9.945601] localhost kernel: usb 3-6: New USB device found, idVendor=0b05, idProduct=18f3, bcdDevice= 1.00
[    9.945983] localhost kernel: usb 3-6: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[    9.946358] localhost kernel: usb 3-6: Product: AURA LED Controller
[    9.946732] localhost kernel: usb 3-6: Manufacturer: AsusTek Computer Inc.
[    9.947100] localhost kernel: usb 3-6: SerialNumber: 9876543210
[    9.965764] localhost kernel: hid-generic 0003:0B05:18F3.0006: hiddev97,hidraw5: USB HID v1.11 Device [AsusTek Computer Inc. AURA LED Controller] on usb-0000:07:00.3-6/input2
[  134.331670] localhost dracut-initqueue[703]: Warning: dracut-initqueue timeout - starting timeout scripts
[  134.861553] localhost dracut-initqueue[703]: Warning: dracut-initqueue timeout - starting timeout scripts
[  135.382539] localhost dracut-initqueue[703]: Warning: dracut-initqueue timeout - starting timeout scripts

I saved the full Logfile of course, but I prefer to not poste it completely public on this list. So if any developer wants it for further inspection, please just let me know, I mail it to you then.

This all happened before I was able to see the menu (Test this media and Boot, Boot, Troubleshooting).

Then I tried the CSM option in the BIOS. I was at least able to pick an option in the boot menu (Test this media and Boot, Boot, Troubleshooting), which then quickly ended in the screenshot "CSM.jpg" in normal boot and also in basic graphics mode form Troubleshooting.

The attachments "CSM_anaconda.log" and "CSM_X.log" are from the installation in CSM Mode.

Any ideas what the problem could be, and/or how to fix it?

The hardware is pretty new (GPU released in August 2019, CPU in December 2019, most mainboards with X570 Chipset in July 2019. So I suspect it's a driver issue.

Also a hint from me: I tried to install Debian and Fedora as well, neither has worked. Seems like only Windows runs on it for now.

I somehow expected that already, but maybe my logs help to identify the problem so this hardware gets usable sooner for others as well.

[Claudia]

January 4, 2020 12:28 PM, "Fabian" <fabian...@gmail.com> wrote:

> Hello,
>
> I recently bought new hardware and tried to install Qubes on it, sadly without any success at all.
>
> I tried it with UEFI only and with CSM (Compatibility Support Module) only, but neither had worked.
>
> The media I used is the recent Qubes 4.0.2 ISO, I had the same issues 2 weeks ago with the Qubes
> 4.0.2 RC3 image.
>
> I copied it to the USB Stick with Rufus (In both MBR/CSM Mode and GPT/UEFI Mode) as well as just
> copying it over to the USB drive and then extract it.
>
> First of all, the Hardware:
>
> CPU: AMD Ryzen 9 3950X
>
> GPU: AMD Radeon RX 5700XT
>
> Mainboard: ASUS TUF Gaming X570-Plus WiFi
>
> I made the mainboard decision because of the price and the good VRMs on the board - most other
> "cheap" x570 mainboards are having big problems with VRM temperatures.
>
> Other parts shouldn't play a significant role here, because I didn't come far enough.
>
> UEFI Problem:
> When I boot the USB Stick in UEFI Mode, the screen simply stays blank after a couple lines of
> output appeared on the screen.
> I then tried what's written on the "UEFI Troubleshooting" site:
> https://www.qubes-os.org/doc/uefi-troubleshooting

Hmm, if you couldn't get the latest Fedora or Debian to run on it, there's probably not much hope for Qubes any time soon. But here are some ideas anyway.

Especially on newer hardware, UEFI mode is strongly recommended if you can get it to work, although there's no guarantee. If you want to try the UEFI route: Try setting rd.debug in the kernel command line, and note the last few lines. This should show you the exact shell command that is timing out. See also: https://fedoraproject.org/wiki/How_to_debug_Dracut_problems#Summary_of_dracut_kernel_command_line_options

Otherwise... Try the "nomodeset" kernel parameter which can often fix X problems. It's worked for me many times. If it works, after installing and updating you can work on getting the graphics driver working later. There's also a text-mode installer ("inst.text" parameter) but I've never tried it and I don't know if it works. https://github.com/QubesOS/qubes-issues/issues/4879

Some generic advice, some users with newer hardware have had to install Qubes on an older machine, update and install kernel-latest, and then move the drive/filesystem to the target machine. https://www.qubes-os.org/doc/software-update-dom0/#kernel-upgrade

Also, you can try installing the 4.0.1 ISO, or even 4.0, and then install updates, at which point it becomes the same as 4.0.2. I had an AMD integrated graphics issue that only affected the 4.0.2 installer but not 4.0.1. You can also try the 4.0.2 RCs as they might have slightly different kernel versions.

Finally, you can try a bleeding-edge Qubes R4.1 pre-release build which has Xen 4.13 and Linux 5.4. Click assets tab for iso: https://openqa.qubes-os.org/tests/5259