2x boot issues: AMD Ryzen 9 5900HX + NVIDIA RTX 3080 Mobile

Issues & Assistance
1

Hi all,

I’d appreciate your help with my problems running cachyos. I am new to this OS but been running arch for several years (and was running on this machine since 2021).

Two separate issues, not sure where one ends and next begins:

  1. cannot boot with my power cable inserted in the laptop (it has to be disconnected, I can re-connect after the machine has booted)

  2. cannot boot without the kernel parameter nomodeset=1 (or it will flake out with a black screen before the login prompt)

I think problem 1 is some sort of acpi issue. This problem has been plaguing me for longer, it also affects other arch distributions and their usb install disks. I can boot usb-installers only if I disconnect the power and set nomodeset=1, just like the installed kernel.

Based on the location(s) of freezing (often, but not always, freezes on message: “BIOS 113-CEZANNE-017”), I suspect problem 2 is an amdgpu issue. This doesn’t come up during nomodeset=1 boots, so it seems a plausible place to start looking.

So far I have tried (separately and in some cases combined):

rcutree.gpu_init_delay=1

amdgpu.dc=1 amdgpu.vm_update_mode=3 amdgpu.runpm=0 amdgpu.dpm=0 amdgpu.dcdebugmask=0x12 amdgpu.modeset=1 (and 0)

iommu=off iommu=gp

and many more, also for NVIDIA but nvidia graphics seem to be working properly.

Errors in dmesg:

[0.319590] ACPI: [Firmware Bug]: BIOS _OSI(Linux) query ignored

When I set acpi=off I do successfully boot, though it immediately initialises shut down again once it has fully booted (cleanly).

Booting with acpi_osi=! acpi_osi=”Windows 2017” as a kernel parameter (manually edited through ‘e’ in limine bootloader) has no effect.

processor.max_cstate=0 no effect, neither to other cstate settings, no difference

[0.324610] PCI: Using host bridge windows from ACPI; if necessary, use “pci=nocrs” and report a bug

Tried this setting, but it just borks my nvidia driver, generating a ‘failed to allocate NvKmsKapi Device’ error that disappears when pci=realloc or is removed altogether.

[5.013345] amdgpu 0000:05:00.0: probe with driver amdgpu failed with error -22
[1.461662] asus_wmi: fan_curve_get_factory_default (0x00110032) failed: -19

The top error I have not really found anything for, the bottom error is apparently generated if you try to set fan curves without having fan control? Thi os second one does not really seem to be a problem.

Bugreport with LTS and nomodeset and older nvidia 580 drivers is here:

Bugreport with regular kernel and nomodeset and nvidia 595 drivers is here:

I appreciate any second pair of eyes on this (in the hope it’s not hardware).

Thanks in advance!

2

Issue 2 is definitely amdgpu, I can modprobe.blacklist=amdgpu and boot without issue 2 (issue 1 remains) and when I run ‘sudo modprobe amdgpu’ I have the exact issue.

Not sure which flags amdgpu is expecting from me to be able to work, or how I would figure out whether this is actually a hardware issue. Continuing to investigate.

With the module blacklisted at boot, my most recent bugreport is here: a697412

3

Stupid question but have you updated the BIOS?

4

Not a stupid question at all, I appreciate the interaction. Afaik I am on the latest version of my BIOS :

Platform Firmware Information
Vendor: American Megatrends International, LLC.
Version: G533QS.335
Release Date: 11/07/2025

But I will doublecheck

5

Issue #2 is a kernel/amdgpu module bug. A temporary workaround is to close the lid of the laptop once the black screen is encountered, and then open it again: Ran into a frozen black screen issue with latest update

6

Thank for linking to this issue, but it doesn’t make a difference in my case, the display stays frozen when I close and re-open the laptop lid.

Neither does adding the video mode as a kernel parameter as suggested here: [SOLVED] Black screen for NVIDIA and AMD hybrid graphics / Kernel & Hardware / Arch Linux Forums

I wouldn’t mind so much (the nvidia driver works fine) but it means my screen doesn’t show up properly (for example) under xrandr which means I can’t do things like set the brightness:

❯ xrandr -d :0
Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 4096 x 4096
None-1 connected primary 1920x1080+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
1920x1080     60.00*+
7

[brief update]

I think there’s a driver issue, but I have not figured out what it is precisely. Switcheroo gives a drm error. It’s the only error clearly associated with the screen freeze (journalctl | grep amdgpu after setting systemd.log_level=debug during boot) I have really been able to find.

❯ /usr/lib/switcheroo-control-check-discrete-amdgpu /dev/dri/card0
_amdgpu_device_initialize: DRM version is 1.0.0 but this driver is only compatible with 3.x.x.

~
❯ /usr/lib/switcheroo-control-check-discrete-amdgpu /dev/dri/card1
_amdgpu_device_initialize: DRM version is 0.0.0 but this driver is only compatible with 3.x.x.

~
❯ /usr/lib/switcheroo-control-check-discrete-amdgpu /dev/dri/renderD128
_amdgpu_device_initialize: DRM version is 0.0.0 but this driver is only compatible with 3.x.x.
8

[update]

I think it’s a power issue of some kind. If I use asusctl profile set quiet, followed by sudo modprobe amdgpu, the screen sets itself up correctly and I can startx and have both gpu’s available and my screen registered as eDP. It will still freeze within a couple of minutes (without error besides those in the post above if I journalctl -f >> logfile.txt ) with or without X. The journalctl also stops dead mid-message, so it’s a full freeze, not just the graphics.

9

The latest kernel (7.0.3) has resolved the black screen issue. Try updating and see if it changes anything.

10

This did not work for me (probably since it is a different issue, as per the previous example where closing and re-opening the lid did not make a difference either).