black screen on boot forced to tty after trying to use nvidia gpu only

berd · 2024-09-03 11:19:18

Hello, I've been using arch for a while now, I ain't a stranger to using guides on the arch wiki to troubleshoot and I usually can figure stuff out by myself, however I encountered a strange issue recently while following an official arch guide on how to only use the nvidia gpu on my primus capable laptop (here)

for context, I'm using the classic sddm and kde6-wayland combo

What I did was quite short, I created the file mentioned in the guide, added all of the text to it, then copied the .xinitrc from /etc/X11/xorg.conf.d/10-nvidia-drm-outputclass.conf to ~/xinitrc, added the two lines mentioned in the guide, restarted, and thus started the issue of me booting to a black screen and having to manually tty and using startplasma-wayland to access my DE, however in doing that it works perfectly normally.

Since then I've tried many things, deleting the .xinitrc to use the default one, make a new .xinitrc setting the session to plasma and then doing exec startplasma-wayland, re-enabling systemctl sddm, nothing changes, still booting to a black screen, I just want to boot into sddm then into plasma as normal.

it's a bit of a bother to have to tty into my DE and not having access to my DM as usual, and I've rarely been stumped by an issue with arch like that before, I'd really appreciate your help in the matter, even though I'm fairly certain it's going to be a trivial issue, cheers !

Lone_Wolf · 2024-09-03 11:50:36

Did you follow the first link in that section and add nvidia_drm.modeset=1 as a kernel parameter to your boot command line ?

(adding it through a file in modprobe.d is NOT enough)

V1del · 2024-09-03 15:13:29

That article is completely irrelevant if you intend to use wayland, and you also can't (and shouldn't want to) try to use the nvidia driver only if you actually have an Optimus setup.

Your intel GPU is the one that the screen is attached to unless you have an external monitor that's directly tied to the nvidia gpu, so any image that the nvidia card would render needs to pass through the intel gpu - always. Trying to bend this around will in the majority of cases lead to generally worse performance. Use prime-run on an as needed basis.

.xinitrc from /etc/X11/xorg.conf.d/10-nvidia-drm-outputclass.conf

that file is not an .xinitrc and that would be wrong in any case, but SDDM shouldn't be using your .xinitrc regardless

You probably did something else you're not reflecting here, can you switch the TTY when in the "black screen" situation (Ctrl+Alt+F2..F3 and so forth)? if you can and you have an internet connection post

sudo journalctl -b | curl -F 'file=@-' 0x0.st
cat /var/log/Xorg.0.log | curl -F 'file=@-' 0x0.st

berd · 2024-09-03 16:26:50

V1del wrote:

That article is completely irrelevant if you intend to use wayland, and you also can't (and shouldn't want to) try to use the nvidia driver only if you actually have an Optimus setup.
Your intel GPU is the one that the screen is attached to unless you have an external monitor that's directly tied to the nvidia gpu, so any image that the nvidia card would render needs to pass through the intel gpu - always. Trying to bend this around will in the majority of cases lead to generally worse performance. Use prime-run on an as needed basis.

oh ok, I wanted to try using my nvidia gpu only because I had issues with primus on an older computer and for some reason didn't want to troubleshoot this on my new laptop, I knew it wasn't as good a choice for the limitations you pointed out, but I really needed to make sure that my gpu heavy applications didn't use the intel integrated chip as they so often tend to use, duly noted, I'll instead set up ways to prime run whatever I need run with the nvidia gpu.

V1del wrote:

.xinitrc from /etc/X11/xorg.conf.d/10-nvidia-drm-outputclass.conf
that file is not an .xinitrc and that would be wrong in any case, but SDDM shouldn't be using your .xinitrc regardless

oh sorry, I copied the wrong command line, I was in a hurry before work, the xinitrc I copied was located in /etc/X11/xinit/xinitrc but as you said, I don't think it should have any impact since I deleted that new .xinitrc anyway and it should fallback to the default one.

V1del wrote:

You probably did something else you're not reflecting here

Most likely, I tried my best to remember everything I did and checked my previous console commands, but it did happen at almost 3 am, sorry for that, imma tty again and post the result in a moment

berd · 2024-09-03 17:04:57

oh, well I guess I'll just become a case for science heh, I remembered that the file the guide told me to modify in /etc/X11/xorg.conf.d/10-nvidia-drm-outputclass.conf, which I initially tried to modify using nano, didn't exist, and I didn't remove it, I just did and sddm started on reboot, plasma too, but the weird thing is that plasma X11 which previously gave me a black screen and kicked me back to sddm, is now the one that works and plasma wayland kicks me back, now I should be exactly back where I should have been before following the guide half assedly so it is a little puzzling but as long as it works I ain't complaining.

Regardless, I'm closing this topic, mark it up as another arch newbie doing their thing and not reading the entire documentation while trying to do something while exhausted in the middle of the night heh.

Thank you for your time and contribution, it is very much appreciated and I wish yall a very pleasant day !

Arch Linux

#1 2024-09-03 11:19:18

black screen on boot forced to tty after trying to use nvidia gpu only

#2 2024-09-03 11:50:36

Re: black screen on boot forced to tty after trying to use nvidia gpu only

#3 2024-09-03 15:13:29

Re: black screen on boot forced to tty after trying to use nvidia gpu only

#4 2024-09-03 16:26:50

Re: black screen on boot forced to tty after trying to use nvidia gpu only

#5 2024-09-03 17:04:57

Re: black screen on boot forced to tty after trying to use nvidia gpu only

Board footer