Issue with display after switching from 3840x1600 to 3840x2160 display

cproo12 · 2024-03-04 16:46:21

Hello. I have been using & learning Arch since mid 2022. Love it so far, but last November, I switched around my setup and moved from an LG 38GL950G-B (3840x1600) to an LG 42C3 (3840x2160).

This issue has been around since November but I am finally just getting too annoyed with having to reboot my computer every time I turn off my monitor, leave for more than 10-15 minutes, and power it back on. Especially since it is an OLED!

to preface: I leave my computer powered on and logged in 24/7, as it just sits in my house, which is safe enough for me.

Say I come home from work, my monitor has been powered off since whenever the last time I used the computer was. Maybe 8 hours, maybe 4 hours. Maybe several days. The computer has been on, is logged in, and programs are running such as Steam auto-updating games, Transmission seeding torrents, or whatever else... but the display is powered off. I will power the display back on, and get no display output UNLESS I switch to a different tty.

Sometimes (maybe 1 out of 5 times) if I switch back to tty0 everything will come up, but most of the time, I have to enter tty1, login to my user acct or root, and systemctl reboot, otherwise I'll just never get into my computer.

Here is my Xorg.0.log.old (old because I had to restart my computer this morning to use it in the first place)

I am not sure what other logs or config files would be helpful to provide to troubleshoot this.

Another issue I am having is my mouse cursor and tty font is huge (some hidpi setting probably) but I tried changing a bunch of things to 96dpi (i use the display at 100% scaling as its 42" on my desk) and could not figure it out. Bonus points if someone can help with that one.

Thanks in advance

seth · 2024-03-04 17:39:16

[  1363.081] (--) NVIDIA(GPU-0): LG Electronics LG TV SSCR2 (DFP-0): 42666.7 MHz maximum pixel clock
[ 50862.990] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50863.016] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50863.016] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50868.437] (--) NVIDIA(GPU-0): LG Electronics LG TV SSCR2 (DFP-0): connected
[ 50868.437] (--) NVIDIA(GPU-0): LG Electronics LG TV SSCR2 (DFP-0): Internal TMDS
[ 50868.437] (--) NVIDIA(GPU-0): LG Electronics LG TV SSCR2 (DFP-0): 42666.7 MHz maximum pixel clock
[ 50890.225] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50890.225] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50890.225] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50890.277] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50890.382] (--) NVIDIA(GPU-0): DFP-0: disconnected
[ 50890.382] (--) NVIDIA(GPU-0): DFP-0: Internal TMDS
[ 50890.382] (--) NVIDIA(GPU-0): DFP-0: 165.0 MHz maximum pixel clock
[ 50892.410] (--) NVIDIA(GPU-0): LG Electronics LG TV SSCR2 (DFP-0): connected

The output seems to disconnect on probably DPMS events.

[    72.702] Kernel command line: BOOT_IMAGE=/vmlinuz-linux root=UUID=5412f30e-ec59-4e8a-b815-6a90e8fb9700 rw loglevel=3 splash

Enable https://wiki.archlinux.org/title/NVIDIA … de_setting - use the "nvidia_drm.modeset=1" kernel parameter (modprobe.conf won't do!), you can try nvidia_drm.fbdev=1 - with recent drivers the results are inconclusive. This will also expose the outputs edid in sys/class/drm what might be enough (or not)

[    72.703] (==) Using config file: "/etc/X11/xorg.conf"

Get rid f that file, it's nvidia-settings generated cruft.
You instead maybe want sth. like

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "GeForce RTX 3090"
EndSection

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Option         "ConnectedMonitor" "DFP-0"
    Option "MetaModes" "3840x2160_120 +0+0"
EndSection

However, the output seems properly re-detected in that log, the issue may just be some zealous output managing daemon in KDE or gnome.
=> What kind of GUI session do you run?

cproo12 · 2024-03-04 17:46:17

I just use awesomewm. Theres a chance some of my .xinit, .zprofile, et cetera might be screwed up because I don't know exactly what I'm doing.

Followed guides but it seems like there are many ways to do one thing.

seth · 2024-03-04 18:46:44

Please try the config to enforce the outputs presence and post your xinitrc.

cproo12 · 2024-03-05 00:51:31

Ok, I deleted the xorg.conf, and made a new file with the exact contents you suggested. I don't need to change "Device0" or anything, right?

I tried 'modprobe nvidia_drm modeset=1 fbdev=1" but "sudo cat /sys/module/nvidia_drm/parameters/modeset" still returns "N"- according to the wiki it should be returning Y, right? I will report back tomorrow morning after my display has sat turned off all night, usually that is when the issue occurs. Today when I came home from work, I was able to get back into my desktop by going to tty1 then back to tty0 instead of restarting.

Here is my .xinitrc

Here is my Xorg.0.log from today, where I was successfully able to switch from tty0 to tty1 and back to tty0.

seth · 2024-03-05 08:55:16

I tried 'modprobe nvidia_drm modeset=1 fbdev=1" but "sudo cat /sys/module/nvidia_drm/parameters/modeset" still returns "N"

Because hte module is probably in the initramfs, please use the kernel parameter -

seth wrote:

(modprobe.conf won't do!

Nothing wrong w/ your xinitrc, do you run picom w/ the glx or the xrender backend?

cproo12 · 2024-03-05 13:39:44

seth wrote:

Because hte module is probably in the initramfs, please use the kernel parameter -
seth wrote:
(modprobe.conf won't do!

Sorry, but I'm not following what you mean by this, is there a different command i should be using to modify the kernel parameter?

I use glx

seth · 2024-03-05 14:07:47

Module options are not https://wiki.archlinux.org/title/Kernel_parameters (linked in the paragraph in the linked nvidia wiki)
You can (and in this case "shall") use kernel parameters to set module options, though.

Edit: try the behavior of the xrender backend, glx is prone to texture corruption through VRAM decay, see https://wiki.archlinux.org/title/NVIDIA … er_suspend

Last edited by seth (2024-03-05 14:09:31)

cproo12 · 2024-03-05 14:26:46

My bad- having noob brain fart. I remembered that I should be editing /etc/grub/default "GRUB_CMDLINE_LINUX_DEFAULT=" and adding "nvidia_drm.modeset=1 & nvidia_drm.fbdev=1", to that, right?

Then running sudo grub-mkconfig -o /boot/grub/grub.cfg & rebooting my machine? going to try that now, I hope I am not being stupid. Sorry just haven't messed with that in a long time so not fresh in my mind!

also note- this morning I was able to get into my desktop by going to tty1 then back to tty0, but again it didn't show up normally upon powering my display on.

edit- "sudo cat /sys/module/nvidia_drm/parameters/modeset" now returns Y, so glad i finally figured that out. Going to go to work and see if the issue is resolved in 8-10 hours when I get home

edit 2- i just forgot something and turned my display back on, (it has been about 12 minutes) and it is still doing the same thing. maybe something to do with sleep? and the PC not coming out of sleep? Going to try disabling DPMS since I don't really care about the sleep functionality anyway (my LG display has it's own annoying screen blanking feature that cannot be disabled)...

Last edited by cproo12 (2024-03-05 14:46:43)

seth · 2024-03-05 15:30:20

Does the system suspend or is it only DPMS?
And please post an updated xorg log.

cproo12 · 2024-03-06 02:40:13

Home from work after full 12 hours of leaving it (and I disabled DPMS before I left, then restarted my PC, xset -q confirmed "Server does not have the DPMS Extension"

Still- "No signal" upon turning on display until I switch to tty1 and back to tty0. So odd!

Here is the Xorg.0.log.

Another point of confusion- it says "Server terminated" at the end of the log file, but I am still actively using the X session, is this even the correct log to be using? /var/log/Xorg.0.log was the current active tty0 session, correct?

seth · 2024-03-06 08:00:41

[   138.009] (==) Log file: "/var/log/Xorg.0.log", Time: Mon Mar  4 10:59:58 2024

And it closed ~22.5h later
Also

[   139.157] (**) NVIDIA(0): DPMS enabled
[   139.158] (II) Initializing extension DPMS

and neither of the nvidia_drm options is present (but the simpledrm device is)

If you're starting w/ xinit/startx, the log is in ~/.local/share/xorg/

cproo12 · 2024-03-06 13:49:40

Ugh I've been uploading the wrong logs then. Have a look at this one. Still had to tty1 back to 0 this morning to get it back on. The only reason it bugs me is because when I switch ttys sometimes my applications crash.

Xorg.0.log:

Xorg.1.log:

seth · 2024-03-06 14:18:32

[   105.257] (==) NVIDIA(0): DPMS enabled

Still had to tty1 back to 0 this morning to get it back on.

Ftr, this is progress, right? It now works reliably and you don't have to reboot anymore?
(Just so we're not chasing the wrong rabbit down his hole)

The only reason it bugs me is because when I switch ttys sometimes my applications crash.

Mind to elaborate? What applications? All? (ie. does the displayserver crash)?

Home from work after full 12 hours of leaving it (and I disabled DPMS before

Since you disable DPMS, I guess you're switching off the monitor?
Is DPMS actually the problem here?
Re-enable it and then run "xset dpms for off", get a cup of coffee, then wiggle the mouse… does the output revive?

cproo12 · 2024-03-06 15:27:02

Sorry, I've left for work for the day but let me answer questions still to be helpful.

Having to reboot the computer seems like it only recently became an issue, and it happens kind of intermittently. It has not happened since I posted this thread after having to reboot. I bet in the future it will happen again. It might depend on which applications I leave open.

Apps that crash or stop working properly after switching displays: Firefox, Discord, Steam (and thus any Steam games I may have left open), and Mullvad's graphical UI program.

when I say I disable dpms, I mean I create /etc/X11/xorg.conf.d/dpms.conf with contents something like:

Section "Monitor"
    Option "DPMS" "false"
EndSection

I remember accidentally blanking the screen with DPMS in the past, and it definitely wakes back up if i shake the mouse. I did this on accident when trying to disable DPMS, typed the wrong command. Now if I typed the command I bet it would just tell me its unavailable, since DPMS is outright disabled. (according to xset q)

Last edited by cproo12 (2024-03-06 15:28:29)

seth · 2024-03-06 15:48:38

blanking the screen with DPMS in the past, and it definitely wakes back up if i shake the mouse

=> See what happens if you re-enable DPMS, dpms the monitor off but keep the power on and then wake the system after a while.

Do the crashing clients crash w/ a SIGSEG or SIGILL or do they "crash" in the way that the windows just display black or garbled content?
In the latter case, can you "uncrash" them by resizing the window?

cproo12 · 2024-03-06 16:13:28

Depends- Sometimes steam closes outright (usually closing games with it), sometimes the UI just stops responding, playing with window geometry, or even reopening does not fix it, but steam is pretty buggy on linux. Especially in awesomewm. So I wouldn't worry about that...

Discord 100% of the time just closes as soon as I exit the WM tty

Firefox sometimes stays open, or it will just close like discord, saying "Firefox quit unexpectedly" after going back in. I don't really care about this, because I never typically need to swap ttys. If I can fix the original problem, I won't need to fix this one.

I'm not sure how to tell what kill signal the programs crashed with, but if you think that's important I can figure it out.

I will try DPMS blanking the display when I get home from work today & provide (the correct) xorg log

Thanks for all the help, by the way. I really appreciate your time.

cproo12 · 2024-03-07 00:49:34

Update- home and still had to switch ttys with DPMS fully disabled, though that was the same behavior as yesterday.

Xorg.0.log

Xorg.1.log

Restarted, re-enabling the DPMS module, but set all timeouts to 0 because I don't want DPMS to do anything (confirmed this is all set properly by xset q)

Blanked the screen, walked off for like 10 minutes, came back shook mouse and it woke right up.

Updated pastebin log after restart in case its even useful

The one thing I notice, in the logs (pre-restart), is line 158, 757, 1009, 1203... Not sure if its relevant but pops up each time I power on the display it seems... based on time stamps.

[ 82148.246] (II) NVIDIA(0): ACPI: failed to connect to the ACPI event daemon; the daemon

seth · 2024-03-07 15:59:04

Blanked the screen, walked off for like 10 minutes, came back shook mouse and it woke right up.

Ok, most likely the only problem is that you switch off the monitor w/ the power button.
You could use dpms for this (or actually S3/4 the system while you're not there - makes Greta happy and cuts the signal from the output what should™ send it to DPMS anyway)

For DPMS off, just a shortcut or icon or whatever w/ "xset dpms off" should usually do - if you're worried that the output gets triggered (spurious input because an earthquake shakes your mouse or so) you could also make that a toogle for an aggressive DPMS setting (so the ouput automatically powers off after 30 seconds unless you "enable" it)

Did we ever try to add the EDID to the drm or the X11 server?
(Sorry, there're two of you and I can't keep apart where we tested what
https://bbs.archlinux.org/viewtopic.php … 2#p2155022

cproo12 · 2024-03-07 16:30:28

Understand. Although I'd like to fix the weird monitor issue, that sounds like an equally good solution.

I never tried adding the EDID and I won't lie I have no idea how. I will try to figure that out tonight and set up a DPMS script as a jerry rig fix.

The only annoying thing about this display is, when I DPMS screen blank, the display stays on for like 5 minutes afterwards doing a stupid LG slideshow through stupid LG pictures. No way to turn it off (beside powering the display off which obviously isn't working too well for me )

I don't want this when I'm going to bed, because my computer lives in my bedroom.

seth · 2024-03-07 21:19:23

display stays on for like 5 minutes afterwards doing a stupid LG slideshow through stupid LG pictures

WTF…
https://www.lgwebos.com/topic/8376-powe … port-dpms/
Sure there isn't a way to configure the input type and tell it that "this is from a PC"?

About the EDID, the linked thread addresses that, ask back if things are unclear.

cproo12 · 2024-03-11 01:51:43

Apologies for letting the thread die a bit. Maybe it was a nice break for you seth I was travelling.

I just copied the EDID ($ sudo cp /sys/class/drm/card0-HDMI-A-1/edid /usr/lib/firmware/edid/lg42c3_edid.bin) and then pointed my Screen section of my xorg.conf to the new .bin file.

Going to upgrade my system as I am several days out of date and reboot, will check back in tomorrow before work in the morning and see if the issue is resolved...

seth · 2024-03-11 09:33:23

If not, attempt to directly add it to the kernel instead of the xorg server.

cproo12 · 2024-03-15 14:22:02

Sorry for the delay's- I have been doing a bit of troubleshooting on my own after reading my ring buffer and journal.

I was getting crazy error spam in my journal from nitrogen (i had a cron job setup to run every 5 minutes to pick a random wallpaper, and if it ran while the display was turned off, it would output hundreds of lines of errors)

I guess I didn't find any real concerning kernel messages but I never thought to check that. Pretty much the entire ring buffer is filled with cron job messages every 5 minutes when it sets a new wallpaper.

I also added the edid to my kernel parameters and it still did not make a difference.

Here are all the logs:

journalctl (pastebin didn't allow this huge file)

ring buffer

Xorg.0.log

Xorg.1.log

might as well also upload some other files I could have possibly screwed up at some point too:

.xinitrc

.zprofile

.zshrc

if there are any other config files I could have screwed up, I can upload those too... not sure where to go next. I guess I could try a brand new arch + awesomeWM install and see if the problem still happens, but that wouldn't tell me *why* it happens

Last edited by cproo12 (2024-03-15 14:27:21)

seth · 2024-03-15 22:31:17

pastebin didn't allow this huge file

https://0x0.st - pastebin is discouraged anyway because it's not globally accessible and justpaste.it is justhorrible (I can't even see how to get the raw text and I cannot check 2MB of logs in a browser text display)

Xorg.0.log shows no issues, Xorg.1.log is vastly dated.
The kernel commandline doesn't show the edid there, though - it's however in the xorg config and used.
Your xinitrc looks fine.

Get used to the idea to leverage DPMS - does the monitor have a displayport input?

Arch Linux

#1 2024-03-04 16:46:21

Issue with display after switching from 3840x1600 to 3840x2160 display

#2 2024-03-04 17:39:16

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#3 2024-03-04 17:46:17

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#4 2024-03-04 18:46:44

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#5 2024-03-05 00:51:31

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#6 2024-03-05 08:55:16

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#7 2024-03-05 13:39:44

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#8 2024-03-05 14:07:47

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#9 2024-03-05 14:26:46

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#10 2024-03-05 15:30:20

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#11 2024-03-06 02:40:13

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#12 2024-03-06 08:00:41

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#13 2024-03-06 13:49:40

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#14 2024-03-06 14:18:32

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#15 2024-03-06 15:27:02

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#16 2024-03-06 15:48:38

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#17 2024-03-06 16:13:28

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#18 2024-03-07 00:49:34

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#19 2024-03-07 15:59:04

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#20 2024-03-07 16:30:28

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#21 2024-03-07 21:19:23

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#22 2024-03-11 01:51:43

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#23 2024-03-11 09:33:23

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#24 2024-03-15 14:22:02

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

#25 2024-03-15 22:31:17

Re: Issue with display after switching from 3840x1600 to 3840x2160 display

Board footer