You are not logged in.
Yesterday I did full update on my system and installed several new packages. I experienced this 3 times already but what happens is when you start the computer, go past the login screen after typing your password the screen is just black without anything on the screen. How to reproduce:
1. Start PC
2. Type your Password
3. Select Gnome on Wayland
4. Press the Login button
5. Screen is black, no mouse or anything
Then I do:
1. Press the power button to stop PC
2. Press the power button again to start PC
3. Type your password
4. Select Gnome on XORG
5. Login
6. Works
Then I can shut down PC and login via Gnome on Wayland without issues for a while
I found this message in journalctl, I don't know if its related
Feb 22 13:34:48 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:48 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:48 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:48 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:48 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:48 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:48 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:48 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:53 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:53 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:53 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:53 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:54 arch systemd[669]: Started Application launched by gnome-shell.
Feb 22 13:34:54 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:54 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:54 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:54 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failed
Feb 22 13:34:54 arch systemd[1]: Starting Time & Date Service...uname -a
Linux arch 6.7.5-arch1-1 #1 SMP PREEMPT_DYNAMIC Sat, 17 Feb 2024 14:02:33 +0000 x86_64 GNU/Linuxgnome-shell --version
GNOME Shell 45.4Last edited by dab9966 (2024-02-23 16:30:19)
Offline
Any news on this? Because I have the same problem, in another thread someone mentioned to unplug the webcam, after two reboots it worked but it stopped working again and I have to use the Xorg session again.
6.7.5-arch1-1
GNOME Shell 45.4
NVIDIA-SMI 545.29.06
GDM 45.0.1Last edited by micronetic (2024-02-23 07:51:15)
Offline
Told you it's a fluke ![]()
I found this message in journalctl, I don't know if its related
Us neither, random lines out of context are pretty much meaningless, but nullptr resolutions are never good.
Please post your complete system journal for the boot:
sudo journalctl -b | curl -F 'file=@-' 0x0.st@micronetic, all of it, incl. the /etc/udev/rules.d/61-gdm.rules mask
For https://wiki.archlinux.org/title/NVIDIA … de_setting make sure to use the nvidia_drm.modeset=1 kernel parameter (not a modprobe config) to get rid of the simpledrm device.
And in dout also post a journal as well.
Offline
Told you it's a fluke
Ah yes, completely forgot. This stands out so much.
Anyway, black screen didn't happen since the time I posted this message originally. Here is log
Offline
I actually meant micronetic and the webcam.
The errors you posted are because you're not using GDM as the gnome developers envisioned you to do.
Also seems /I/ only "envisioned" having posted https://wiki.archlinux.org/title/GDM#Wa … DIA_driver - you'll still need those services and probably also add the gdm udev mask.
Offline
Told you it's a fluke
nvidia_drm.modeset=1 kernel parameter (not a modprobe config) to get rid of the simpledrm device.
Yeah it was a fluke ![]()
Wait a moment, I see you and the Wiki says there is an underscore, I had this in my Kernel parameters for a very long time:
nvidia-drm.modeset=1so I changed it to the underscore.
Those were my steps for enabling Wayland:
sudo nano /etc/default/grub
nvidia_drm.modeset=1
sudo grub-mkconfig -o /boot/grub/grub.cfg
sudo ln -s /dev/null /etc/udev/rules.d/61-gdm.rulesStill not working, here is my journal: https://pastebin.com/80pgksVP
Last edited by micronetic (2024-02-23 10:12:03)
Offline
You're facing a very different situation from the OP
Feb 23 10:57:19 emnet systemd-coredump[1021]: Process 947 (mutter-x11-fram) of user 120 dumped core.
Stack trace of thread 947:
#0 0x00007d7c1f64332c n/a (libc.so.6 + 0x8d32c)
#1 0x00007d7c1f5f26c8 raise (libc.so.6 + 0x3c6c8)
#2 0x00007d7c1f5da4b8 abort (libc.so.6 + 0x244b8)
#3 0x00007d7c1ba35f68 n/a (libvulkan.so.1 + 0x11f68)
#4 0x00007d7c053ed8b1 n/a (zink_dri.so + 0x9ed8b1)
#5 0x00007d7c04a3b11e n/a (zink_dri.so + 0x3b11e)
#6 0x00007d7c0501dee8 n/a (zink_dri.so + 0x61dee8)
#7 0x00007d7c05039ad3 n/a (zink_dri.so + 0x639ad3)
#8 0x00007d7c04a4aa6e n/a (zink_dri.so + 0x4aa6e)
#9 0x00007d7c04a4adb5 n/a (zink_dri.so + 0x4adb5)
#10 0x00007d7c1bfd0e56 n/a (libEGL_mesa.so.0 + 0x17e56)
#11 0x00007d7c1bfd628f n/a (libEGL_mesa.so.0 + 0x1d28f)
#12 0x00007d7c1bfc3ec5 n/a (libEGL_mesa.so.0 + 0xaec5)
#13 0x00007d7c2003599e n/a (libgtk-4.so.1 + 0x43599e)
#14 0x00007d7c20008752 n/a (libgtk-4.so.1 + 0x408752)
#15 0x00007d7c200353d4 gdk_display_prepare_gl (libgtk-4.so.1 + 0x4353d4)
#16 0x00007d7c20002dd5 gdk_x11_display_open (libgtk-4.so.1 + 0x402dd5)
#17 0x00007d7c2002fa00 gdk_display_manager_open_display (libgtk-4.so.1 + 0x42fa00)
#18 0x00007d7c1fd8d58f gtk_init_check (libgtk-4.so.1 + 0x18d58f)
#19 0x00007d7c1fd8d7ee gtk_init (libgtk-4.so.1 + 0x18d7ee)
#20 0x0000596ac3ce0070 n/a (mutter-x11-frames + 0x3070)
#21 0x00007d7c1f5dbcd0 n/a (libc.so.6 + 0x25cd0)
#22 0x00007d7c1f5dbd8a __libc_start_main (libc.so.6 + 0x25d8a)
#23 0x0000596ac3ce0175 n/a (mutter-x11-frames + 0x3175)You're for some reason using libEGL_mesa and that derails wayland and X11 alike.
printenv | curl -F 'file=@-' 0x0.stOffline
You're facing a very different situation from the OP
You're for some reason using libEGL_mesa and that derails wayland and X11 alike.
printenv | curl -F 'file=@-' 0x0.st
That's weird, do you need this one?
https://0x0.st/H54G.txt
This is everything I have installed:
pacman -Qqe
amd-ucode
base
chromium
code
dosfstools
efibootmgr
file-roller
gdm
git
github-cli
gnome-console
gnome-control-center
grub
linux
linux-firmware
linux-headers
loupe
mtools
nano
nautilus
networkmanager
nodejs
noto-fonts
nvidia-open
sudo
ufw
xdg-desktop-portal-gnome
xdg-user-dirs
xdg-utilsLast edited by micronetic (2024-02-23 13:54:17)
Offline
I just enabled GDM on Wayland, I didn't remember it was so aesthetic, perhaps they changed the UI. But happy to say I am no longer using SDDM.
However, I started experiencing new issues, simiar to @micronetic.
Here is new log https://0x0.st/H5Jq.txt
Here is list of envs if it helps https://0x0.st/H5JT.txt
lots of stack trace issues.
All I did was to
- add nvidia.NVreg_PreserveVideoMemoryAllocations=1 to /boot/loader/entries/arch.conf under options.
- install gdm via pacman
- systemctl enable gdm and disable sddm
- enable nvidia services:
sudo systemctl enable nvidia-suspend.service
sudo systemctl enable nvidia-hibernate.service
sudo systemctl enable nvidia-resume.serviceLast edited by dab9966 (2024-02-23 14:56:21)
Offline
You're running gnome on wayland, though
XDG_CURRENT_DESKTOP=GNOME
WAYLAND_DISPLAY=wayland-0
XDG_SESSION_TYPE=waylandThe coredumps are from "mutter-x11-frames", does gnome on wayland actually start for you right now?
For the mutter-x11-frames thing, https://forums.developer.nvidia.com/t/m … tis/283055
=> Try to downgrade mesa to the last 23.x version.
Offline
How do you mean does it start for you? Why wouldn't it? It is working fine
Librewolf only runs on Wayland too, so if it doesn't start that means Wayland isn't running. Good way of checking.
3 = Get session name from running loginctl
loginctl show-session 3 -p Type
Type=waylandAlso downgrading mesa from 1.24.0 to 1.23 worked Thank you. Log is way more readable without all these stack trace issues.
Offline
For me it is still not working and I can't downgrade to mesa 23 because I don't have it in the cache or so.
Offline
@micronetic I have this version I can upload online for you if you want but I don't know if its going to be compatible with your OS, let me know
file:///var/cache/pacman/pkg/mesa-1:23.2.1-2-x86_64.pkg.tar.zst
Offline
Online
@dab9966 for clarification, downgrading mesa did not only remove the stack traces but you also don't get a black screen anymore?
In case, please always remember to mark resolved threads by editing your initial posts subject - so others will know that there's no task left, but maybe a solution to find.
Thanks.
Offline
@dab9966 for clarification, downgrading mesa did not only remove the stack traces but you also don't get a black screen anymore?
In case, please always remember to mark resolved threads by editing your initial posts subject - so others will know that there's no task left, but maybe a solution to find.
Thanks.
Hmm yes, it was more of an intermittent issue. I will mark this as resolved for now, haven't experienced it again since posting
Offline
Just one more bump. The first post with messages about NULL pointer were pinpointed to an extension: https://github.com/aunetx/blur-my-shell. Open issue here: https://github.com/aunetx/blur-my-shell/issues/485. Disabling extension got rid of these messages. It might be what made my gnome-shell process memory utilisation around 6GB, but I cannot be sure yet, as I am testing extensions to see if there is memory leak in any of them.
Feb 22 13:34:48 arch gnome-shell[752]: invalid (NULL) pointer instance
Feb 22 13:34:48 arch gnome-shell[752]: g_signal_handler_disconnect: assertion 'G_TYPE_CHECK_INSTANCE (instance)' failedthanks
Offline
Okay, since the latest Linux (6.7.6-arch1-1) and Nvidia (545.29.06) update it is working without problems anymore, can log in to the Wayland session like always.
Last edited by micronetic (2024-02-24 12:29:25)
Offline
FWIW, today I also ran into some weird issues involving Wayland, Brave, Chrome, and VSCode where basically complete windows would be invisible. Switching to X11, things were working as before (like yesterday). So I went ahead and installed nvidia, nvidia-utils, and lib32-nvidia-utils from extra-testing and multilib-testing. Turns out nVidia - despite their astronomic gain in market cap - have broken something in the 545 driver. Running on the 550 version (from testing) everything's back to normal.
Offline
FWIW, today I also ran into some weird issues involving Wayland, Brave, Chrome, and VSCode where basically complete windows would be invisible. Switching to X11, things were working as before (like yesterday). So I went ahead and installed nvidia, nvidia-utils, and lib32-nvidia-utils from extra-testing and multilib-testing. Turns out nVidia - despite their astronomic gain in market cap - have broken something in the 545 driver. Running on the 550 version (from testing) everything's back to normal.
For Code - OSS I created the file "code-flags.conf" inside the .configs folder with this inside the file and it works good for me:
--enable-features=UseOzonePlatform
--ozone-platform=waylandPlus Code - OSS Settings:
"window.titleBarStyle": "custom",
"editor.smoothScrolling": trueSame goes for Chromium, chrome://flags/ search for Wayland and set it to Wayland:
Preferred Ozone platform - WaylandI am on: Linux (6.7.6-arch1-1) and Nvidia (545.29.06)
Last edited by micronetic (2024-02-24 12:36:18)
Offline
Sure, when I use all those settings, things are also working on the older driver, but with the most recent version, things are working without configuring those settings explicitly, so something is broken in 545.
Last edited by Smoerrebroed (2024-02-24 22:12:22)
Offline