Try the beta driver.
https://ftp.icm.edu.pl/pub/Linux/dist/rpmfusion/nonfree/fedora/updates/testing/41/x86_64/repoview/
Installing the beta drivers, along side changing the GSK_RENDERER setting sadly had no effect.
I have now reverted back to stable drivers.
What worked for me was to reset all Gnome settings:
mv $HOME/.config/dconf/user{,.bak}
I noticed it was working for an account I rarely used so I guess I had some setting in my regular account that was causing gnome-shell to crash in Wayland.
1 Like
I did find this in a journalctl:
Nov 01 09:30:07 Framework audit[24150]: ANOM_ABEND auid=1000 uid=1000 gid=1000 ses=16 subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 pid=24150 comm=âgnome-shellâ exe=â/usr/bin/gnome-shellâ sig=11 res=1
I wonder if you would see a similar thing?
This worked for me. Thanks for your help everyone!
I would like to note that after you ran the following command, you will need to delete the user file, but leave the user.bak file in the directory.
Out of the abundance of precaution, I also did the same with the rootâs user profile.
dconf reset -f /
This will reset all the settings in dconf/gsettings database, which is utilized by the Gnome system & GTK applications.
2 Likes
That would be an easier way of doing it.
Upgraded to 41 in AMD RAEDON. Wayland option is not working, coming back to login (kind of loop). Xorg working.
The above fix worked: dconf reset -f / after running from Xorg. Then login back with Wayland
Log:
Nov 03 13:49:39 fedora gnome-shell[2808]: Running GNOME Shell (using mutter 47.1) as a Wayland display server
Nov 03 13:49:39 fedora gnome-shell[2808]: Enabling experimental feature 'scale-monitor-framebuffer'
Nov 03 13:49:39 fedora gnome-shell[2808]: Thread 'KMS thread' will be using real time scheduling
Nov 03 13:49:39 fedora gnome-shell[2808]: Device '/dev/dri/card1' prefers shadow buffer
Nov 03 13:49:39 fedora gnome-shell[2808]: Added device '/dev/dri/card1' (amdgpu) using atomic mode setting.
Nov 03 13:49:39 fedora gnome-shell[2808]: Created gbm renderer for '/dev/dri/card1'
Nov 03 13:49:39 fedora gnome-shell[2808]: Boot VGA GPU /dev/dri/card1 selected as primary
Nov 03 13:49:39 fedora gnome-shell[2808]: Obtained a high priority EGL context
Nov 03 13:49:39 fedora wireplumber[2766]: [0:00:42.223310161] [2766] INFO Camera camera_manager.cpp:325 libcamera v0.3.2
Nov 03 13:49:39 fedora systemd[2635]: Starting xdg-permission-store.service - sandboxed app permission store...
Nov 03 13:49:39 fedora systemd[2635]: Started xdg-permission-store.service - sandboxed app permission store.
Nov 03 13:49:40 fedora audit[2808]: ANOM_ABEND auid=1000 uid=1000 gid=1000 ses=3 subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 pid=2808 comm="gnome-shell" exe="/usr/bin/gnome-shell" sig=11 res=1
Nov 03 13:49:40 fedora kernel: show_signal_msg: 70 callbacks suppressed
Nov 03 13:49:40 fedora kernel: gnome-shell[2847]: segfault at 20 ip 00007f09f856e94a sp 00007f09bdffe790 error 4 in libmutter-15.so.0.0.0[7f09f8400000+194000] likely on CPU 15 (core 7, socket 0)
Nov 03 13:49:40 fedora kernel: Code: ff ff 48 89 c7 48 89 c3 e8 93 18 ea ff 4c 89 f7 e8 9b eb e9 ff 48 8b 0d bc f5 10 00 4c 89 ee 48 89 df 48 89 c2 e8 96 79 ea ff <49> 8b 74 24 20 48 89 df e8 59 3f ea ff 48 89 df 5b 41 5c 41 5d 41
Nov 03 13:49:40 fedora systemd-coredump[2863]: Process 2808 (gnome-shell) of user 1000 terminated abnormally with signal 11/SEGV, processing...
Nov 03 13:49:40 fedora systemd[1]: Created slice system-systemd\x2dcoredump.slice - Slice /system/systemd-coredump.
Nov 03 13:49:40 fedora audit: BPF prog-id=85 op=LOAD
Nov 03 13:49:40 fedora audit: BPF prog-id=86 op=LOAD
Nov 03 13:49:40 fedora audit: BPF prog-id=87 op=LOAD
Nov 03 13:49:40 fedora systemd[1]: Started systemd-coredump@0-2863-0.service - Process Core Dump (PID 2863/UID 0).
Nov 03 13:49:40 fedora audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-coredump@0-2863-0 comm="systemd" exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
Nov 03 13:49:40 fedora abrt-dump-journal-core[1282]: Failed to obtain all required information from journald
Nov 03 13:49:41 fedora systemd-coredump[2864]: [đĄ] Process 2808 (gnome-shell) of user 1000 dumped core.
It would be interesting if someone managing to solve the issue with the workaround of resetting the userâs dconf database, and who has kept the old db as well, could run a diff on the 2 files, to try to figure out what change might be causing this.
A new bug report against GNOME Shell, or maybe a comment on this existing one, which seems similar, might be of help in fixing this issue upstream.
hi! what worked for me was rebuilding the nvidia driver using akmod:
sudo akmods --rebuild
then a new start and now gnome/wayland works again 
1 Like
These havenât worked for me. I tried: dconf reset -f / as well as: mv $HOME/.config/dconf/user{,.bak} and deleting the user file while leaving the user.bak in the directory. I did not do this with the rootâs user profile. Any ideas to check next?
on my system i think the problem was that the nvidia driver hadnât been properly installed during upgrade. to fix this, i ran sudo akmods --rebuild and then rebooted
No nvidia on my computer. I have seen that is helping those with nvidia drivers though.
It works for for a while, but after logging out and logging back in. It happens again.
Basically, I have to repeat the dconf reset -f / steps multiple times.
I wonder if your issue comes back after you set a different display scaling than the default one, and fractional scaling (i.e. not multiple of 100%, e.g. 175% instead of 200%) in particular?
If so, you could try disabling fractional display scaling, by first setting up a non-fractional one (e.g. 200%), and then running the following command:
gsettings set org.gnome.mutter experimental-features "[]"
Currently, I have the scaling set to the default (100%).
The only scaling I did was the fonts, since I found the fractional scaling to have some glitches in the UI.
1 Like
According to this bug report, it seems to be connected to an accessibility issue.
I managed to reproduce it myself, when setting the value of the key enable in schema org.gnome.desktop.a11y.keyboard to true.
By settings it to false (from a tty console), issue gets fixed.
gsettings set org.gnome.desktop.a11y.keyboard enable false
Might be one of the culprits.
In x11, I went into settings â accessibility â pointing and clicking â then turned off Mouse Keys. This allows me to then log into Wayland. When I turn Mouse Keys back on, it goes back to the login loop and unable to get in.
2 Likes
Can confirm. This also happened to me, too.
I also tested it on my Framework 13 laptop, and that also caused it to loop back to the login screen. Disabling it again fixed it.