Hello!
I am changing the video card of a desktop PC from an Nvidia to an AMD. Fedora 39 is installed on the PC.
Before making the change I looked for information here on the forum and followed the directions in this post: Fedora 38: migrating from NVidia to AMD - #2 by barryascott
For me, however, it was almost a disaster!
Basically after the video card change the login screen and the Gnome shell crash at startup. I can only use the terminal tty.
I have already uninstalled the Nvidia drivers and deleted the nvidia-setting configuration file.
I also tried restarting gdm but it always crashes.
Can anyone help me solve this problem?
Please post the output of inxi -Fzxx so we have system details.
Are you using wayland or x11?
Look in the system journal and the user journal for errors at the time things go wrong. What so you find?
I haven’t installed the inxi command, nor can I install it now since I can’t access Gnome to connect to wi-fi networks.
With the video card and Nvidia drivers I was using X11.
I tried giving simple commands like journalctl and dmesg, but I wouldn’t know how to search better. Can you suggest more specific commands for me to use?
Also during system startup, pressing ESC, the outputs on the screen are all positive (OK in green).
To see the errors while booting from today:
journalctl --no-tail --no-hostname -S "today" -p err
If you remove -p err you could substitute with nvidia/amd to see if there are still some driver errors.
This is the output of the command you suggested:
set 20 18:50:18 kernel: ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.SAT0.SPT4_GTF.DSSP], AE_NOT_FOUND (20230628/psargs-330)
set 20 18:50:18 kernel: ACPI Error: Aborting method \_SB.PCI0.SAT0.SPT4_GTF due to previous error (AE_NOT_FOUND) (20230628/psparse-529)
set 20 18:50:18 kernel: ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.SAT0.SPT0_GTF.DSSP], AE_NOT_FOUND (20230628/psargs-330)
set 20 18:50:18 kernel: ACPI Error: Aborting method \_SB.PCI0.SAT0.SPT0_GTF due to previous error (AE_NOT_FOUND) (20230628/psparse-529)
set 20 18:50:18 kernel: ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.SAT0.SPT4_GTF.DSSP], AE_NOT_FOUND (20230628/psargs-330)
set 20 18:50:18 kernel: ACPI Error: Aborting method \_SB.PCI0.SAT0.SPT4_GTF due to previous error (AE_NOT_FOUND) (20230628/psparse-529)
set 20 18:50:18 kernel: ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PCI0.SAT0.SPT0_GTF.DSSP], AE_NOT_FOUND (20230628/psargs-330)
set 20 18:50:18 kernel: ACPI Error: Aborting method \_SB.PCI0.SAT0.SPT0_GTF due to previous error (AE_NOT_FOUND) (20230628/psparse-529)
set 20 16:50:24 smartd[913]: Device: /dev/sda [SAT], 14 Currently unreadable (pending) sectors
set 20 16:50:24 smartd[913]: Device: /dev/sda [SAT], 17338 Offline uncorrectable sectors
set 20 16:50:24 smartd[913]: Device: /dev/sdb [SAT], 1 Offline uncorrectable sectors
set 20 16:50:31 gnome-session-binary[4483]: Unrecoverable failure in required component org.gnome.Shell.desktop
Substituting the parameters with nvidia or amd to the command you suggested I get this output:
Unknown log level nvidia/amd
P.S: regarding SSD errors, I am aware of them, but so far they have never caused such problems and besides these are secondary disks on which there is only data. The Fedora system is installed on another healthy SSD.
I found that the following path still has the log files and packages related to the Nvidia drivers:
/var/cache/akmods/nvidia-470xx
Would deleting them solve it?
I assume you do have some other device with access to this forum and other internet sites.
You will find that diagnosing and fixing issues in linux often require use of command-line tools. Are you able to boot a Live USB installer? If that works you can install inxi (it goes in a RAM disk, so will no survive a reboot). If the Live USB installer fails to provide a GUI you can enable WiFi with command-line tools, in particular iw.
Please read man journalctl to understand what the options mean. I know from personal experience that it is all too easy when writing posts to omit or mistype essential details, so you should always make sure you understand what suggested commands are meant to do before running them. This has become more import now that click-bait sites are using AI-generated content to provide “quick and easy” answers to frequently asked questions.
I’m guessing you are meant to use the -g (grep) option with either nvidia or amd.
From this evidence your /dev/sda has failed and needs replaceing.
With the output of smartctl -A /dev/sda I could comment further.
This has been noted, but should not be causing Fedora crashes:
I would be surprised if fedora works with damaged files in the file system.
if you have the newer version of fpaste installed can you do:
fpaste --sysinfo-short --printonly
if not:
cat /proc/cmdline
lspci | grep -i -E 'vga' | cut -b1-7 | xargs -i lspci -vnnks {}
Yes, I have another computer with which I access this forum so I can solve the problem.
Yes, I can boot and use a Live version from USB. I recently downloaded Fedora 40 KDE to test it. If this version is ok, since I have Gnome on my system instead, I can boot this Live version, install inxi and provide the required output.
Through iw, after connecting to a network, can I try reinstalling gdm or the Gnome shell as a possible solution to the problem?
Thank you for the suggestion. I will definitely read and try to better understand the commands that will be recommended. Usually when I find on random online guides about the commands I delve deeper, but when I read them on this forum I trust them a little more because I think there are people more knowledgeable than me.
I try using the -g (grep) option instead of -p.
Using this command:
journalctl --no-tail --no-hostname -S “today” -g amd
I get an output of 207 lines in which there do not seem to be any errors in red. I will try somehow to export it and share it.
Instead, using this command:
journalctl --no-tail --no-hostname -S “today” -g nvidia
I get this output:
--No entries--
Those errors on sda appeared after restoring a partition via Rescuezilla, otherwise they were not there before. I gather that the Rescuezilla restore was not done well.
I assume that by deleting the restored partition they should go away. Also because the disk is healthy and from the SMART status it shows 97% life.
Thanks for the helpfulness! As soon as I resolve the Gnome crash I will share with you the SMART status of the disk.
Can this output, on the other hand, indicate something useful to be able to solve the main problem?
Sep 20 16:50:31 gnome-session-binary[4483]: Unrecoverable failure in required component org.gnome.Shell.desktop
The suggested fpaste command gives me a message saying that the syntax is incorrect.
Instead, the other two commands have the following outputs:
cat /proc/cmdline
BOOT_IMAGE=/@/boot/vmlinuz-6.8.11-200.fc39.x86_64 root=UUID=fa7f4375-9aca-4c2c-a122-8a0abaa53c18 ro rootflags=subvol=0
lspci | grep -i -E 'vga' | cut -b1-7 | xargs -i lspci -vnnks {}
VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Navi 14 [Radeon RX 5500/5500M / Pro 5500M] [1002:7340] (rev c5) (prog-if 00 [VGA controller])
Subsystem: Sapphire Technology Limited Device [1da2:e421]
Flags: bus master. fast devsel, latency 0, IRQ 49
Memory at e0000000 (64-bit, prefetchable) [size=256M]
Memory at f0000000 (64-bit, prefetchable) [size=2M]
I/O ports at e000 [size=256]
Memory at f7b00000 (32-bit, non-prefetchable) [size=512K]
Expansion ROM at 000c0000 [disable] [size=128K]
Capabilities: <access denied>
Kernel driver in use: amdgpu
Kernel modules: amdgpu
Did you already try moving the monitors.xml file?
mv ~/.config/monitors.xml ~/.config/monitors.old
If you had a specific setup for gdm you may need to look at that as well, I believe that is in /var/lib/gdm/.config
The monitor.xml file is nonexistent.
I don’t think I have a specific gdm configuration.
I tried to access the path you suggested but it is impossible even using sudo.
AFAIK, there should be one under your user config directory. ~/.config
ls -l ~/.config/*.xml
you should be able to see if there anything under the gdm config with :
sudo ls -l /var/lib/gdm/.config/
that file may not be there.
what does the below indicate:
systemctl status gdm.service
journalctl -b -u gdm
Thanks
1 Like
I confirm that there is no .xml file in the home .config folder.
In the /var/lib/gdm/.config/ path, however, these folders are present:
total 0
drwx------. 1 gdm gdm 8 21 giu 19.09 dconf
drwx------. 1 gdm gdm 26 13 feb 2023 gnome-session
drwx------. 1 gdm gdm 6 13 feb 2023 ibus
drwx------. 1 gdm gdm 32 28 mar 17.09 pulse
Ooops misread sorry about that, what about the output of the gdm.service journal and systememctl. Thanks.