gdm unable too login. It is bouncing back to the login screen

odror · 2024-12-08 19:16:14

Hi

I have tried everything that I can think. I am not able to login to gnome-session.

A different user can login.
I removed the entire .config directory I am still not able to login.
I rollback my system to 2 days ago. I am still not able to login.
I reinstalled the nvidia driver it did not help.

Any ideas how to fix

odror · 2024-12-08 19:31:28

correction
Now any user cannot login.
I get this error in journalctl

gsd-media-keys[13191]: Unable to get default sink

is this related

odror · 2024-12-08 20:35:00

I have an update.

I rolled back mesa and mutter and the issue is resolved. I am guessing that there is some incompatibility with the latest nvidia and mutter or mesa.

Any ideas?

seth · 2024-12-08 20:56:21

Hybrid system? AMD? https://bbs.archlinux.org/viewtopic.php?id=301581 ?
Roll back from which to what version?

There's probably some https://wiki.archlinux.org/title/Core_d … _core_dump or session error in the journal?

odror · 2024-12-09 00:00:46

I have a mixed intel+Nvidia GPU not amd.
i7-13900k and nvidia RTX 4070Ti

After waking up from suspend The problem came back. I rolled back nvidia-dkms to 565.57.01 and now it is OK. I am not sure that the latest mesa and mutter are the issue. (currently I have mesa 1:24.2.7-1 and mutter 47.2-1) I think it is the nvidia driver in my particular situation
with Intel and Nvidia GPU running concurrently is failing with the new nvidia driver.

I would switch to wayland, but it is too limiting in my setup with 5 4k monitors. I think mesa can take only 4 monitors. ( I have 5 monitors because 3 are for a VM and 2 for the host)

Also the issue did not exist with gnome classic (X11)

seth · 2024-12-09 07:25:08

seth wrote:

There's probably some https://wiki.archlinux.org/title/Core_d … _core_dump or session error in the journal?

odror · 2024-12-10 01:30:52

The issue happen when either I upgrade to mesa mesa-1:24.3.1-2 or nvidia 565.77. It is an X11 issue only. Wayland is ok.

I get the following error in journalctl.

Dec 09 17:19:59 i9-13900k-arch /usr/lib/gdm-x-session[7266]: /etc/gdm/Xsession: Beginning session setup...
Dec 09 17:19:59 i9-13900k-arch /usr/lib/gdm-x-session[7285]: localuser:dror being added to access control list
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7266]: /etc/gdm/Xsession: Setup done, will execute: /usr/bin/gnome-session
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE)
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) Backtrace:
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) unw_get_proc_name failed: no unwind info found [-10]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) 0: /usr/lib/Xorg (?+0x0) [0x5e725962789c]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) unw_get_proc_name failed: no unwind info found [-10]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) 1: /usr/lib/libc.so.6 (?+0x0) [0x77565a7f01d0]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) unw_get_proc_name failed: no unwind info found [-10]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) 2: /usr/lib/nvidia/xorg/libglxserver_nvidia.so (?+0x0) [0x775646d3a7a5]
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE)
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) Segmentation fault at address 0x0
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE)
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: Fatal server error:
Dec 09 17:20:00 i9-13900k-arch /usr/lib/gdm-x-session[7254]: (EE) Caught signal 11 (Segmentation fault). Server aborting

Last edited by odror (2024-12-10 01:32:21)

seth · 2024-12-10 08:37:07

It crashes in the nvidia GLX X11 module, so the downgrade/wayland situation makes sense.
Do you have the xorg log for that failure?
(If not r/n, you can also start by posting the Xorg log of the working session - just for a general overview of the supposed setup)
gdm should™ store the entire xorg log in the journal, though: https://wiki.archlinux.org/title/Xorg#General

odror · 2024-12-10 21:16:23

I get the same error in the X11 log file.

seth · 2024-12-10 22:30:59

https://bbs.archlinux.org/viewtopic.php?id=57855
The point is the entire xorg log to see what constellation *leads* to this error…

odror · 2024-12-10 22:35:14

It is uploaded to

http://0x0.st/Xhvp.txt

seth · 2024-12-10 23:02:35

[ 12346.309] (==) Log file: "/home/dror/.local/share/xorg/Xorg.1.log", Time: Mon May 27 14:51:38 2024

That log is severely dated.

odror · 2024-12-10 23:06:13

I am sorry I send the wrong one.. It should be Xorg.0.log.old not Xorg.1.log.old

http://0x0.st/XhvU.txt

Last edited by odror (2024-12-10 23:07:01)

seth · 2024-12-10 23:24:53

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

Post that file, there's some weird

[   704.983] (WW) modeset(G0): Option "AllowEmptyInitialConfiguration" is not used
[   704.983] (WW) modeset(G0): Option "PrimaryGPU" is not used
[   704.983] (WW) modeset(G0): Option "Stereo" is not used
[   704.983] (WW) modeset(G0): Option "nvidiaXineramaInfoOrder" is not used
[   704.983] (WW) modeset(G0): Option "metamodes" is not used
[   704.983] (WW) modeset(G0): Option "SLI" is not used
[   704.983] (WW) modeset(G0): Option "MultiGPU" is not used
[   704.983] (WW) modeset(G0): Option "BaseMosaic" is not used

(I guess it's to facilitate the reverse prime setup)
Then move that file away and see whether the issue remains (you'll probably have to "prime-run glxinfo" or so to invoke the nvidia GLX)

The older log had

[ 12347.050] (WW) NVIDIA(0): Failed to request fliplock.

Probably prime-sync related, https://wiki.archlinux.org/title/PRIME# … ronization

odror · 2024-12-11 01:56:35

my xorg.conf is below. If I remove it. mesa/intel will render my monitors. Since I have 5 4k monitors, mesa fails with an error ( i get the white error screen), since it can only handle 4 monitors. I have had this issue before. It was solved with this xorg.conf, which is now not working after the mesa upgrade.

Section "OutputClass"
    Identifier "intel"
    MatchDriver "i915"
    Driver "modesetting"
EndSection

Section "OutputClass"
    Identifier "nvidia"
    Driver "nvidia"
    Option "AllowEmptyInitialConfiguration"
    Option "PrimaryGPU" "yes"
EndSection

Section "Screen"
    Identifier     "Screen0"
    DefaultDepth    24
    Option         "Stereo" "0"
    Option         "nvidiaXineramaInfoOrder" "DFP-1"
    Option         "metamodes" "HDMI-0: nvidia-auto-select +0+0 {ForceCompositionPipeline=On, ForceFullCompositionPipeli
ne=On}, DP-0: nvidia-auto-select +3840+0 {ForceCompositionPipeline=On, ForceFullCompositionPipeline=On, AllowGSYNCCompat
ible=On}, DP-4: nvidia-auto-select +7680+0 {ForceCompositionPipeline=On, ForceFullCompositionPipeline=On}"
    Option         "SLI" "Off"
    Option         "MultiGPU" "Off"
    Option         "BaseMosaic" "off"
    SubSection     "Display"
        Depth       24
    EndSubSection
EndSection

seth · 2024-12-11 07:54:00

'key - I don't really think this causes it (I had hoped for modulepath manipulations or explicit module loading) but comment the screen section (you'll lose the preconfigured layout but that would be fixable by other means)

Assuming that's not it, let's have a look at the GDM situation.
It probably defaults to wayland (or would crah as well), does X11 likewise crash if you https://wiki.archlinux.org/title/GDM#Use_Xorg_backend resp. what happens if you don't use GDM at all but start the session w/ xinit/startx? (See the last link below on what an xinitrc should look like to not end up w/ a broken session)

odror · 2024-12-12 01:58:18

Removing the screen section did resolve the issue with the newest mesa driver, but when updating to the latest nvidia driver 564.77-1 the issue came back. So I had to downgrade the nvidia driver. my log file after the upgrade of the nvidia driver is:

http://0x0.st/XhCs.txt

The is a separate nvidia driver issue. I am using kernel 6.10.10-arch1-1 #1 SMP PREEMPT_DYNAMIC. If it makes any difference.

seth · 2024-12-12 11:01:03

https://github.com/sddm/sddm/issues/200 … 2481204286 ?
You do run into the permission error and despite (?) GDM the xorg log shows a rootless server…

odror · 2024-12-13 02:53:59

I tried adding

Section "Screen"
    Option         "SidebandSocketPath" "/tmp"
EndSection

As suggested. It did not work.

I am not sure what is the issue and how I can fix it.

seth · 2024-12-13 09:26:24

seth wrote:

does X11 likewise crash if you
* https://wiki.archlinux.org/title/GDM#Use_Xorg_backend resp. what happens if you
* don't use GDM at all but start the session w/ xinit/startx? (See the last link below on what an xinitrc should look like to not end up w/ a broken session)

Also please post the xorg log of that to illustrate your recent changes.

odror · 2025-01-02 00:30:24

I was off for 2 weeks. The issue is not resolved. in fact the new systemd 257-1 causes the same issue independently.

I disabled gdm and enabled sddm. The issue is resolved. It seems like and issue of gdm on xorg only for the newest packages of nvidia or systemd.

seth · 2025-01-02 10:32:36

Originally

seth wrote:

It crashes in the nvidia GLX X11 module, so the downgrade/wayland situation makes sense.

so any relation to systemd and GDM doesn't make a whole lot of sense.
Please post an up-to-date system journal and X11 log (covering the GDM failures) to illustrate the status quo.

odror · 2025-01-03 05:09:59

The log files is available bellow. The issue is with GDM with the latest version of systemd or the latest version of Nvidia. No issues with SDDM.

http://0x0.st/8z52.txt

seth · 2025-01-03 09:58:27

"NVreg_PreserveVideoMemoryAllocations=1" is nonsense, should be "nvidia.NVreg_PreserveVideoMemoryAllocations=1" but is now default anyway

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

Post that file - and your system journal.

[   511.801] (II) xfree86: Adding drm device (/dev/dri/card1)
[   511.801] (II) Platform probe for /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/drm/card1
[   511.801] (II) systemd-logind: got fd for /dev/dri/card1 226:1 fd 14 paused 0
[   511.801] (II) xfree86: Adding drm device (/dev/dri/card0)
[   511.801] (II) Platform probe for /sys/devices/pci0000:00/0000:00:02.0/drm/card0
[   511.801] (II) systemd-logind: got fd for /dev/dri/card0 226:0 fd 15 paused 0
[   511.802] (**) OutputClass "nvidia" ModulePath extended to "/usr/lib/nvidia/xorg,/usr/lib/xorg/modules,/usr/lib/xorg/modules"
[   511.802] (**) OutputClass "nvidia" setting /dev/dri/card1 as PrimaryGPU
[   511.814] (--) PCI: (0@0:2:0) 8086:a780:1043:8882 rev 4, Mem @ 0x656d000000/16777216, 0x4000000000/268435456, I/O @ 0x00008000/64, BIOS @ 0x????????/131072
[   511.814] (--) PCI:*(1@0:0:0) 10de:2782:1043:88dd rev 161, Mem @ 0x78000000/16777216, 0x6000000000/17179869184, 0x6400000000/33554432, I/O @ 0x00007000/128, BIOS @ 0x????????/524288
[   511.814] (--) PCI: (6@0:0:0) 10de:13f1:103c:1153 rev 161, Mem @ 0x76000000/16777216, 0x6550000000/268435456, 0x6560000000/33554432, I/O @ 0x00005000/128, BIOS @ 0x????????/524288

You've 3 GPUs (2 nvidia, one intel) and 2 cards, the intel GPU is card0 but then

[   511.834] (II) Applying OutputClass "nvidia" to /dev/dri/card0
[   511.834] 	loading driver: nvidia
…
[   511.837] (II) modeset(G0): using drv /dev/dri/card0
…
[   514.301] (EE) 0: /usr/lib/Xorg (?+0x0) [0x5b39a3bbf89c]
[   514.301] (EE) unw_get_proc_name failed: no unwind info found [-10]
[   514.301] (EE) 1: /usr/lib/libc.so.6 (?+0x0) [0x79f7eedaf1d0]
[   514.301] (EE) unw_get_proc_name failed: no unwind info found [-10]
[   514.301] (EE) 2: /usr/lib/nvidia/xorg/libglxserver_nvidia.so (?+0x0) [0x79f7e838e7a5]
[   514.301] (EE) 
[   514.302] (EE) Segmentation fault at address 0x0

you end up running the nvidia GLX module against the modesetting driver on the intel IGP.
That's not related to GDM or systemd.

odror · 2025-01-03 21:26:34

I am not an expert, but I thought that nvidia can render monitors attached to the intel card (Nvidia Prime). I am getting the following on all the monitors including the ones that are attached to the integrated intel card.

$ glxinfo | grep "OpenGL renderer"
OpenGL renderer string: NVIDIA GeForce RTX 4070 Ti/PCIe/SSE2

Also this does not explain Why I do not have an issue once I was able to login succefuly using SDDM, and it does not explain why it used to work before the upgrading systemd and nvidia.

Arch Linux

#1 2024-12-08 19:16:14

gdm unable too login. It is bouncing back to the login screen

#2 2024-12-08 19:31:28

Re: gdm unable too login. It is bouncing back to the login screen

#3 2024-12-08 20:35:00

Re: gdm unable too login. It is bouncing back to the login screen

#4 2024-12-08 20:56:21

Re: gdm unable too login. It is bouncing back to the login screen

#5 2024-12-09 00:00:46

Re: gdm unable too login. It is bouncing back to the login screen

#6 2024-12-09 07:25:08

Re: gdm unable too login. It is bouncing back to the login screen

#7 2024-12-10 01:30:52

Re: gdm unable too login. It is bouncing back to the login screen

#8 2024-12-10 08:37:07

Re: gdm unable too login. It is bouncing back to the login screen

#9 2024-12-10 21:16:23

Re: gdm unable too login. It is bouncing back to the login screen

#10 2024-12-10 22:30:59

Re: gdm unable too login. It is bouncing back to the login screen

#11 2024-12-10 22:35:14

Re: gdm unable too login. It is bouncing back to the login screen

#12 2024-12-10 23:02:35

Re: gdm unable too login. It is bouncing back to the login screen

#13 2024-12-10 23:06:13

Re: gdm unable too login. It is bouncing back to the login screen

#14 2024-12-10 23:24:53

Re: gdm unable too login. It is bouncing back to the login screen

#15 2024-12-11 01:56:35

Re: gdm unable too login. It is bouncing back to the login screen

#16 2024-12-11 07:54:00

Re: gdm unable too login. It is bouncing back to the login screen

#17 2024-12-12 01:58:18

Re: gdm unable too login. It is bouncing back to the login screen

#18 2024-12-12 11:01:03

Re: gdm unable too login. It is bouncing back to the login screen

#19 2024-12-13 02:53:59

Re: gdm unable too login. It is bouncing back to the login screen

#20 2024-12-13 09:26:24

Re: gdm unable too login. It is bouncing back to the login screen

#21 2025-01-02 00:30:24

Re: gdm unable too login. It is bouncing back to the login screen

#22 2025-01-02 10:32:36

Re: gdm unable too login. It is bouncing back to the login screen

#23 2025-01-03 05:09:59

Re: gdm unable too login. It is bouncing back to the login screen

#24 2025-01-03 09:58:27

Re: gdm unable too login. It is bouncing back to the login screen

#25 2025-01-03 21:26:34

Re: gdm unable too login. It is bouncing back to the login screen

Board footer