Bug 475490 - [NVIDIA] Immediate segfault in kwin_wayland upon initiation of Wayland session by SDDM; followed by prompt return to SDDM greeter
Summary: [NVIDIA] Immediate segfault in kwin_wayland upon initiation of Wayland sessio...
Status: RESOLVED FIXED
Alias: None
Product: kwin
Classification: Plasma
Component: wayland-generic (show other bugs)
Version: 5.27.8
Platform: Arch Linux Linux
: NOR crash
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2023-10-12 00:23 UTC by Gary S. Martin
Modified: 2023-10-12 21:37 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments
Complete boot journal for Wayland attempt and X11 login (367.53 KB, text/x-log)
2023-10-12 00:23 UTC, Gary S. Martin
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Gary S. Martin 2023-10-12 00:23:24 UTC
Created attachment 162237 [details]
Complete boot journal for Wayland attempt and X11 login

SUMMARY
***
NOTE: If you are reporting a crash, please try to attach a backtrace with debug symbols.
See https://community.kde.org/Guidelines_and_HOWTOs/Debugging/How_to_create_useful_crash_reports
***


STEPS TO REPRODUCE
1. Install Archlinux per https://wiki.archlinux.org/title/Installation_guide
2. Install SDDM per https://wiki.archlinux.org/title/SDDM
3. Install Plasma per https://wiki.archlinux.org/title/KDE#Plasma
4. Install Wayland per https://wiki.archlinux.org/title/Wayland
5. Install additional prerequisites per https://community.kde.org/Plasma/Wayland/Nvidia
6. Boot system to SDDM greeter
7. Select "Plasma (Wayland)," enter credentials, then select "login"

OBSERVED RESULT
The greeter disappears, and the screen goes blank except for an underscore in the first column of the first row. After a short period, the underscore is replaced by a cursor in the middle of the screen (which does respond to mouse movement), followed very shortly by the reappearance of the greeter.

Oct 11 15:22:59 TehachapiMtn dbus-daemon[780]: [session uid=1000 pid=780] Activa
ting service name='org.kde.KSplash' requested by ':1.7' (uid=1000 pid=1084 comm=
"/usr/bin/startplasma-wayland")
Oct 11 15:22:59 TehachapiMtn systemd[669]: Created slice User Background Tasks S
lice.
Oct 11 15:22:59 TehachapiMtn systemd[669]: Reached target Session services which
 should run early before the graphical session is brought up.
Oct 11 15:22:59 TehachapiMtn systemd[669]: Starting KDE Window Manager...
Oct 11 15:22:59 TehachapiMtn systemd[669]: Starting Baloo File Indexer Daemon...
Oct 11 15:22:59 TehachapiMtn systemd[669]: Started KDE Window Manager.
Oct 11 15:22:59 TehachapiMtn systemd[669]: Starting KDE Config Module Initialization...
Oct 11 15:22:59 TehachapiMtn systemd[669]: Starting Splash screen shown during boot...
Oct 11 15:22:59 TehachapiMtn systemd[669]: Started Baloo File Indexer Daemon.
Oct 11 15:22:59 TehachapiMtn dbus-daemon[780]: [session uid=1000 pid=780] Activating via systemd: service name='org.freedesktop.portal.Desktop' unit='xdg-desktop-portal.service' requested by ':1.11' (uid=1000 pid=1133 comm="/usr/bin/kcminit_startup")
Oct 11 15:22:59 TehachapiMtn systemd[669]: Starting Portal service...
Oct 11 15:23:00 TehachapiMtn kwin_wayland[1128]: No backend specified, automatically choosing drm
Oct 11 15:23:00 TehachapiMtn kernel: kwin_wayland[1128]: segfault at 0 ip 000064d066a1550b sp 0000716de9708fe0 error 4 in libnvidia-allocator.so.535.113.01[64d066a00000+1f000] likely on CPU 3 (core 1, socket 0)
Oct 11 15:23:00 TehachapiMtn kernel: Code: 44 24 08 ff 95 e0 00 00 00 48 83 c4 10 48 89 df 5b 5d 41 5c e9 26 d5 fe ff 66 0f 1f 44 00 00 53 48 8b 87 38 01 00 00 48 89 fb <48> 8b 00 48 85 c0 74 02 ff d0 48 8b bb c8 00 00 00 48 85 ff 74 05
Oct 11 15:23:00 TehachapiMtn systemd[1]: Created slice Slice /system/systemd-coredump.
Oct 11 15:23:00 TehachapiMtn dbus-daemon[780]: [session uid=1000 pid=780] Activating via systemd: service name='org.freedesktop.portal.Documents' unit='xdg-document-portal.service' requested by ':1.14' (uid=1000 pid=1137 comm="/usr/lib/xdg-desktop-portal")
Oct 11 15:23:00 TehachapiMtn systemd[1]: Started Process Core Dump (PID 1145/UID 0).
Oct 11 15:23:00 TehachapiMtn systemd[669]: Starting flatpak document portal service...
Oct 11 15:23:00 TehachapiMtn dbus-daemon[780]: [session uid=1000 pid=780] Activating via systemd: service name='org.freedesktop.impl.portal.PermissionStore' unit='xdg-permission-store.service' requested by ':1.15' (uid=1000 pid=1150 comm="/usr/lib/xdg-document-portal")
Oct 11 15:23:00 TehachapiMtn systemd[669]: Starting sandboxed app permission store...
Oct 11 15:23:00 TehachapiMtn dbus-daemon[780]: [session uid=1000 pid=780] Successfully activated service 'org.freedesktop.impl.portal.PermissionStore'
Oct 11 15:23:00 TehachapiMtn systemd[669]: Started sandboxed app permission storEXPECTED RESULT


SOFTWARE/OS VERSIONS
Windows: 
macOS: 
Linux/KDE Plasma: Archlinux (updated today)/ 6.5.7-hardened1-1-hardened (64-bit)
(available in About System)
KDE Plasma Version:  5.27.8-4
KDE Frameworks Version: 5.110.0
Qt Version:  5.15.11
NVIDIA Proprietary Driver Version: 535.113.01

ADDITIONAL INFORMATION
This computer ran KDE Plasma/Wayland for about a year without significant issues before replacing the previous AMD GPU with the NVIDIA RTX4060 installed 6 weeks ago.
Comment 1 Zamundaaa 2023-10-12 11:18:03 UTC
Please attach the output of drm_info on your PC
Comment 2 Gary S. Martin 2023-10-12 11:44:26 UTC
(In reply to Zamundaaa from comment #1)
> Please attach the output of drm_info on your PC

http://0x0.st/H4_h.txt
Comment 3 Gary S. Martin 2023-10-12 11:48:05 UTC
I have also submitted this to linux-bugs@nvidia.com.
Comment 4 Zamundaaa 2023-10-12 11:52:34 UTC
You seem to have some setup issue, there should never be simpledrm + another drm driver at the same time.
As for the actual crash, do you have the environment variables set the arch wiki suggests?
> GBM_BACKEND=nvidia-drm
> __GLX_VENDOR_LIBRARY_NAME=nvidia
If so, I'd recommend you to delete them, they just break things.
Comment 5 Gary S. Martin 2023-10-12 13:28:16 UTC
(In reply to Zamundaaa from comment #4)
> You seem to have some setup issue, there should never be simpledrm + another
> drm driver at the same time.
> As for the actual crash, do you have the environment variables set the arch
> wiki suggests?
> > GBM_BACKEND=nvidia-drm
> > __GLX_VENDOR_LIBRARY_NAME=nvidia
> If so, I'd recommend you to delete them, they just break things.

#GBM_BACKEND=nvidia-drm
#__GLX_VENDOR_LIBRARY_NAME=nvidia
#LIBVA_DRIVER_NAME=nvidia

Yes, they were set as recommended in the Archlinux wiki. I commented out the three lines above in /etc/environment and rebooted and I am now able to start a Wayland session. I am seeing occasional incidences of flickering of this browser window and some occasional weird shifting of this text as I type, but Wayland starts and seems to be mostly stable now.

New output of drm_info: http://0x0.st/H4Lw.txt
Comment 6 Gary S. Martin 2023-10-12 13:30:35 UTC
(In reply to Gary S. Martin from comment #5)
> (In reply to Zamundaaa from comment #4)
> > You seem to have some setup issue, there should never be simpledrm + another
> > drm driver at the same time.
> > As for the actual crash, do you have the environment variables set the arch
> > wiki suggests?
> > > GBM_BACKEND=nvidia-drm
> > > __GLX_VENDOR_LIBRARY_NAME=nvidia
> > If so, I'd recommend you to delete them, they just break things.
> 
> #GBM_BACKEND=nvidia-drm
> #__GLX_VENDOR_LIBRARY_NAME=nvidia
> #LIBVA_DRIVER_NAME=nvidia
> 
> Yes, they were set as recommended in the Archlinux wiki. I commented out the
> three lines above in /etc/environment and rebooted and I am now able to
> start a Wayland session. I am seeing occasional incidences of flickering of
> this browser window and some occasional weird shifting of this text as I
> type, but Wayland starts and seems to be mostly stable now.
> 
> New output of drm_info: http://0x0.st/H4Lw.txt

FYI, the browser is Firefox. The flickering does not seem to involve the desktop image or the console window that is to the right of the Firefox window.
Comment 7 Gary S. Martin 2023-10-12 13:33:46 UTC
(In reply to Gary S. Martin from comment #6)
> (In reply to Gary S. Martin from comment #5)
> > (In reply to Zamundaaa from comment #4)
> > > You seem to have some setup issue, there should never be simpledrm + another
> > > drm driver at the same time.
> > > As for the actual crash, do you have the environment variables set the arch
> > > wiki suggests?
> > > > GBM_BACKEND=nvidia-drm
> > > > __GLX_VENDOR_LIBRARY_NAME=nvidia
> > > If so, I'd recommend you to delete them, they just break things.
> > 
> > #GBM_BACKEND=nvidia-drm
> > #__GLX_VENDOR_LIBRARY_NAME=nvidia
> > #LIBVA_DRIVER_NAME=nvidia
> > 
> > Yes, they were set as recommended in the Archlinux wiki. I commented out the
> > three lines above in /etc/environment and rebooted and I am now able to
> > start a Wayland session. I am seeing occasional incidences of flickering of
> > this browser window and some occasional weird shifting of this text as I
> > type, but Wayland starts and seems to be mostly stable now.
> > 
> > New output of drm_info: http://0x0.st/H4Lw.txt
> 
> FYI, the browser is Firefox. The flickering does not seem to involve the
> desktop image or the console window that is to the right of the Firefox
> window.

The whole Firefox window just disappeared for about ten seconds and then came back.
Comment 8 Gary S. Martin 2023-10-12 21:37:43 UTC
The original crash issue was resolved by commenting out the setting of the three NVIDI-Wayland-related environment variables (GBM_BACKEND, GLX_VENDOR_LIBRARY_NAME, andLIBVA_DRIVER_NAME. My flickering issue does not seem to be a duplicate of any previously reported bugs, so I will open a new bug for the flickering issue.