[SOLVED] Several crashes GNOME with NVIDIA cards

amhairghin · 2024-04-02 06:51:14

Just note that if you update mesa to the latest version (1:24.0.4-2) it is highly probable that GNOME will not start (NVIDIA card and X11). Simply downgrading to 1:24.0.4-1 fixes the problem. All the problems derived from the GNOME 46 update (mainly discussed here https://gitlab.gnome.org/GNOME/mutter/-/issues/3384) are still present, but at least it starts.

UPDATE: I initially blamed it on the mutter upgrade, but I was checking that I downgraded both at the same time. And the culprit was not the mutter upgrade to 46.0-2 but mesa.

Last edited by amhairghin (2024-04-09 08:36:54)

seth · 2024-04-02 07:27:33

Many other users have faced insta-crasheson or after login w/ the previous version, you might have been shielded and now run into that?
https://bbs.archlinux.org/viewtopic.php?id=294385

Edit: do you still have a backtrace/log of the crash in https://wiki.archlinux.org/title/Core_d … _core_dump or previous system journals?

Last edited by seth (2024-04-02 07:29:38)

amhairghin · 2024-04-02 08:04:31

seth wrote:

Edit: do you still have a backtrace/log of the crash in https://wiki.archlinux.org/title/Core_d … _core_dump or previous system journals?

The coredump does not seem to shed much light: https://lesma.eu/qilugaru

I can't do much testing right now either because the machine is the one I use for work and I'm on business hours.

The only interesting part I see in the journal is this one (which doesn't say much):

abr 02 08:21:35 hell gnome-shell[1696]: GNOME Shell started at Tue Apr 02 2024 08:21:34 GMT+0200 (hora de verano de Europa central)
abr 02 08:21:35 hell gnome-shell[1696]: Registering session with GDM
abr 02 08:21:35 hell org.gnome.Shell.desktop[1125]: X connection to :0 broken (explicit kill or server shutdown).
abr 02 08:21:35 hell gsd-sharing[1213]: Error releasing name org.gnome.SettingsDaemon.Sharing: La conexión está cerrada
abr 02 08:21:35 hell gsd-rfkill[1238]: Error releasing name org.gnome.SettingsDaemon.Rfkill: La conexión está cerrada
abr 02 08:21:35 hell gsd-smartcard[1241]: Error releasing name org.gnome.SettingsDaemon.Smartcard: La conexión está cerrada
abr 02 08:21:35 hell gsd-print-notif[1232]: Error releasing name org.gnome.SettingsDaemon.PrintNotifications: La conexión está cerrada
abr 02 08:21:35 hell gsd-datetime[1245]: Error releasing name org.gnome.SettingsDaemon.Datetime: La conexión está cerrada
abr 02 08:21:35 hell gsd-screensaver[1258]: Error releasing name org.gnome.SettingsDaemon.ScreensaverProxy: La conexión está cerrada
abr 02 08:21:35 hell gsd-a11y-settin[1272]: Error releasing name org.gnome.SettingsDaemon.A11ySettings: La conexión está cerrada
abr 02 08:21:35 hell gsd-housekeepin[1277]: Error releasing name org.gnome.SettingsDaemon.Housekeeping: La conexión está cerrada
abr 02 08:21:35 hell gsd-sound[1263]: Error releasing name org.gnome.SettingsDaemon.Sound: La conexión está cerrada
abr 02 08:21:35 hell gsd-media-keys[1250]: Error releasing name org.gnome.SettingsDaemon.MediaKeys: La conexión está cerrada
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:84
abr 02 08:21:35 hell polkitd[961]: Unregistered Authentication Agent for unix-session:1 (system bus name :1.23, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale es_ES.UTF-8) (disconnected from bus)
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:70
abr 02 08:21:35 hell telegram-desktop[1841]: QPainter::begin: Paint device returned engine == 0, type: 2
abr 02 08:21:35 hell telegram-desktop[1841]: QWidget::render: Cannot render with an inactive painter
abr 02 08:21:35 hell gnome-shell[1696]: Window manager warning: MetaSyncRing: Sync object is not ready -- were events handled properly?
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:68
abr 02 08:21:35 hell gnome-shell[1696]: Window manager warning: MetaSyncRing: Sync object is not ready -- were events handled properly?
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:67
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:64
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:66
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) UnloadModule: "libinput"
abr 02 08:21:35 hell /usr/lib/gdm-x-session[1067]: (II) systemd-logind: releasing fd for 13:65
abr 02 08:21:39 hell kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:4:0:1173
abr 02 08:21:39 hell gdm[918]: Gdm: GdmCommon: process (pid:997, command 'gdm-session-worker [pam/gdm-launch-environment]') isn't dying after 5 seconds, now ignoring it.
abr 02 08:21:41 hell /usr/lib/gdm-x-session[1578]: (II) event20 - MOSART Semi. 2.4G Wireless Mouse: SYN_DROPPED event - some input events have been lost.
abr 02 08:21:41 hell kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:6:0:1173
abr 02 08:21:41 hell /usr/lib/gdm-x-session[1067]: (II) NVIDIA(GPU-0): Deleting GPU-0
abr 02 08:21:41 hell /usr/lib/gdm-x-session[1067]: (II) Server terminated successfully (0). Closing log file.

Specifically the following lines, but they don't tell me anything other than that it has failed.

abr 02 08:21:39 hell kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:4:0:1173
abr 02 08:21:41 hell kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:6:0:1173

seth · 2024-04-02 08:23:25

That's different from all the other recent gnome crashes and the backtrace looks more like https://bbs.archlinux.org/viewtopic.php?id=285940 which seems to have been caused by an input device (wacom tablet)

Please post your Xorg log, https://wiki.archlinux.org/title/Xorg#General

As a generic advice, try to enable

Option  "TripleBuffer"          "True"

(but that's only because of the buffer swap in the backtrace)

I'n not sure the mesa update is directly relevant here and I'm also not sure what to make out of that journal snippet

08:21:35 hell org.gnome.Shell.desktop[1125]: X connection to :0 broken
…
08:21:41 hell /usr/lib/gdm-x-session[1067]: (II) Server terminated successfully (0)

But this might be a blur of the GDM and session instaces of gnome-shell

amhairghin · 2024-04-02 09:07:16

You won't believe it, I just updated again and applied these updates:

[2024-04-02T10:56:04+0200] [ALPM] upgraded xz (5.6.1-2 -> 5.6.1-3)
[2024-04-02T10:56:04+0200] [ALPM] upgraded libarchive (3.7.2-1 -> 3.7.2-2)
[2024-04-02T10:56:04+0200] [ALPM] upgraded util-linux-libs (2.40-1 -> 2.40-2)
[2024-04-02T10:56:04+0200] [ALPM] upgraded mesa (1:24.0.4-1 -> 1:24.0.4-2)
[2024-04-02T10:56:04+0200] [ALPM] upgraded sdl2 (2.30.1-1 -> 2.30.2-1)
[2024-04-02T10:56:04+0200] [ALPM] upgraded syncthing (1.27.4-1 -> 1.27.5-1)
[2024-04-02T10:56:04+0200] [ALPM] upgraded util-linux (2.40-1 -> 2.40-2)

And now everything works correctly (there are still the input lag problems in GNOME 46 but that's another story). The only thing I can think of is that the problem was derived from sdl2. I don't know, I'm puzzled.

seth · 2024-04-02 12:35:44

Does mutter even link libsdl2?

Please post your Xorg log, https://wiki.archlinux.org/title/Xorg#General

amhairghin · 2024-04-02 13:23:21

seth wrote:

Please post your Xorg log, https://wiki.archlinux.org/title/Xorg#General

Here is: https://lesma.eu/xonasiko

But for now it's working...

seth · 2024-04-02 13:58:20

You've an

[    17.535] (II) modeset(G0): glamor X acceleration enabled on AMD Radeon Graphics (radeonsi, raphael_mendocino, LLVM 17.0.6, DRM 3.57, 6.8.2-arch2-1)

but there're not outputs attached to it and you're running on the nvidia chip in a reverse prime setup.
If this flares up again and you can, I'd try to disable the AMD chip.
In doubt nudge the system to use nvidia, https://wiki.archlinux.org/title/Vulkan … initialize - espeically if you've amdvlk or vulkan-radeon installed.

M-Reimer · 2024-04-02 14:21:30

I also run into issues with the latest mesa update.

Nvidia 1070 with latest nvidia driver. No additional GPU.

I still get gdm but after logging in I just see that "X" cursor and everything hangs.

I've downgraded mesa for now...

I can provide logs if needed but please tell me where to find the relevant logs.

Last edited by M-Reimer (2024-04-02 14:23:34)

amhairghin · 2024-04-02 14:22:40

Yes, it is an AMD motherboard with integrated graphics that is not really used. Maybe I would have to disable it in the BIOS. In any case when it failed if it was correctly detecting the NVIDIA card.

The only thing I can think of is to let it pass and if it happens again I will upload a copy of the Xorg log to see if it tells us something.

seth · 2024-04-02 14:27:55

@M-Reimer, sure that's mesa related?
"Gnome46 crashes/hands on login" errors are very common right now (because there was a gnome update…)
You'd have to look into the system journal, there's typically not even a coredump and there're two prevalent patterns, one telling you that the screensaver state could not be determined, the other one that "META_CURRENT_TIME used to choose focus window" (see the other thread linked in #2)

M-Reimer · 2024-04-02 14:34:43

Not sure anymore. Probably not a mesa problem on my side. But I'll have to do a few more reboots to be sure. Thanks for the hint.

amhairghin · 2024-04-03 15:49:46

Well, the problem happened again. This time I have extracted the Xorg log. https://lesma.eu/kixapeku

It seems to have something to do with this https://bbs.archlinux.org/viewtopic.php?id=294385 because if I remove telegram from the autostart it doesn't crashes at startup. Although if after logging in I open an application quickly it still crashes.

seth · 2024-04-03 15:59:10

The X11 server ran fine for ~30s on the nvidia driver and then terminated cleanly.
You'll have to look at the journal, the session process (gnome) died (and 30s hints at the 25s dbus timeout)

The boys at https://bbs.archlinux.org/viewtopic.php?id=294385 figured that autostarting gnome-terminal would allow them to run whatever else immediately after and the gnome session completely stopped crashing.

amhairghin · 2024-04-03 16:30:29

The two problems are definitely related. In my particular case if I put Telegram Desktop on autostart it works, but with version 4.16.0-1, if I upgrade Telegram Desktop to 4.16.1-1 GNOME crashes. That's why it stopped working for me now, because I got the Telegram desktop update.

It is clear that the release of GNOME 46 is being complicated.

seth · 2024-04-03 16:35:10

Every major gnome update is complicated
What about 4.16.0-2 (the Qt 6.7 rebuild)?

amhairghin · 2024-04-03 17:06:54

seth wrote:

What about 4.16.0-2 (the Qt 6.7 rebuild)?

It does not work for me. It gives me a library error:

telegram-desktop: /usr/lib/libQt6Core.so.6: version `Qt_6.7' not found (required by telegram-desktop)

I understand that it will be because it is still in testing (and I don't have that repository activated, just in case).

seth · 2024-04-03 17:09:55

4.16.0-2 is not in testing but requires Qt 6.7 (like 4.16.1-1 does)?

amhairghin · 2024-04-03 17:16:11

seth wrote:

4.16.0-2 is not in testing but requires Qt 6.7 (like 4.16.1-1 does)?

As far as I can see 4.16.1-1 should not require Qt 6.7 since it is not in testing. It must have been compiled it with Qt 6.6. In fact there is a telegram-desktop-4.16.1-2 that is in testing and possibly with Qt 6.7.

seth · 2024-04-03 17:40:18

https://gitlab.archlinux.org/archlinux/ … mmits/main
The only explanation is that there was a 4.16.0-2 build against 6.7 and an 4.16.1-1 against 6.6 at the same time.

It's however moot - the curious part would have been whether the Qt6 dependency had brought some update in creating the window that was triggering the gnome crash.
If you make sure to first autostart gnome-terminal and then telegram, does it still crash? (You could add a 5s sleep delay to the telegram autostart service)

amhairghin · 2024-04-05 19:30:10

I finally updated my Arch today (I didn't have time to do it until now), there has been an xorg update (from 21.1.11-2 to 21.1.12-1) and it seems that I don't have the crash anymore.

The input delay still remain but seems that there is already a solution with this mutter change: https://gitlab.gnome.org/GNOME/mutter/- … uests/3685 (I haven't tried it personally, but in the comments they say that it solves the problem for everyone).

amhairghin · 2024-04-09 08:38:28

After 46.0-3 mutter update, no more crashes nor delays :-)

Arch Linux

#1 2024-04-02 06:51:14

[SOLVED] Several crashes GNOME with NVIDIA cards

#2 2024-04-02 07:27:33

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#3 2024-04-02 08:04:31

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#4 2024-04-02 08:23:25

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#5 2024-04-02 09:07:16

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#6 2024-04-02 12:35:44

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#7 2024-04-02 13:23:21

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#8 2024-04-02 13:58:20

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#9 2024-04-02 14:21:30

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#10 2024-04-02 14:22:40

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#11 2024-04-02 14:27:55

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#12 2024-04-02 14:34:43

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#13 2024-04-03 15:49:46

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#14 2024-04-03 15:59:10

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#15 2024-04-03 16:30:29

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#16 2024-04-03 16:35:10

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#17 2024-04-03 17:06:54

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#18 2024-04-03 17:09:55

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#19 2024-04-03 17:16:11

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#20 2024-04-03 17:40:18

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#21 2024-04-05 19:30:10

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

#22 2024-04-09 08:38:28

Re: [SOLVED] Several crashes GNOME with NVIDIA cards

Board footer