Frequent Wayland crashes with only a few applications running

Running bazzite-nvidia-open with Intel i5-3570K, nVidia GTX 1650, 2 monitors. nVidia driver version 580.82.07 I am very technically familiar with Linux but new to Bazzite. I am loving it so far, but have seen 5+ Wayland crashes in the last 2 days, which is a big headache because I run lots of terminal windows with editors, and they all get killed due to the Wayland restart. One of the only thing sthat DOESN'T get killed is System Monitor, if it was running. edit This appears to be resolved by disabling the media player widget in the systray. It may be unrelated, but I also changed both monitors to use a solid color instead of an image background.
46 Replies
AgentElrond
AgentElrondOP4w ago
On previous days, I would just be using the system normally with a few applications open (default terminals, Firefox, Discord, sometimes a Wine application) Sleep hasn't been involved previously, but today's crash was immediately after I resumed from sleep, the desktop appeared normal, but when I clicked between windows or something of that nature, I immediately had a crash and restart.
AgentElrond
AgentElrondOP4w ago
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x300c, surface: 0x5580c9d8a000
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5580cf4799a0
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5580cf4799a0
flatpak[449886]: 16:42:08.969 › The resource https://discord.com/assets/189422196a4f8b53.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/ce3b8055f5114434.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/cb2006dbced0e246.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/7a6a566c2e88a35d.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/e52f0cba712e2fb4.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/dd24010f3cf7def7.woff2 was preloaded using link preload but n>
kernel: kwin_wayland[413257]: segfault at 0 ip 0000000000000000 sp 00007ffcfbd8d678 error 14 likely on CPU 3 (core 3, socket 0)
kernel: Code: Unable to access opcode bytes at 0xffffffffffffffd6.
audit[413257]: ANOM_ABEND auid=1000 uid=1000 gid=1000 ses=18 subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 pid=413257 co>
systemd-coredump[465208]: Process 413257 (kwin_wayland) of user 1000 terminated abnormally with signal 11/SEGV, processing...
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x300c, surface: 0x5580c9d8a000
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5580cf4799a0
plasmashell[413476]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5580cf4799a0
flatpak[449886]: 16:42:08.969 › The resource https://discord.com/assets/189422196a4f8b53.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/ce3b8055f5114434.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/cb2006dbced0e246.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/7a6a566c2e88a35d.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/e52f0cba712e2fb4.woff2 was preloaded using link preload but n>
flatpak[449886]: 16:42:08.970 › The resource https://discord.com/assets/dd24010f3cf7def7.woff2 was preloaded using link preload but n>
kernel: kwin_wayland[413257]: segfault at 0 ip 0000000000000000 sp 00007ffcfbd8d678 error 14 likely on CPU 3 (core 3, socket 0)
kernel: Code: Unable to access opcode bytes at 0xffffffffffffffd6.
audit[413257]: ANOM_ABEND auid=1000 uid=1000 gid=1000 ses=18 subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 pid=413257 co>
systemd-coredump[465208]: Process 413257 (kwin_wayland) of user 1000 terminated abnormally with signal 11/SEGV, processing...
audit: BPF prog-id=440 op=LOAD
audit: BPF prog-id=441 op=LOAD
audit: BPF prog-id=442 op=LOAD
systemd[1]: Started systemd-coredump@23-465208-0.service - Process Core Dump (PID 465208/UID 0).
audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-coredump@23-465>
systemd-coredump[465209]: [🡕] Process 413257 (kwin_wayland) of user 1000 dumped core.

Module libspa-videoconvert.so from rpm pipewire-1.4.7-1.fc42.bazzite.0.0.git.6161.88a10687.x86_64
Module org.kde.desktop.so from rpm kf6-qqc2-desktop-style-6.17.0-1.fc42.x86_64
Module kwin_windowswitcher.so from rpm kwin-6.4.4-5.fc42.x86_64
Module KirigamiPlasmaStyle.so from rpm libplasma-6.4.4-1.fc42.x86_64
Module libwebpmux.so.3 from rpm libwebp-1.5.0-2.fc42.x86_64
audit: BPF prog-id=440 op=LOAD
audit: BPF prog-id=441 op=LOAD
audit: BPF prog-id=442 op=LOAD
systemd[1]: Started systemd-coredump@23-465208-0.service - Process Core Dump (PID 465208/UID 0).
audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=systemd-coredump@23-465>
systemd-coredump[465209]: [🡕] Process 413257 (kwin_wayland) of user 1000 dumped core.

Module libspa-videoconvert.so from rpm pipewire-1.4.7-1.fc42.bazzite.0.0.git.6161.88a10687.x86_64
Module org.kde.desktop.so from rpm kf6-qqc2-desktop-style-6.17.0-1.fc42.x86_64
Module kwin_windowswitcher.so from rpm kwin-6.4.4-5.fc42.x86_64
Module KirigamiPlasmaStyle.so from rpm libplasma-6.4.4-1.fc42.x86_64
Module libwebpmux.so.3 from rpm libwebp-1.5.0-2.fc42.x86_64
If anyone has any specific suggestions or questions, please let me know and thanks so much for your time. I found https://discuss.kde.org/t/with-fedora-42-especially-kernel-6-15-10-kwin-dumps-core-when-resuming-from-suspend/38942/12 which looks somewhat similar, and I've superstitiously changed my wallpaper from default to a solid color, just in case. I will report if I see any further crashes.
porkloin
porkloin4w ago
new system? or are you coming from another distro/installation?
AgentElrond
AgentElrondOP4w ago
Brand new install, clean disk
porkloin
porkloin4w ago
i guess what i mean is, is the hardware validated in another OS to be running fine?
AgentElrond
AgentElrondOP4w ago
used the bazzite-nvidia-open ISO, haven't customized anything other than installing a couple apps I think Sorry yes, ran in Windows 10 for years with no crashes
porkloin
porkloin4w ago
👍
AgentElrond
AgentElrondOP4w ago
Things I have NOT tried: * Changing Bazzite versions * Changing Wayland versions * Changing nVidia driver versions * Turning off Wayland / Plasma (I want to keep KDE if possible) Steam was NOT running in the latest crash shortly after resume from sleep, random note. Discord was, using the latest official Linux client
porkloin
porkloin4w ago
unfortunately changing nvidia driver/bazzite version/wayland version in isolation isn't really doable since they're packaged into the OS "immutably" on bazzite, but nvidia driver issues in particular i wouldn't be surprised to find at root i know a lot of people have weird issues with nvidia drivers 🙁
AgentElrond
AgentElrondOP4w ago
understandable, I just was surprised to see so many wayland crashes in 2 days, which unfortunately is bordering on unusable because I lose all my terminals every time
porkloin
porkloin4w ago
if you want to isolate KDE plasma you could download the live ISO of the gnome version, run that for a day or two from live env (annoying because no persistence) and see if you get any crashes at all there
AgentElrond
AgentElrondOP4w ago
understood thank you for your time, by the way
porkloin
porkloin4w ago
np! it does look like kwin is crashing. it sounds like the rest of your system stays up though? just out of curiosity, are you using any kwin scripts or extensions or anything like that? custom themes?
AgentElrond
AgentElrondOP4w ago
No extension, no custom theme, no scripts, clean install The system stays up, but I lose all applications (except, oddly, System Monitor if it was running) My session also stays logged in. The last time I looked at the journalctl crash log, it said "The Wayland connection broke" but I didn't dig in to see if there was a full segfault log.
porkloin
porkloin4w ago
interesting. i don't have enough nvidia experience to know what is "expected" but i can't imagine that's a normal amount of just random desktop crashing. do you have hdr on in kde?
AgentElrond
AgentElrondOP4w ago
I'm a little confused why libspa-videoconvert.so is being used, unless it's related to Discord. no HDR I always have several Firefox windows, and maybe 3-4 terminals, and usually Discord.
porkloin
porkloin4w ago
GitHub
KDE desktop is turning black and crashing. · Issue #3234 · ublue-...
Describe the bug KDE desktop is crashing - I think my KDE version is outdated how to fix it? An issue started today. I am using my computer normally doing light tasks like web browsing. Then I see ...
AgentElrond
AgentElrondOP4w ago
crash log looks very similar Behavior not exactly the same I think? I don't notice the taskbar flash, I just see it go away entirely. Black screen with blinking text cursor in both monitors, like a full Wayland restart. And all apps get killed I do indeed have media player in system tray enabled (by default), I'll turn that off just for fun @porkloin huh, disabling media player in system tray while youtube was playing made the taskbar vanish for a few seconds, then come back, but applications stayed running. I'll try this for a while.
porkloin
porkloin4w ago
yeah it's worth a shot since some people seemed to indicate that as a source for crashes :\
AgentElrond
AgentElrondOP4w ago
I also see https://github.com/ublue-os/aurora/issues/992#issuecomment-3317265893 Which mentions it's fixed in Plasma 6.5.0 and 6.4.6 (I'm using Plasma 6.4.4) Is there a simple way in Bazzite to query which Qt version is installed?
porkloin
porkloin4w ago
rpm -qa | grep qt or something like that or just ctrl f through the output of rpm -qa oh there's like a million things with qt in them, lol
AgentElrond
AgentElrondOP4w ago
I see a bunch of 6.9.2 in my qt libraries I think for now I'm going to try this, for science purposes: * I changed my desktop to solid black, not the default cube room (some reports mentioned JPEG desktops and crashes). * I disabled the system tray media player widget. Thanks for pointing me in interesting directions
porkloin
porkloin4w ago
the last comment in that thread indicates that the bug was fixed in plasma 6.5 which might be coming with upstream fedora 43 which should be out very soon https://fedorapeople.org/groups/schedule/f-43/f-43-key-tasks.html the schedules are kinda annoying to read but anywhere between oct 28 and nov 11 we should see next fedora version which would include the update. you could maybe try to downgrade but that might be dicey. on bazzite it's much easier to just rebase to an old image from something like F41 depending on when the regression was introduced
AgentElrond
AgentElrondOP4w ago
I'll see how it goes before I do anything more drastic an upgrade would be ideal Are new Fedora versions immediately available in a desktop Bazzite build, or is it normal for there to be a delay?
porkloin
porkloin4w ago
since you mentioned you're new on bazzite but have linux experience overall, the model on immutable is kinda odd, the preference is to treat the OS and deps like DE as one big blob and just rebase back and forth rather than trying to do overlays/patches/overrides hence "atomic" os, treat the whole thing as one big deployment
AgentElrond
AgentElrondOP4w ago
I did see that's a big difference with the immutable model, yes So I didn't know if I needed to do an entire Bazzite system upgrade to get a newer Fedora version or whatever ( once it's out )
porkloin
porkloin4w ago
it's sorta weird, even minor updates for bazzite are a complete redeployment of the entire os image
AgentElrond
AgentElrondOP4w ago
I can see the appeal because it's more controllable, but also the downside if I'm an expert user and need to fix anything myself
porkloin
porkloin4w ago
yeah. you can layer stuff if you want to do patch-style overrides but it makes getting new updates slow so most people avoid it typically if i want to get a different version of a package my decision tree is like:
- can i just get it in a VM/distrobox? (not doable for "core os functionality" like DE/WM)
- if yes, get it in distrobox
- if no:
- is the dependency an older version of a package i already have?
- if yes, rollback to an older bazzite image with that package version
- if no, layer via rpm-ostree if possible
- can i just get it in a VM/distrobox? (not doable for "core os functionality" like DE/WM)
- if yes, get it in distrobox
- if no:
- is the dependency an older version of a package i already have?
- if yes, rollback to an older bazzite image with that package version
- if no, layer via rpm-ostree if possible
for stuff that is bleeding edge, we're kinda outta luck sometimes like for example you can't just upgrade to latest kwin/plasma since we're stuck with whatever we can get from upstream package management from fedora project, and there's no "bazzite rawhide" requivalent
AgentElrond
AgentElrondOP4w ago
so for new fedora versions, I need to wait until there's a new "updated bazzite atomic image" or somesuch?
porkloin
porkloin4w ago
yep, so for minor or major version updates, it's always ujust update and then the entire os gets upgraded in-place, then any layers you have with rpm-ostree get installed on top. https://github.com/ublue-os/bazzite/releases shows all the releases
GitHub
Releases · ublue-os/bazzite
Bazzite makes gaming and everyday use smoother and simpler across desktop PCs, handhelds, tablets, and home theater PCs. - ublue-os/bazzite
porkloin
porkloin4w ago
it does look like they're testing f43 on the unstable tag currently, but ymmv there the cool thing about this model though is that you can rebase and "pin" on any release listed in the github releases page. so if something breaks you can rollback to a version from whenever, even if it's like last year and it'll be identical to when that version came out and all the deps will be swapped too
AgentElrond
AgentElrondOP4w ago
@porkloin for future reference, the easiest way to see stats on stuff (other than graphics driver version) is System Settings -> "About this System" That includes stuff like KDE plasma version 6.4.4, Qt version 6.9.2
Peteydog
Peteydog4w ago
for future reference, open up info centre and Qt version is right between the KDE frameworks version and the kernel version
AgentElrond
AgentElrondOP4w ago
@porkloin @Peteydog if you're curious, I haven't had a single Plasma / Wayland crash since changing to a solid color desktop (instead of the default blue/purple cube room), and disabling the default media player systray widget. I assume it's the widget, but I'm keeping both changes out of superstition. 😂 It might be worth disabling that widget by default if anyone else runs into those crashes, but maybe it's rare enough not to matter.
porkloin
porkloin4w ago
Nice!
AgentElrond
AgentElrondOP4w ago
@porkloin I spoke too soon, lost all my applications today (several Firefox windows, terminals, and Discord) while I wasn't even listening to media
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5624a365d5f0
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: Could not create EGL surface (EGL error 0x300b)
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x0
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: error marshalling arguments for import_timeline: dup failed: Too many open files
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: Error marshalling request for wp_linux_drm_syncobj_manager_v1.import_timeline: Too many open files
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5624a5d98b80
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: The Wayland connection experienced a fatal error: Too many open files
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5624a365d5f0
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: Could not create EGL surface (EGL error 0x300b)
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x0
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: error marshalling arguments for import_timeline: dup failed: Too many open files
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: Error marshalling request for wp_linux_drm_syncobj_manager_v1.import_timeline: Too many open files
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: qt.qpa.wayland: eglSwapBuffers failed with 0x3000, surface: 0x5624a5d98b80
Oct 27 14:28:29 DESKTOP-VV1DUT5 plasmashell[2080]: The Wayland connection experienced a fatal error: Too many open files
Oct 27 14:28:34 DESKTOP-VV1DUT5 systemd-coredump[82470]: [🡕] Process 1909 (kwin_wayland) of user 1000 dumped core.

Module libspa-videoconvert.so from rpm pipewire-1.4.9-1.fc42.x86_64
Oct 27 14:28:34 DESKTOP-VV1DUT5 systemd-coredump[82470]: [🡕] Process 1909 (kwin_wayland) of user 1000 dumped core.

Module libspa-videoconvert.so from rpm pipewire-1.4.9-1.fc42.x86_64
porkloin
porkloin4w ago
that is a weird error
AgentElrond
AgentElrondOP4w ago
Yet again in libspa-videoconvert.so
porkloin
porkloin4w ago
what browser do you use?
AgentElrond
AgentElrondOP4w ago
Firefox, standard Bazzite version I think I rebooted a day or so ago, so it hasn't been running for weeks or anything
porkloin
porkloin4w ago
when i was looking at it last week i was seeing some stuff about libspa-videoconvert getting used for some browsers for media playback and stuff super odd
AgentElrond
AgentElrondOP4w ago
ulimit -a shows open files set to 1024 as https://discussion.fedoraproject.org/t/troubles-increasing-soft-open-file-limit/148383 describes which does seem really low depending on how some apps behave I'm not (knowingly) using explicit sync, but I do have nVidia https://forums.developer.nvidia.com/t/fd-leak-with-explicit-sync-and-kde-plasma/317293/31 talks about some potential leaks with nvidia drivers that cause unavoidable crashes, no idea what makes them happen faster, if that's what's going on. I haven't searched any github repos yet, but I don't see any obvious reports besidse those 2 threads.
Peteydog
Peteydog4w ago
i'm on the intel arc b580 for context
AgentElrond
AgentElrondOP3w ago
Just for data recording purposes, someone in #bazzite mentioned crashing, seemingly due to the systray media widget, on AMD. To emphasize, that appears to be the original reason I made this thread, and is distinct from the "Too many open files" issue above which has only occurred once thus far. @porkloin just crashed a second time, too many open files again, libspa-videoconvert yet again Was just listening to a long youtube video with terminals and firefox and discord open I did some brief searching but haven't found a good approach. If I understand correctly, Bazzite ships with an open file limit of 1024 (soft) and 1048576 (hard). I want a way to monitor the open file count periodically so I can detect when it gets close to the limit but BEFORE the crash occurs. That would let me maybe identify what's leaking, e.g. if it's the nVidia driver or something else. For better visibility and more focus, I'm going to close this issue and open a new one regarding the open file limit.

Did you find this page helpful?