Since the 6.2.3 update I now had several times one of my Monitor completely freeze when playing a Youtube Video in my browser (Firefox).
Most times full screen but it even happened once just playing in the window and while most times at the same (second) Monitor at least one time it happened with the third.
Sound continues but I can not move my Mousepointer on that screen or interact with it any other way. Even killing Firefox does not help to Unfreeze.
What works is to log out that unfreezes the Monitor and after relogin everything works again like nothing ever happened.
Just took a look at the log for yesterdays incident and all I got in that one to two minutes time frame until realizing that it is not just a still image shown but frozen is:
kwin_wayland_drm: Pageflip timed out! This is a kernel bug
18 times, nothing else.
Google found a relatively recent KDE bugtracker entry about that, but they talk about that it is in the AMD driver part of the Kernel (and fixed now).
Well, I do not have anything AMD or use that driver.
Around the same time when 6.2.3 came out I got a update to Nvidia 565.57.01 as well and while /usr/bin/kwin_wayland runs with 2MiB used on that driver (according to nvidia-smi) I do not prime-run my browser, the intel iGPU with its i915 driver should be in use for that video.
Kernel got probably updated at least twice or trice since.
Operating System: Arch Linux
KDE Plasma Version: 6.2.3
KDE Frameworks Version: 6.8.0
Qt Version: 6.8.0
Kernel Version: 6.11.9-arch1-1 (64-bit)
Graphics Platform: Wayland
Processors: 32 × 13th Gen Intel® Core™ i9-13900HX
Memory: 62.5 GiB of RAM
Graphics Processor: Mesa Intel® Graphics
Manufacturer: TUXEDO
Product Name: TUXEDO Gemini Gen2
Little bit more “plasmashell” error output in the Log today:
Nov 21 19:19:05 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: atomic commit failed: Das Argument ist ungĂĽltig
Nov 21 19:25:01 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:06 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:11 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:16 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:21 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:26 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:31 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:36 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:41 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:46 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:51 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:25:56 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:01 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:20 xxxxxxxx plasmashell[1716]: KPackageStructure of KPluginMetaData(pluginId:"org.kde.plasma.kclock_1x2", fileName: "/usr/share/plasma/plasmoids/org.kde.plasma.kcl>
Nov 21 19:26:22 xxxxxxxx plasmashell[1716]: KPackageStructure of KPluginMetaData(pluginId:"org.kde.plasma.kclock_1x2", fileName: "/usr/share/plasma/plasmoids/org.kde.plasma.kcl>
Nov 21 19:26:23 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:28 xxxxxxxx plasmashell[1716]: KPackageStructure of KPluginMetaData(pluginId:"org.kde.plasma.kclock_1x2", fileName: "/usr/share/plasma/plasmoids/org.kde.plasma.kcl>
Nov 21 19:26:33 xxxxxxxx plasmashell[1716]: The cached device pixel ratio value was stale on window update. Please file a QTBUG which explains how to reproduce.
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x0
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: kf.plasma.quick: Exposed with no visual parent. Window positioning broken.
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: The cached device pixel ratio value was stale on window expose. Please file a QTBUG which explains how to reproduce.
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: kf.windowsystem.wayland: Failed to recreate shadow for ToolTipDialog(0x64a30d30f090)
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: qt.qpa.wayland: eglSwapBuffers failed with 0x300d, surface: 0x0
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: error creating screencast "Could not find window id {fbf0f065-9a4b-4881-b397-be52fe3b2ee0}"
Nov 21 19:26:36 xxxxxxxx plasmashell[1716]: error creating screencast "Could not find window id {fbf0f065-9a4b-4881-b397-be52fe3b2ee0}"
Nov 21 19:26:36 xxxxxxxx systemd[1394]: app-firefox@af524077d93d4ef0b9ce8acda2ef6de0.service: Consumed 41min 21.590s CPU time, 4.2G memory peak.
Nov 21 19:26:36 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:41 xxxxxxxx plasmashell[1716]: KPackageStructure of KPluginMetaData(pluginId:"org.kde.plasma.kclock_1x2", fileName: "/usr/share/plasma/plasmoids/org.kde.plasma.kcl>
Nov 21 19:26:41 xxxxxxxx plasmashell[1716]: The cached device pixel ratio value was stale on window update. Please file a QTBUG which explains how to reproduce.
Nov 21 19:26:46 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:51 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:26:52 xxxxxxxx plasmashell[1716]: The cached device pixel ratio value was stale on window update. Please file a QTBUG which explains how to reproduce.
Nov 21 19:26:56 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
Nov 21 19:27:01 xxxxxxxx kwin_wayland[1480]: kwin_wayland_drm: Pageflip timed out! This is a kernel bug
On the other hand, there are also quite a few issues reported around the i915 drivers:
Seemingly all from folks running Arch, the first one of those (10395) with some workaround suggestions - hopefully one of those might help in your case?
I’ve been seeing the same for 6.2.2 and 6.2.3 for about a month. I’m running Bazzite with Nvidia drivers (565.57.01).
Journalctl shows the same error message when the freeze happens. Toggling the TTY back and forth seems to work around the issue temporarily. I’ve also noticed that when I’m connected to a second machine with Barrier/Input-Leap that the second machine continues to get input and that it no longer perceives any resolution size on the KDE host machine to move the cursor too, effectively locking it to the client.
I’ve noticed this happen while streaming music in electron flatpak apps (Tidal)
With 6.2.4 and Kernel 6.12 (and whatever got a update as well) it is now, from almost daily, down to approximately once a week but still happening.
And I wonder if I just should wait and hope that it magically vanishes with the following updates?
How to even approach and find the culprit of a bug that is, even for me, impossible to reliably reproduce (as in reliably/directly to trigger)
I mean steps to reproduce:
Watch YT video and do nothing else, if nothing happens (even after hours of playtime), watch next and so on and on, and maybe or maybe not you stumble upon it. I mean, who in the right mind would do this?
As the only thing I tried I worked my way though the arch wiki to make sure everything for the Nvidia and Intel GPU is set up correctly and especially for Intel it is rather complicated as to what exactly apples to my integrated GPU and what of the mentioned (sub-)features mine even has.
But that has not fully helped, it is way way less frequent but still happens occasionally, for example every time I start to think “oh so many days since the last Kernel and or Nividia driver update, maybe it is gone/fixed now” I can be sure it happens just minutes after
And by the way not only YT even other streaming services.
Ya, in my case I don’t have an (active) Intel IGP. Only a couple nVidia cards.
I’m going to boot back into my Bazzite setup to see if it’s happening there, too. It’s happening frequently in Arch - and ya, with any/every streaming service. Even VLC will do it.
Same thing happens in Bazzite. So it’s not just an Arch issue.
I was able to force the issue to happen in Bazzite by starting OBS. So maybe it’s something with pipewire?
If I am not mistaken, not looked myself but seen at another thread, the Kernel log should now contain a link to the Nvidia Forums aka “where to complain” maybe it is worth a try. At this point they should already know but to add more pressure.
Found today another way to recover without logging out “SUPER+L” to Lock the system and then “ESC” to turn off the screens. It takes roughly 10 seconds until the frozen one turns off but after that it reactivates on (the next) mouse move and I can unlock back in.
I wonder if I just can turn off the screens kscreen-doctor --dpms off wait until they go black and then reactivate them all with a mouse move. That way I have not even (to Lock and) Unlock the whole system. I have set that command to a key combination, and now have to wait for the next occurrence to test that out. If that works that would probably be the fastest way to recover.
First the bad news,
neither Kernel 6.13.8-arch1-1 or Nvidia 570.133.07 fixed the issue.
Then the bad news,
just/only turning off the screen, with kscreen-doctor, did (not) work.
Well it unfroze the affected screen but at the same time crashed kwin to its core, literary a crashdump of core dot kwin dot something was created, together with a huge amount of “it’s dead Jim” (? or along that line) including background processes, that I decided to restart just to be sure everything is running again.
Strange that with additionally locking the screens nothing crashed, I will definitely retry that next time to see if it was only a fluke, though.
I have disabled all sleep and hibernation modes from the beginning just because of bad experiences in the Past, they may work on this particular Setup/Hardware, never tried.
In my “docked” Multi-Monitor home setup they are not really necessary as well, after 10 Minutes of inactivity without anything that would block (eg. video playing) the system locks and after 30 Minutes the Monitors turn off but even that nearly never happens, almost always I unlock it before.
Sure when i take the Notebook away with me sleep and or hibernation would be nice sometimes but in my case that is only a single digit percentage of use.
Or in short; Yes randomly, without entering any sleep mode first, back in November it was almost daily, now it not had happened since that last big crash 10 days ago. which is already unusual long (maybe I should better not think about it to not trigger it , knocking on wood).
Knock on wood, but I switched back over to KDE (from Hyprland) when the 570.133.07 dropped, and so far, I haven’t had a freeze! Even after several suspends/sleeps!
It still won’t let me test the lock + screen blank method.
Aka, still not a single freeze since.
Maybe a little early after I just got that new driver as well. but if I’m lucky maybe, just maybe, all that is left is related to the (unused) sleep mode (and if not now get fixed later).
Unfortunately, while it was a new record time without a freeze for sure, it just happened again.
And lock + screen blank did not work (this time).
It not only refused to blank my second external not frozen Monitor as well even the logout and back in that usually worked without a issue in the past caused many crashes so that I had to restart completely. From that point of view it got worse but that could be caused by me trying to blank the screens first(?).
If it lasts as long as last time i can live with it (as i could until now) and still hope for the best.
Bummer, and for sure not sleep mode related here,
Edit: Forgot something that was different this time .
Not directly before, earlier that session, I accidentally turned off one of the Monitors trying to access its OSD-Menu. Because things swapped around had to turn off the second one as well and turn them back on in the “correct” order for everything getting back to its right place.
Maybe that triggered something similar to the Monitors waking up after sleep mode and the freeze later that day? Don’t know, but maybe worth mentioning.
Ah!
I came to post that I think I’ve found one of the triggers for my issue - I use a KVM switch and if I switch some of my monitors over to a different machine while my Linux box is sleeping, when I wake the box up I get the issue, almost guaranteed.
Which, to the PC, looks like I’ve unplugged/turned off the monitor.
So if the same thing is happening to you, then I think we’ve found the root cause!
Not that knowing that helps us fix the issue, lol, but at least we know what’s causing it.
, knocking on wood).