I’m glad the issue with startplasma-wayland was mentioned here. I am also using mesa-nonfree and can’t start KDE Plasma. I’m sorry that I’m talking about it, but I’d like mods to realize that it was very valuable for a real forum reader for it to have been discussed in this announcement thread, however unofficial the status of that repo is.
Just some thoughts for those who use nonfree mesa on the unstable branch. You wouldn’t have had any issues if you were using proper Arch.
The Manjaro company has clearly stated that there is no interest in bringing back Mesa HW acceleration for patent encumbered codecs.
I switched to proper Arch the day they dropped HW acceleration and would advise anyone else to do so, who relies on that. Otherwise you will hunt down problems on every Mesa update in the future.
Manjaro should also put a big red banner on the Homepage, which tells people clearly about the status quo of Mesa, especially for AMD users.
7 Likes
regarding mesa-nonfree
any idea why i could use “downgrade” to install 24.1.3 to fix this which seems to be more uptodate than manjaro unstable?
Because downgrade takes packages from Arch?
I fully agree and migrated to arch just now.
3 Likes
huh turns out i should be using “manjaro-downgrade”, but it still downloads arch packages (???)
kde segfaults on last upgrade, i have to switch back to testing for the moment
jul 06 18:30:53 alex-b450aoruselite systemd-coredump[1811]: [🡕] Process 1762 (ksmserver) of user 1000 dumped core.
Stack trace of thread 1762:
#0 0x0000706270ca8e44 n/a (libc.so.6 + 0x94e44)
#1 0x0000706270c50a30 raise (libc.so.6 + 0x3ca30)
#2 0x0000706270c384c3 abort (libc.so.6 + 0x244c3)
#3 0x000070627128c643 n/a (libQt6Core.so.6 + 0x8c643)
#4 0x000070627128cde7 _ZNK14QMessageLogger5fatalEPKcz (libQt6Core.so.6 + 0x8cde7)
#5 0x00007062718daa45 n/a (libQt6Gui.so.6 + 0xdaa45)
#6 0x000070627197b038 _ZN22QGuiApplicationPrivate21createEventDispatcherEv (libQt6Gui.>
#7 0x0000706271348c65 _ZN23QCoreApplicationPrivate4initEv (libQt6Core.so.6 + 0x148c65)
#8 0x000070627197b0ce _ZN22QGuiApplicationPrivate4initEv (libQt6Gui.so.6 + 0x17b0ce)
#9 0x0000706271973cdd _ZN15QGuiApplicationC2ERiPPci (libQt6Gui.so.6 + 0x173cdd)
#10 0x00005bc8308aa4bd n/a (ksmserver + 0x94bd)
#11 0x0000706270c39c88 n/a (libc.so.6 + 0x25c88)
#12 0x0000706270c39d4c __libc_start_main (libc.so.6 + 0x25d4c)
#13 0x00005bc8308acd05 n/a (ksmserver + 0xbd05)
Stack trace of thread 1807:
#0 0x0000706270d1c39d __poll (libc.so.6 + 0x10839d)
#1 0x000070627071e8fd n/a (libglib-2.0.so.0 + 0xbe8fd)
#2 0x00007062706bbf95 g_main_context_iteration (libglib-2.0.so.0 + 0x5bf95)
#3 0x00007062715a2cbd _ZN20QEventDispatcherGlib13processEventsE6QFlagsIN10QEventLoop17>
#4 0x000070627134f01e _ZN10QEventLoop4execE6QFlagsINS_17ProcessEventsFlagEE (libQt6Cor>
#5 0x000070627143a410 _ZN7QThread4execEv (libQt6Core.so.6 + 0x23a410)
#6 0x0000706272185e4e n/a (libQt6DBus.so.6 + 0x2de4e)
#7 0x00007062714c9747 n/a (libQt6Core.so.6 + 0x2c9747)
#8 0x0000706270ca6ded n/a (libc.so.6 + 0x92ded)
#9 0x0000706270d2a0dc n/a (libc.so.6 + 0x1160dc)
ELF object binary architecture: AMD x86-64
System started but it leaves me with no graphic environment, switching back to testing solves the issue momentarily.
Some more related logs
jul 06 18:30:53 alex-b450aoruselite gmenudbusmenuproxy[1977]: qt.qpa.xcb: could not connect to display
jul 06 18:30:53 alex-b450aoruselite gmenudbusmenuproxy[1977]: qt.qpa.plugin: From 6.5.0, xcb-cursor0 or libxcb-cursor0 is needed to load the Qt xcb>
jul 06 18:30:53 alex-b450aoruselite gmenudbusmenuproxy[1977]: qt.qpa.plugin: Could not load the Qt platform plugin "xcb" in "" even though it was f>
jul 06 18:30:53 alex-b450aoruselite gmenudbusmenuproxy[1977]: This application failed to start because no Qt platform plugin could be initialized. >
Available platform plugins are: eglfs, minimal, linuxfb, vkkhrdisplay, offscreen, way>
jul 06 18:30:53 alex-b450aoruselite systemd[1543]: app-org.kde.plasma\x2dfallback\x2dsession\x2drestore@autostart.service: Skipped due to 'exec-con>
jul 06 18:30:53 alex-b450aoruselite systemd[1543]: Condition check resulted in Plasma Session Restore being skipped.
EDIT 2024/07/08: This seems related to my issue https://forum.manjaro.org/t/unstable-manjaro-community-mesa-nonfree-codecs/135012/148
I have indeed mesa non-free installed, I have not tested if installing simple mesa from manjaro would resolve the issue
EDIT 2024/07/09: Related issue with same log errors
Qt Platform Plugins Won’t Initialize After Upgrade - Support / System Updates - Manjaro Linux Forum
in the latest iso (Plasma) that boots X11 apart from persisting problem of not changing the menu to Application Menu from Application Launcher, when I logout to switch to Wayland the screen remain black with cursor in the middle, I’ve to hit Ctrl+Alt+Backspace to have the login screen and the change to Wayland, login out of Wayland works
Edit: Intel only iGPU
7 posts were split to a new topic: Paprefs and pipewire-pulse are in conflict
A post was merged into an existing topic: Paprefs and pipewire-pulse are in conflict
tested several unstable isos, for me what are fine are 20240702-03 thay boot into wayland and I’m not having issue with log out or menu change
btw is there a reason that all the isos, included yesterday iso, are keeping kernel 6.8 that is EOL?
This update completely broke sddm for me, throwing me into tty on boot.
I used systemctl enable --force sddm.service, which now makes my system boot into KDE again, but logging out and shutting down from KDE is still broken.
What update? Your profile says you’re using the stable branch.
I’ve been using unstable for a while now, my profile is just outdated. I updated yesterday on the 13th.
A post was split to a new topic: Updated from stable to unstable and now shutdown doesn’t work
Continuing down the rabbit hole of nvidia 555, I’m still getting the same FF and wake crashes (same signatures) while on the newly-released kernel 6.10 (linux610-nvidia (555.58.02-1). The weird part about the ff crashes is they don’t seem to have a trigger but don’t appear to be video or streaming-related that I can tell. Just doing regular online shopping is enough to kill it. Very odd. But, I’m not prepared to downgrade just yet.
A post was split to a new topic: Unable to play minecraft
Today I received the egl-wayland 2:1.1.14-1 update, which adds explicit sync, which leads to firefox crashing (perhaps some other applications are also affected), at the moment the fix is not ready, although the problem has been known for ~ 2 months. Rolling back to the previous version solves the problem.
UPD: It looks like they are trying to quickly resolve the problem. The arch repository has backported fixes for firefox, but it looks like fixes from mesa are still needed. The problem also affects thunderbird and tor browser.
1 Like
Explicit sync with Nvidia is currently still a hit and miss. A lot of libraries and other stuff needs to be added to make it work. That is why we have isolated the 555 driver series to our unstable branch so far. Also Plasma 6.1 in combination creates issues with it. Same goes for that. It might get more dynamic with 560 driver series when Nvidia starts to support newer cards only with their open-source drivers and older still with their proprietary ones. Currently I would avoid to use Nvidia graphics on Linux until all parties had catched up and Nvidia starts to rollout working drivers again.
4 Likes
Seeing a lot of issues with the KDE plasma 6.1.3 upgrade. (I didn’t have any issues with 6.1.2 or 6.1.1.)
I have a main panel at the top of both screens, and a secondary panel at the bottom of both screens.
The panel at the bottom is fit-to-width, auto-hide, and has, from left to right, an app launcher, 14 app shortcut widgets, a workspace switcher, clipboard plasmoid, weather plasmoid, and sticky note plasmoid.
I’m seeing the following problems:
- When the widget was set to auto-hide initially, it wouldn’t un-hide. I could get the two auto-hide panels to show by going into app configuration mode, however all of my panels would freeze shortly afterward requiring me to run
killall -9 plasmashell && plasmashell --replace &. This continued until I eventually set them to always show. - When in always show mode, the two bottom panels now load, but only show the application launcher icon. I can get the rest of the panels to show by entering the panel config view, however they’re still fairly unstable and crash eventually.
- When the screens go to sleep for a while, or I turn the screens off, when I return, the panels are often crashed, requiring running
plasmashell --replace & - krunner stops launching from my hotkey randomly. It sometimes returns randomly after some hours.
- This may not be new in 6.1.3, but the status bar lock key indicator seems to be printing this error ~50 times in the logs, unsure of the cause (I’ve noticed it because I’ve been watching to logs now):
org.kde.plasma.keyboardindicator: Unknown key Qt::Key_Alt
I’ve seen various errors since the update associated with the crashes, however the behavior, and errors, are often inconsistent.
When it crashes on monitor resume, it gives me:
...
org.kde.pulseaudio: No object for name "@DEFAULT_SOURCE@"
org.kde.pulseaudio: No object for name "alsa_output.pci-0000_01_00.1.hdmi-surround.monitor"
trying to show an empty dialog
xdg_surface#473: error -1: invalid window geometry size (0x0)
The Wayland connection experienced a fatal error: Protocol error
This is an error I received when I was trying to make auto-hide work:
file:///usr/share/plasma/shells/org.kde.plasma.desktop/contents/configuration/PanelConfiguration.qml:536:9: QML SpinBox: Binding loop detected for property "implicitWidth"
file:///usr/share/plasma/plasmoids/org.kde.panel/contents/ui/ConfigOverlay.qml:329:17: QML SpinBox: Binding loop detected for property "implicitWidth"
kf.plasma.quick: Exposed with no visual parent. Window positioning broken.
kf.windowsystem.wayland: Failed to recreate shadow for PlasmaQuick::AppletPopup_QML_199(0x649471715b60, name="popupWindow")
kf.plasma.quick: Exposed with no visual parent. Window positioning broken.
kf.plasma.quick: Exposed with no visual parent. Window positioning broken.
file:///usr/share/plasma/plasmoids/org.kde.panel/contents/ui/main.qml:88: TypeError: Cannot read property 'index' of null
file:///usr/share/plasma/shells/org.kde.plasma.desktop/contents/configuration/PanelConfiguration.qml:536:9: QML SpinBox: Binding loop detected for property "implicitWidth"
$ org.kde.plasma.containmentlayoutmanager: Trying to take space not available BasicAppletContainer_QMLTYPE_39_QML_77(0x64946982b830, id="appletContainer", parent=0x649469eb7170, geometry=2607,1069 773x54)
org.kde.plasma.containmentlayoutmanager: Trying to take space not available BasicAppletContainer_QMLTYPE_39_QML_77(0x64946a384930, id="appletContainer", parent=0x649469eb7170, geometry=2601,1064 773x54)
org.kde.plasma.containmentlayoutmanager: Trying to take space not available BasicAppletContainer_QMLTYPE_39_QML_77(0x649469fab000, id="appletContainer", parent=0x649469eb7170, geometry=2369,1071 773x54)
Seeing this frequently, likely associated with the lock key indicator:
file:///usr/share/plasma/plasmoids/org.kde.plasma.keyboardindicator/contents/ui/main.qml:61:13: Unable to assign [undefined] to bool
js: Error with Permissions-Policy header: Unrecognized feature: 'ch-ua-form-factors'.
There are various other error, but overall, the KDE Plasma 6.1.3 release has been quite problematic, which is a shame because 6.1.2, 6.1.1, and 6.1.0 were all pretty good (generally after some fiddling about).
I can still use my desktop at least, with the exception of the failing krunner, and the lack of being able to auto-hide my bottom panel. The actual crashes are much less frequent in this setup, but still happen, and relaunching works, as long as I get the bottom panel to render after the restart by entering panel config mode.
Edit (next day, after restarting, following the update to the new qt6 wayland and restarting):
I restarted again after the qt6 wayland update, hoping it was resolved, however now, when trying to exit the panel configuration mode, plasmashell will reliably freeze. I’ve also tried setting the panel to “fill width” now, and it shows the full panel from the start of plasmashell running (there was a bug on the kde bug tracker that vaguely sounded related to this, in that “fit to content” was causing issues).
But, having the panel crash when trying to exit panel config mode is still quite an issue (in addition to the other new requirements regarding limited panel configuration options due to crashes). It doesn’t appear to always crash when trying to exit panel configuration mode, just sometimes. :<
Edit (again):
Forgot to say, also, system settings won’t launch now. “Quick Settings” will appear in my taskbar, but I can’t get the window to actually show. Right clicking on it and selecting “Move” also doesn’t make it appear, so it isn’t just off the display somewhere. Quite the disaster of an update, fr D: (Bonus edit on this one! settings still opens if i open a specific settings menu from the application picker, like “Display Configuration”. It just doesn’t open correctly if I try to open the main settings application)
Example experience returning to the computer after a few hours (from just now):
Sometimes when the screen comes back on, plasmashell isn’t crashed, but only one monitor will turn on. Since regular settings won’t open anymore, I try to open display settings, however it isn’t possible to navigate to the window (it’s invisible), which requires killing it from the terminal. After reopening it in a second workspace (to be safe), it says both displays are connected and active (they are not, according to the workspace switcher, my mouse, etc.), so I reactivate the display manually, then I change the refresh rate so I can get the “revert settings” menu to get the monitor to turn on. After this, the krunner launcher no longer starts from my hotkey. Also, about half of my windows are positioned underneath my “always on top” main taskbar at the top of the screen, meaning I need to use my move hotkey + mouse to make them usable again, or trigger the move dialogue via right clicking the taskbar.
1 Like