[Testing Update] 2023-04-08 - Kernels, Plasma 5.27.4, Phosh, Pipewire, KDE-git

Known issues and solutions

This is a wiki post; please edit as necessary.
Please, consider subscribing to the Testing Updates Announcements RSS feed

Please RTFT (Read This Fine Thread) first before reporting the same issues over and over again!

:arrow_right: 2023-04-08


Info about recent GRUB changes

[Unstable Update] 2023-02-17 - Plasma 5.27 LTS, GNOME, Python - #51 by philm

Info about recent NVIDIA packaging changes

[Testing Update] 2023-03-25 - Kernels, Firefox, Deepin, GNOME, NVIDIA, Wine - #42 by philm


Possible kernel fs block code regression in 6.2.3 umounting usb drives

With kernel 6.2.3 when you simply plug in a usb external drive, mount it and umount it, then the journal has a kernel Oops. A bug report, that includes the journal output was submitted at. 217174 – Plugging in usb external drive, mount and umount causes kernel Oops Discussion can be found here: Re: Possible kernel fs block code regression in 6.2.3 umounting usb drives — Linux Stable Kernel Updates

kernel 5.15.99+ Black/empty screen during boot with i915 Intel gfx. No ways to do anything

Seems there is a regression introduced with 5.15.99 kernel update: [i915]drm:add_taint_for_CI [i915]] CI tainted:0x9 by intel_gt_init+0xae/0x2d0 [i915] (#8284) · Issues · drm / i915 / kernel · GitLab. My thoughts here: [Testing Update] 2023-03-11 - Kernels, Mesa 22.3.7, Phosh, LibreOffice 7.4.6 - #31 by philm


Switch to the base-devel meta package requires manual intervention

2023-02-12 - Robin Candau

On February 2nd, the base-devel package group has been replaced by a meta package of the same name.
If you installed the base-devel package group prior to this date, explicitly re-install it to get the new base-devel package installed on the system:

pacman -Syu base-devel

Arch Linux - News: Switch to the base-devel meta package requires manual intervention


Very slow shutdown when using KDE Plasma Wayland

The issue is sddm that can not be finished until the default timeout 90 sec.
The solution: Install sddm-git

The known bug reports:

PHP 8.2 update and introduction of legacy branch

2023-01-13 - Pierre Schmitz

The php packages have been updated to the latest version 8.2. In addition to this, a new set of php-legacy packages have been introduced. These will follow the oldest but still actively supported PHP branch. This enables users to install the latest version but also introduces the option to use third party applications that might rely on an older version. Both branches will be kept up to date following our rolling release model. php and php-legacy can be installed at the same time as the latter uses a -legacy suffix for its binaries and configuration files.

In addition to this, the php7 packages have been removed as they reached end of life. The imap extension will no longer be provided as it relies on the c-client library which has been abandoned for many years.

Arch Linux - News: PHP 8.2 update and introduction of legacy branch


Some games like Dota 2 are freezing

The issue is caused by update libx11 1.8.3-4
The solution: Upgrade to libx11 1.8.3-5

VLC crashes when playing a video file

It looks like there has been a regression introduced with the new version of libva (2.17.0), which is used by VLC media player. If you get the following error when executing vlc in a terminal:

libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null)

you may try to change the Output option in Preferences > Video to VDPAU output instead of Automatic (or whatever value it had been set to before).

Downgrading the libva package from 2.17.0-1 to 2.16.0-1 may also work, but it is a more short-term workaround that should not be kept for too long, as sooner or later, the older package won’t be compatible with the other packages. Consider to do the solution above first.


pacman-mirrors was temporarily renamed to manjaro-mirrors (the changes were reverted)

This was done to avoid confusion with an AUR / WSL package with the same name and avoid possible breakage. It should be a seamless transition.



KDE Wayland sometimes hangs with mesa 22.3.1 and AMD GPU


dbus-x11 demoted to AUR

dbus-x11 was installed out of the box on some flavors of Manjaro as of a year ago, but is no longer needed. If you have it installed, please replace it with dbus using the following:

sudo pacman -Syu dbuswiki post
XFCE becomes unusable if libxfce4ui-nocsd is still installed

So far XFCE Classic project hasn’t updated yet to 4.18 series. We pushed 4.17.0 version of the package, but don’t know if that helps in that situation. It is recommended to install the regular libxfce4ui package to avoid issues and comment on this issue: libxfce4ui 4.18.0 released · Issue #15 · Xfce-Classic/libxfce4ui-nocsd · GitHub


python-cairo 1.23.0-2 update requires manual intervention

The python-cairo package prior to version 1.23.0-2 was missing the compiled python module. This has been fixed in 1.23.0-2, so the upgrade will need to overwrite the untracked pyc file that was created. If you receive this error

python-cairo: /usr/lib/python3.10/site-packages/cairo/__pycache__/__init__.cpython-310.pyc exists in filesystem

when updating, use

pacman -Syu --overwrite /usr/lib/python3.10/site-packages/cairo/__pycache__/\*

to perform the upgrade.


Pop Shell issues with Xorg session with Mutter 43.1 (fixed with mutter 43.1-1.0)

Gnome 43 (X11) GTK4 windows failed to render · Issue #1541 · pop-os/shell · GitHub

Window actors call update_stage_views() without allocation after non-interactive resize on X11 (#6054) · Issues · GNOME / gnome-shell · GitLab


signature from "Manjaro Build Server <build_at_manjaro_dot_org>" is unknown trust

Seems our signing key for our Build Server had the 2022-10-28 as expire date set when generated. This is now fixed for chroot builds and fresh as also current existing installs via manjaro-keyring 20221028-4 Simply update your system to get that package, which got pushed to all our branches.


Kernel 6.1-rc# might break backlight control on old/weird laptops


Firefox 106.0 may have issues with screen sharing on GNOME with Wayland


Nvidia 3060RTX cards may have issues with the 515.76 driver (fixed with 520.56.06)
Removing python2 from the repositories

2022-09-23 - Jelle van der Waa

Python 2 went end of life January 2020. Since then we have been actively cutting down the number of projects depending on python2 in our repositories, and we have finally been able to drop it from our distribution. If you still have python2 installed on your system consider removing it and any python2 package.

If you still require the python2 package you can keep it around, but please be aware that there will be no security updates. If you need a patched package please consult the AUR, or use an unofficial user repository.*

Arch Linux - News: Removing python2 from the repositories

* Note: Unofficial user repositories are not supported

This update broke Firefox launching on Wayland (MOZ_ENABLE_WAYLAND=1) for me. Happens with both stable and dev edition.

I would like to point out that some things have been changed in yay v12.

The most noticeable for me:

  • new update menu style by default (I’ve used it before, and I like it)
  • you can’t clean orphans with yay -c (must do yay -Yc)
  • yay currently does not show out-of-date AUR packages (fixed in yay 12.0.4)
1 Like

I didn’t use that variable and my Firefox works fine on KDE Wayland since last two months (when I switched to Wayland permanently).

Does Features > Window Protocol on about:support list wayland or xwayland? If xwayland, maybe you could try quitting it and launching with MOZ_ENABLE_WAYLAND=1 to see if it crashes for you too?

I checked, and I don’t have such variable there at all. It shows:

Window protocol: **xwayland**.
Driver WebGL 2 WSI info: 
EGL_MESA_configless_context EGL_MESA_drm_image EGL_MESA_image_dma_buf_export EGL_MESA_query_driver EGL_NOK_texture_from_pixmap EGL_WL_bind_**wayland_display** 
EGL_EXT_platform_x11 EGL_EXT_platform_device EGL_MESA_platform_surfaceless EGL_EXT_explicit_device EGL_KHR_**platform_wayland** EGL_EXT_**platform_wayland** EGL_KHR_platform_gbm EGL_MESA_platform_gbm EGL_EXT_platform_xcb
IsWebglOutOfProcessEnabled: 1

Ah, I just realized, I don’t use stock firefox but firefox-appmenu-bin - this may do the whole difference. It’s a version with applied patches for global menus.

Anyway, I’m not an expert and since it works for me, I don’t want to mess it up.

Also, I don’t have any special Wayland related global environmental variables.

I use a laptop with Nvidia hybrid graphics and Wayland works pretty great.

Maybe you don’t need that variable? Did you add this manually? In the past we had to use several variables to enable Wayland and Nvidia related things like EGL strings, but it’s not needed anymore.

Moderator edit: In the future, please use proper formatting: [HowTo] Post command output and file content as formatted text

After updating to latest KDE 5.27.4, every GTK app I launch gives me extra warnings that were’nt there before:

(liferea:1348): Gtk-WARNING **: 14:11:45.318: Theme parsing error: gtk.css:1649:16: '-gtk-icon-size' is not a valid property name
(liferea:1348): Gtk-WARNING **: 14:11:45.318: Theme parsing error: gtk.css:1652:16: '-gtk-icon-size' is not a valid property name
(WebKitWebProcess:1372): Gtk-WARNING **: 14:11:46.381: Theme parsing error: gtk.css:1649:16: '-gtk-icon-size' is not a valid property name
(WebKitWebProcess:1372): Gtk-WARNING **: 14:11:46.381: Theme parsing error: gtk.css:1652:16: '-gtk-icon-size' is not a valid property name
(WebKitWebProcess:1372): Gtk-WARNING **: 14:11:46.772: Theme parsing error: gtk.css:1649:16: '-gtk-icon-size' is not a valid property name
(WebKitWebProcess:1372): Gtk-WARNING **: 14:11:46.772: Theme parsing error: gtk.css:1652:16: '-gtk-icon-size' is not a valid property name

Probably nothing to worry about, isn’t it ? :slight_smile: Restoring to KDE 5.27.3 fixes it…

And the buttons are too big (minimize, maximize, close).

Not a big deal, I found this bug report 468203 – GTK3 window buttons are too big and blurry on Wayland with Breeze GTK

Looks like it will be fixed in 5.27.5 (a month from now).

I believe Arch maintainers will cherry-pick a fix to breeze-gtk-5.27.4-2, so we don’t have to wait too long.

Edit: I temporarily downgraded to 5.27.3:

sudo pacman -U https://archive.archlinux.org/packages/b/breeze-gtk/breeze-gtk-5.27.3-1-any.pkg.tar.zst

Firefox does not break here when launching, I use the the same variable and about:support reports wayland (I use Gnome) Does starting firefox from the terminal give any leads as to why this breaks?

mesa 23.0.2-2 + egl-wayland 2:1.1.11-4 should fix the issue.



2 posts were split to a new topic: Update took two Timeshift snapshots

I’m guessing this is an upstream bug: after installing this update, pipewire-pulse goes into a loop, constantly crashing and restarting, if pipewire-zeroconf is installed. (I use it to output audio to a home server.) Haven’t had time to examine this very much, but this is what I’m getting:

> journalctl --user --unit pipewire-pulse.service
Apr 08 18:36:11 DrFunkenstein systemd[936]: Started PipeWire PulseAudio.
Apr 08 18:36:11 DrFunkenstein pipewire-pulse[3061]: 536870913
Apr 08 18:36:11 DrFunkenstein pipewire-pulse[3056]: mod.zeroconf-publish: error id:34 seq:66 res:-2 (No such file or directory): no global 56
Apr 08 18:36:11 DrFunkenstein pipewire-pulse[3056]: mod.zeroconf-publish: error id:0 seq:93 res:-2 (No such file or directory): unknown resource 34 o
Apr 08 18:36:11 DrFunkenstein pipewire-pulse[3063]: 536870914
Apr 08 18:36:11 DrFunkenstein pipewire-pulse[3059]: 536870915
Apr 08 18:36:41 DrFunkenstein pipewire-pulse[3056]: mod.pulse-tunnel: failed to connect: Timeout
Apr 08 18:36:41 DrFunkenstein pipewire-pulse[3056]: mod.zeroconf-discover: Can't load module: Connection timed out
Apr 08 18:36:46 DrFunkenstein pipewire-pulse[3056]: mod.pulse-tunnel: failed to connect: Connection refused
Apr 08 18:36:46 DrFunkenstein pipewire-pulse[3056]: mod.zeroconf-discover: Can't load module: Connection refused
Apr 08 18:36:47 DrFunkenstein systemd-coredump[3245]: [🡕] Process 3056 (pipewire-pulse) of user 1000 dumped core.
                                                      Stack trace of thread 3056:
                                                      #0  0x00007f99b9cf25f0 n/a (n/a + 0x0)
                                                      ELF object binary architecture: AMD x86-64
Apr 08 18:36:47 DrFunkenstein systemd[936]: pipewire-pulse.service: Main process exited, code=dumped, status=11/SEGV
Apr 08 18:36:47 DrFunkenstein systemd[936]: pipewire-pulse.service: Failed with result 'core-dump'.
Apr 08 18:36:47 DrFunkenstein systemd[936]: pipewire-pulse.service: Scheduled restart job, restart counter is at 5.
Apr 08 18:36:47 DrFunkenstein systemd[936]: Stopped PipeWire PulseAudio.
Apr 08 18:36:47 DrFunkenstein systemd[936]: Started PipeWire PulseAudio.

For now, my only solution has been to uninstall pipewire-zeroconf. Can anyone think of a workaround?

EDIT: Seems the solution was to use context.modules instead of context.exec to load modules in the configuration files in /etc/pipewire/pipewire-pulse.conf.d/. See https://docs.pipewire.org/page_pipewire_modules.html for examples.

The patch for breeze-gtk fixed the icons being too large, though they are still spaced as though they were the large size, leading to some weird gaps.

I believe the new version of yay has been producing some unexpected output lately. I just did a double-take thinking firefox was downgraded to the aur, lol :smiling_face_with_tear::

$ yay
:: Synchronizing package databases...
 core is up to date
 extra is up to date
 community is up to date
 multilib is up to date
:: Searching AUR for updates...
:: Searching databases for updates...
:: 2 Packages to upgrade.
2  extra/firefox             111.0.1-1 -> 112.0-1
1  extra/firefox-i18n-en-us  111.0.1-1 -> 112.0-1
==> Packages to exclude: (eg: "1 2 3", "1-3", "^4" or repo name)
 -> Excluding packages may cause partial upgrades and break systems
==> ^C


Hm, thanks for that but I still find the new default behavior from yay very odd. I can see why exclusions are presented for aur packages but to include those selections on normal repo packages even with dropping the nocombinedupgrade flag?..seems like that’s giving the non-dev user easier opportunities for a partially updated system. Maybe I’m missing reasonable thinking here and perhaps this is a “stepping stone” to transition to the combined updates behavior. @stasadev, any additional commentary regarding this?

Did you see the most recent addition to the news post I linked?

yay 12.0.4 reintroduces --nocombinedupgrade for yay -Syu , handling all repository updates before running yay’s upgrade and dependency gathering.

I did, I’m just trying to understand the logic of the new exclusions behavior. I like using yay for my aur helper and I’ll continue experimenting with different flags.