cairo-dock-team team mailing list archive

[Mamoru TASAKA <1943052@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Downstream bug: https://bugzilla.redhat.com/show_bug.cgi?id=2000812

On Fedora 34 KDE Plasma wayland session, using cairo-dock
https://github.com/Cairo-Dock/cairo-dock-
core/commit/6c569e67a2a366e7634224a0133ede51755629cb ,

`$ cairo-dock -c`

causes segfault at starup as atatched. This does not occur with GNOME +
wayland - so apparently this is related to wayland implementation on KDE
plasma.

So what is happening here is that cairo-dock calls
wl_output_add_listener(), expecting that "geometry" information is to be
returned first before "mode" information is gained. GNOME + wayland
returns the information in this order, but KDE plasma + wayland returns
in the opposite order.

Currently I don't know this behavior is allowed or not. But perhaps cairo-dock side can workaround this issue by allocating g_desktopGeometry.pScreens on _output_mode_cb() when the pointer is still NULL, perhaps? (at least this workaround seems to work for me).
 

You can try Fedora 34 KDE Plasma session from
https://spins.fedoraproject.org/kde/download/index.html

** Affects: cairo-dock-core
     Importance: Undecided
         Status: New

** Attachment added: "gdb log for segfault"
   https://bugs.launchpad.net/bugs/1943052/+attachment/5523803/+files/gdb-cairo-dock-nocolor.log.txt

-- 
You received this bug notification because you are a member of Cairo-
Dock Devs, which is subscribed to Cairo-Dock Core.
https://bugs.launchpad.net/bugs/1943052

Title:
  segfault on KDE Plasma wayland

Status in Cairo-Dock Core:
  New

Bug description:
  Downstream bug: https://bugzilla.redhat.com/show_bug.cgi?id=2000812

  On Fedora 34 KDE Plasma wayland session, using cairo-dock
  https://github.com/Cairo-Dock/cairo-dock-
  core/commit/6c569e67a2a366e7634224a0133ede51755629cb ,

  `$ cairo-dock -c`

  causes segfault at starup as atatched. This does not occur with GNOME
  + wayland - so apparently this is related to wayland implementation on
  KDE plasma.

  So what is happening here is that cairo-dock calls
  wl_output_add_listener(), expecting that "geometry" information is to
  be returned first before "mode" information is gained. GNOME + wayland
  returns the information in this order, but KDE plasma + wayland
  returns in the opposite order.

  Currently I don't know this behavior is allowed or not. But perhaps cairo-dock side can workaround this issue by allocating g_desktopGeometry.pScreens on _output_mode_cb() when the pointer is still NULL, perhaps? (at least this workaround seems to work for me).
   

  You can try Fedora 34 KDE Plasma session from
  https://spins.fedoraproject.org/kde/download/index.html

To manage notifications about this bug go to:
https://bugs.launchpad.net/cairo-dock-core/+bug/1943052/+subscriptions