Control: severity -1 important Control: tags -1 + moreinfo Control: user pkg-gnome-maintain...@lists.alioth.debian.org Control: usertags -1 + wayland
On Tue, 08 Oct 2019 at 13:59:12 -0400, Zack Weinberg wrote: > On my computer, after upgrading to gnome-shell 3.34, I consistently get a > gnome-shell catastrophic failure about 10 seconds after logging in, but only > when running under Wayland. Does this still happen in newer versions, such as 3.34.4 (in testing) or 3.36.1 (in unstable)? Do you mean that you literally log in, wait for 10 seconds without doing anything, and the shell crashes; or is there something else you do that is or might be the trigger? Do you have any GNOME Shell extensions enabled? If yes, which ones, and does this still happen if you disable them? (You can disable them using gnome-tweaks, or using the new gnome-shell-extension-prefs for 3.36.x.) What GPU(s) do you have? If you have an NVIDIA device, are you using the proprietary nvidia-graphics-drivers or the open source Mesa/Nouveau drivers? It seems there's a workaround (X11 mode), and this doesn't seem to be happening to many people, so I'm reducing the severity. > This error appears in the logs: > > gnome-shell[2608]: JS ERROR: TypeError: actor.get_meta_window(...) is null > destroyWindowDone@resource:///org/gnome/shell/ui/windowManager.js:1773:26 > onStopped@resource:///org/gnome/shell/ui/windowManager.js:1742:34 > makeEaseCallback/<@resource:///org/gnome/shell/ui/environment.js:73:13 > easeActor/<@resource:///org/gnome/shell/ui/environment.js:126:60 > gnome-shell[2608]: JS ERROR: TypeError: actor.get_meta_window(...) is null > destroyWindowDone@resource:///org/gnome/shell/ui/windowManager.js:1773:26 > onStopped@resource:///org/gnome/shell/ui/windowManager.js:1742:34 > makeEaseCallback/<@resource:///org/gnome/shell/ui/environment.js:73:13 > easeActor/<@resource:///org/gnome/shell/ui/environment.js:126:60 > gdm-wayland-session[2571]: Received signal:15->'Terminated' > > I don't see anything else that is obviously related, but I could have > missed something. If the Shell is crashing, there should be more than this. If this is still reproducible in newer versions, please attach the systemd journal from the time you log in to a few seconds after the Shell crashes. If you can get a backtrace by using systemd-coredump (see https://wiki.debian.org/HowToGetABacktrace) then that would also give us a lot more information to work with. Thanks, smcv
