GNOME Bugzilla – Bug 756994
GDM 3.18 login screen cursor much too large on Wayland with hidpi monitor
Last modified: 2018-05-24 11:18:53 UTC
Hi folks, I'm on Arch Linux, running Gnome 3 with GDM. Haven't had to touch any settings until today when I rebooted for the first time in several weeks and GDM popped up with a double- or possibly even triple-sized cursor, the usual grey background, no login box, and didn't respond to any input at all (e.g., couldn't move the mouse). I chrooted from a live USB and tried reverting a few gdm and wayland and gnome-session packages but nothing seemed to do the trick other than uncommenting WaylandEnable=false in /etc/gdm/custom.conf. On checking Arch sources, it doesn't seem like the Wayland packages themselves have been updated in a long time, so this would seemingly be down to one of the 3.18 changes, but I can't isolate it. Technically this would be considered a HiDPI machine (it's a Dell Venue 11 Pro, 1080p, something like 200 DPI), and I've made some tweaks to default font sizes and scaling factors via the Gnome tweak tool, but I'm not using the integer HiDPI multiplier setting (under "Windows" in the tweak tool) because I found it to be overkill. No idea if gdm reads these settings or not, just mentioning it in case it's helpful. I have my kernel version locked to 3.18 due to https://bugzilla.kernel.org/show_bug.cgi?id=94281 but I doubt that's relevant either. Is there any other information I can provide that would be helpful? Basically, it seems like now that Wayland is *trying* to be DPI-aware, something is breaking.
Bizarrely, if I check systemctl, it seems like geoclue (of all things) is reporting as having dumped core prior to all of the hard restarts I had to do after gdm ending up in this state. This is a long shot, but has anything changed in terms of how gdm calls networkmanager or modemmanager when using Wayland rather than Xorg? This machine has both wifi and LTE; the latter comes up under USB and isn't currently used (I don't have a SIM card in here) but Gnome does detect it properly under normal circumstances, and the wifi is probably connecting in the background while the login screen is displayed.
It seems like a duplicate of bug 754814, but I have to admit that my geoclue also crashes frequently. Here is what I found by coredumpctl: Sat 2015-10-24 10:42:21 CST 10907 0 0 6 * /usr/lib/geoclue2/geoclue Sat 2015-10-24 10:45:27 CST 12161 0 0 6 * /usr/lib/geoclue2/geoclue Sat 2015-10-24 10:46:07 CST 12677 0 0 6 * /usr/lib/geoclue2/geoclue Sat 2015-10-24 10:46:13 CST 12817 0 0 6 * /usr/lib/geoclue2/geoclue coredumpctl gdb 10907(it's all the same when debugging other 3 dumps) Core was generated by `/usr/lib/geoclue2/geoclue -t 5'. Program terminated with signal SIGABRT, Aborted.
+ Trace 235617
I'm seeing similar geoclue crashes. Even tried disabling the service but that does nothing ( I don't need location to be known ). I have them consistently on each boot but it doesn't seem to stop anything. I do get an occasional random no-errors stall on gdm tho. It just gets stuck in text mode on boot with no info and journalctl -b shows no error. NOt sure if that's related to this tho.
OK folks, please report the geoclue crashes to the geoclue developers on bugzilla.freedesktop.org. It's super unlikely that would be related to the cursor size problem.
-- GitLab Migration Automatic Message -- This bug has been migrated to GNOME's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.gnome.org/GNOME/gdm/issues/240.