Archlinux user. Updating from 0.8.0 to 0.9.0 introduces heavy lags on two of my machines (both intel) running KDE. Moreover a software that depends on seatd stops working unless I start it from my terminal: rog-control-center and cosmic-greeter falls in a restart loop.
Even if you used seatd, its only job is to open a handful of device files. It does nothing after your display server is started, and has no involvement in using them.
You likely use logind, in which case what you’re using from here is just libseat - that is, we’re not even responsible for opening the files, just for asking to have them opened.
In other words, I can’t see how it could possibly induce “lag”, nor does it have any effect on stuff like “rog-control-center” that cannot use libseat.
For cosmic greeter, if it’s started on a background VT. libseat no longer auto-switches the VT, so either use the current VT or insert a chvt when starting.
thanks for answering. I can't really say much except that going back with archlinux to 30 October and updating each package one by one except seatd have my system working normally, but as soon as I update seatd to 0.9.0 all of that minifests.
rog-control-center is flagged as a package that requires seatd by pacman.
When I say "seatd" I mean the package and libseatd is part of that package.
Anyway I am using sddm mainly and cosmic-greeter was used just to do a test on packages depending on seatd as suggested by the package manager.
The roc thing sounds like an accidental dependency as it wouldn’t be able to use the library. I’ll check that later.
What display server are you running? Try to reproduce with nothing running (including greeters/greetd stopped), preferably with the session started directly by logging into an empty VT and just running your display server from the shell there. Then check things like htop for CPU.
busctl --user monitorandsudo busctl monitorwill tell you if something is happening on dbus, which is how we talk to logind.You could also do a quick test with e.g. sway to see if it’s display server specific.
As mentioned, libseat doesn’t do anything when the system is up and running other than give a heads up if it has to pause because you’re switching VT, but maybe something is getting confused by a change in behavior and burning CPU or hammering the API or something.
The roc thing sounds like an accidental dependency
And indeed it was: https://gitlab.com/asus-linux/asusctl/-/merge_requests/204
Well both laptops were asus... Thank you very much for finding that!