02:26 <RoundDuckKira> no j`ey

11:50 <u154ss> Still on my ailing M2 Mini again.. Did a fresh install of Debian (Glanzmann) with a minimal flavour - xinit, blackbox, lightdm and terminator - still the same Xorg error(s) as reported last week. Full logs - xorg - https://paste.debian.net/1304797/ dmesg - https://paste.debian.net/1304798/ journalctl - https://paste.debian.net/1304799/, https://paste.debian.net/1304800/ and https://paste.debian.net/1304801/

11:52 <u154ss> I know this is a bother, but I would appreciate any bright ideas.

11:58 <leio> feels like something in DRI2 stuff is messing up memory badly, hitting a crash and then backtrace is just gibberish and nonsense

11:59 <leio> did wayland work?

12:01 <u154ss> @leio as reported previously, on the M2 Mini, Fedora/KDE Plasma (Wayland) works. Fedora(XFCE) works. Debian more-or-less-minimal fails, as does Debian/XFCE - same errors.

12:07 <leio> I was asking about wayland on that debian install

12:13 <u154ss> @leio No, I do not even get that far.

12:15 <u154ss> @leio as my preferred DE is XFCE, wayland wil not be relevant is my case anyway.

12:21 <u154ss> wil not be relevant is my case anyway = will not be relevant in my case anyway.

12:52 <leio> it is relevant for figuring out what is wrong

12:53 <leio> as-is it seems like we don't even know if it's Xorg-specific or not

12:57 <j`ey> yeah trying wayland seems like a good idea

12:57 <j`ey> that's an older kernel, 6.5, maybe there's an issue with that and the mesa you have

13:30 <u154ss> Update - just tried Debian/Gnome (Wayland-ready) on the M2 Mini. Same results - backtrace with offset 0x1070000051e in Xorg.0.log. Bah!

13:32 <j`ey> is that still with lightdm? if that's gdm maybe gdm is still X11 based

13:34 <u154ss> @j'ey I installed Gnome using tasksel. The DM will be whatever Gnome uses for default. I do not think it is Lightdm. GDM most likely.

13:35 <j`ey> presumably xorg if xorg crahses

13:37 <u154ss> @j'ey, I will have to port the whole build chain from the M1 Mini to the M2 Mini and try and build a 6.6.x kernel. The stuff from Glanzmann is from November 2023.

13:38 <j`ey> dunno the easiest way to get a full wayland DE+login manager, maybe KDE

13:39 <u154ss> @j'ey Gnome is supposed to be "ready", I could try KDE though.. will take some time.

13:39 <j`ey> seems that GDM itself is using xorg, even if gnome is using wayland

13:40 <u154ss> @j'ey what a f**king mess.

13:40 <u154ss> will report back.

13:44 <leio> GDM uses wayland

13:44 <leio> unless it's a hybrid nvidia card or some such

13:44 <leio> in fact GDM can only start wayland sessions (including GNOME itself) when GDM itself is using wayland - if it uses Xorg, it only allows choosing xorg sessions

13:44 <leio> testing GDM and GNOME in Xorg doesn't really give any additional debugging information here

13:45 <leio> it falling back to Xorg instead of Wayland may be a symptom; journal might tell something.

13:45 <leio> i.e. wayland crashes too, fallbacks to Xorg, that crashes too, and you see just the Xorg crash but should be looking at the wayland using one too

13:46 <leio> (GDM uses gnome-shell to draw the UI; this shuts down on timeout when the active VT isn't on GDM to save resources, but only with wayland)

13:49 <u154ss> @all starting from scratch again, this time with Debian/KDE. I am not sure what version of KDE debian will pull from the repos though. Surely not as "polished" as Fedora though.

14:20 <u154ss> @all The Debian/KDE install did not work either. Strangely, Xorg.0.log has the same error text and offset value as my previous efforts. I took a photo from the recovery console - https://ibb.co/DGGZrfc

14:23 <leio> You shouldn't even have a Xorg.0.log for wayland testing though?

14:25 <leio> don't start a DE as root; if you want to start it from command line instead of something like sddm, you may need to take care of many things manually like I see a struggle in the photo

14:25 <u154ss> @leio would have thought so, yes.

14:26 <leio> something like `dbus-launch --exit-with-session startplasma-wayland` maybe, but I haven't used that ever

14:26 <leio> as NOT root.

14:30 <u154ss> @leio whether with Debian/XFCE or Debian/DKE, I do not even get as far enough to login with another userid.

14:31 <u154ss> DKE = KDE.

14:32 <j`ey> you can use the 'su' command to change user from the recovery shell to try that manuallly

14:32 <u154ss> @j'ey have tried already.

14:32 <leio> or disable the desktop manager service in the rescue shell and boot into proper system

14:32 <leio> didn't realize that's at play here

14:45 <u154ss> @leio from the rescue console, I can find no references to kde or plasma using systemctl and grepping.

14:46 <leio> sddm?

14:52 <u154ss> @leio no references to sddm using systemctl.

14:53 <u154ss> @leio there is also no dbus-launch binary on this system.

14:56 <u154ss> @leio https://packages.debian.org/sid/dbus-x11 is where this binary is found, but as I have to network in the recovery console, I am snookered here.

14:57 <leio> I don't think you see sddm with just `systemctl` as it's not started due to rescue session

14:57 <leio> u154ss: thus look at `systemctl list-unit-files` maybe

14:57 <leio> `dbus-run-session -- startplasma-wayland` then maybe; or maybe it's just not working because even system dbus session isn't working due to rescue shell

14:59 <leio> maybe `systemctl disable sddm` will tell you that it removes a symlink and then it shouldn't start after reboot to normal mode

15:01 <leio> (as in, maybe startplastma-wayland already would take care of the user dbus session that dbus-run-session/dbus-launch concern itself with, but the problem is elsewhere with dbus, e.g. lacking system bus if this is a /bin/bash rescue shell)

15:03 <j`ey> or sudo systemctl set-default multi-user.target, to boot to a tty

15:07 <u154ss> @all back in a few mins.

15:21 <u154ss> @j'ey systemctl set-default multi-user.target - am at a console now.

15:29 <u154ss> @all photo - https://ibb.co/12kXB8p something amiss.

18:51 <leio> don't use dbus-launch, as you found it's X11-specific; use dbus-run-session then, if it's even needed. Not saying that's your problem - doesn't look like it

18:52 <leio> might as well just try weston or something to start with

18:52 <leio> oh, they aren't here right now