13:24 <Guest49> ifreund: would it be helpful in the current state of rwm to test it out using rijan or some other wm?

14:55 <leon-p> testing is always helpful, although I suppose trying to write a non-trivial WM is more useful as it helps discovering design bugs which are arguably worse than code bugs

15:20 <jpnt> hi everyone, im using a hybrid gpu laptop (amdgpu and nvidia), recently I noticed that helvum and pwvucontrol (both GTK, made in Rust) crash the compositor and send me back to the tty. Im on the latest river dev version. Running 'fuser -v /dev/nvidia*' I can see that river is using my dedicated card, which is not optimal. I dont know if someone can help me or give me any tips to solve this

15:21 <jpnt> also, does river support GPU resets without crashing the compositor?

15:22 <leon-p> river logs and a trace would be helpful. if you are on a distribution with systemd, you can get a trace with `coredumpctl gdb` and then in the gdb repl call `bt`

15:22 <jpnt> the crash is some assertion error, Im guessing its because of gpu switching or some weird stuff like that

15:22 <jpnt> oof im running runit, will see what i can get

15:23 <leon-p> if this error is reproducable, can you maybe do so with a debug build of river and `-log-level debug`?

15:41 <ifreund> jpnt: river has some code that attempts to recover from GPU resets

15:42 <ifreund> it's not very well tested though since on my desktop with an rx 580 the kernel fails to recover

20:57 <jpnt> hey so i got a log (log-level debug). i opened helvum and it blows up. same thing for pwvucontrol, another gtk rust app... very weird stuff https://pastebin.com/etiKPasE

21:05 <jpnt> probably its related to this: https://github.com/swaywm/sway/issues/8194