08:53tiredchiku[d]: at some point, helldivers 2 got fixed
08:53tiredchiku[d]: on nvk
09:14tiredchiku[d]: quake champions too
09:14tiredchiku[d]: elite dangerous still crashes in the same spot though
10:02tiredchiku[d]: so, plasma doesn't launch when I set NOUVEAU_USE_ZINK=1, unless I also set MESA_VK_DEVICE_SELECT to my gpu's pci id
10:03tiredchiku[d]: gnome, sway, and i3wm all launch fine without the Mesa env var
10:03tiredchiku[d]: would this be more of a plasma bug or more a mesa/zink bug
10:03tiredchiku[d]: just confused as to where to file the issue
10:07karolherbst[d]: could be a GPU selection issue
10:07karolherbst[d]: do you have multiple GPUs?
10:07tiredchiku[d]: nope
10:07tiredchiku[d]: just the 3070
10:07tiredchiku[d]: not even building swrast
10:10karolherbst[d]: weird...
10:10karolherbst[d]: did you check logs?
10:11tiredchiku[d]: journalctl doesn't have anything significant
10:11tiredchiku[d]: neither does dmesg
11:11marysaka[d]: how do you start it tiredchiku[d]
11:14tiredchiku[d]: tried both, via sddm and `startplasma-wayland` in a tty
11:14tiredchiku[d]: also `startx /usr/bin/startplasma-x11` fwiw
11:17marysaka[d]: tiredchiku[d]: hmm I don't have issues starting it on tty with `NOUVEAU_USE_ZINK=1` and doing:
11:17marysaka[d]: XDG_SESSION_TYPE=wayland dbus-run-session startplasma-wayland
11:18tiredchiku[d]: will try that the next time I boot into that installation
11:24tiredchiku[d]: marysaka[d]: still a nope
11:26tiredchiku[d]: also leads to a full system render freeze btw
11:26tiredchiku[d]: can't even swap to a different tty
11:30marysaka[d]: That's quite weird..
11:35tiredchiku[d]: kernel 6.11-rc5
11:35tiredchiku[d]: mesa-git
11:35tiredchiku[d]: exact commit 1537b9355aa
11:36tiredchiku[d]: a couple days old
11:37tiredchiku[d]: if you're on stable/tagged release, you probably won't run into it because this commit caused it https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30647/diffs?commit_id=8bfbcd5b28d3b4c764ba29000945647eb3fdc2c3
11:50marysaka[d]: Okay yeah I can reproduce on main
11:51tiredchiku[d]: mhm
11:51tiredchiku[d]: been meaning to file an issue for it since friday but
11:51tiredchiku[d]: I'm not sure where it should go
11:53marysaka[d]: I think zink but also add NVK tag to it
12:03tiredchiku[d]: https://gitlab.freedesktop.org/mesa/mesa/-/issues/11859 don't think I have the permissions to add tags
12:04marysaka[d]: tiredchiku[d]: Are you building with vk layer? Looking at the MR, mike patch requires the changes on the layer
12:04tiredchiku[d]: I am
12:06marysaka[d]: tiredchiku[d]: are you certain the loader used is the one you are building
12:08tiredchiku[d]: yes
12:08tiredchiku[d]: there's no other on my system
12:12marysaka[d]: tiredchiku[d]: For now as a workaround also set NODEVICE_SELECT=1 in your env
12:12marysaka[d]: it works here but on my side it was picking the system one
12:13tiredchiku[d]: I mean, as a workaround I've set MESA_VK_DEVICE_SELECT=10de:2484, which also works
12:14marysaka[d]: That's true... NODEVICE_SELECT seems to actually make it loads forever?
12:14tiredchiku[d]: ```[sidpr@infiltrator /]$ sudo find . -type f -name "libVkLayer_MESA_device_select.so"
12:14tiredchiku[d]: ./home/sidpr/documents/pkgs/mesa-git/src/_build64/src/vulkan/device-select-layer/libVkLayer_MESA_device_select.so
12:14tiredchiku[d]: ./home/sidpr/documents/pkgs/mesa-git/src/_build32/src/vulkan/device-select-layer/libVkLayer_MESA_device_select.so
12:14tiredchiku[d]: ./home/sidpr/documents/pkgs/mesa-git/pkg/mesa-tkg-git/usr/lib/libVkLayer_MESA_device_select.so
12:14tiredchiku[d]: ./home/sidpr/documents/pkgs/mesa-git/pkg/lib32-mesa-tkg-git/usr/lib32/libVkLayer_MESA_device_select.so
12:14tiredchiku[d]: find: ‘./run/user/1000/doc’: Permission denied
12:14tiredchiku[d]: find: ‘./run/user/1000/gvfs’: Permission denied
12:14tiredchiku[d]: ./usr/lib/libVkLayer_MESA_device_select.so
12:14tiredchiku[d]: ./usr/lib32/libVkLayer_MESA_device_select.so
12:15tiredchiku[d]: 4 of those find results are just from my build env (makepkg)
14:38valentineburley[d]: konstantinseurer[d]: gfxstrand[d] Would you be ok with doing https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30821/diffs?commit_id=77f1a2657b50576fb4a9a5c4c71caf793778322f for LVP too?
16:05gfxstrand[d]: valentineburley[d]: Yeah, that's fine.
16:09zmike[d]: I think konstantinseurer[d] is handling lavapipe
16:14valentineburley[d]: That was his suggestion yesterday, but I'll leave it up to him then
16:27zmike[d]: oh then do whatever he says
16:48lyudess: airlied: jfyi been seeing this happen every now and then when my nvidia card resumes, any way I can get more info on what might be happening (or whether this is a known issue?) https://paste.centos.org/view/d1be4852
16:51magic_rb[d]: Have to you tried doing a full poweroff? Ive found the gpu can get into an invalid state and a reboot doesnt cut it
17:13Lyude: magic_rb[d]: I mean it's not about getting the GPU working again, that's easy enough :P. I'm trying to look into this more so that we can fix issues like this, since if we're getting exceptions like that randomly that means there's a bug somewhere to be fixed
17:20magic_rb[d]: Oh yeah i can send you a buttload of exceptions :)
17:21magic_rb[d]: But ive no clue how to fix any
18:19gfxstrand[d]: No, it's not
18:51airlied: Lyude: that should be fixed by 9c685f61722d30a22d55bb8a48f7a48bb2e19bcc nouveau: set placement to original placement on uvmm validate.
19:12Lyude: airlied: cool ^^
20:03samantas5855[d]: pavlo_kozlenko[d]: we got fermi nvk before gta6?
20:03samantas5855[d]: nevermind this is a kepler for some weird reason
20:25tiredchiku[d]: do you have the lib installed
20:27tiredchiku[d]: libdrm
20:27tiredchiku[d]: hm
20:27tiredchiku[d]: just a sec
20:28tiredchiku[d]: `gcc -I /usr/include/libdrm 1.c`
21:13asdqueerfromeu[d]: pavlo_kozlenko[d]: You don't need xf86drm I think
23:29gfxstrand[d]: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31105