08:27 pabs3: got another nouveau_pushbuf_data crash in Xwayland
08:54 pabs3: seems to happen immediately after gnome-shell screen unlock...
09:50 karolherbst: pabs3: do you have a stacktrace of that?
10:06 pabs3: karolherbst: systemd journal (inc gnome-shell/kernel nouveau messages) and gdb backtrace in https://paste.debian.net/hidden/2b76f52f/
10:06 karolherbst: pabs3: ohh...
10:06 karolherbst: pabs3: nah.. that just means the gpu context crashed
10:07 karolherbst: I'll have an idea
10:07 pabs3: confirmed via journalctl that it happened just after gnome-shell screen unlock
10:07 pabs3: Jun 26 16:11:10 gdm-password][880311]: gkr-pam: unlocked login keyring
10:08 karolherbst: pabs3: what gpu is that on? and do you mind compiling nouveau or the kernel yourself?
10:09 pabs3: hmm, it doesn't happen on every unlock though
10:09 pabs3: 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK107 [GeForce GT 740] [10de:0fc8] (rev a1)
10:09 pabs3: its old hardware so compiles are slow, but I don't mind trying patches
10:10 karolherbst: pabs3: patches to help debug the issue :p
10:10 karolherbst: if glamor crashes the GPU context we should figure out why
10:11 karolherbst: and we can do something to help out with that
10:11 pabs3: no problem with testing such patches, I'll be glad to have the crashes go away :)
10:12 pabs3: I'll likely be applying them to the version that is in Debian though, so I can simply recompile/reinstall and overwrite Debian's packages
10:12 pabs3: Debian bullseye I mean
10:13 pabs3: mesa 20.1.1-1/20.1.2-1 and linux 5.7.6-1
10:13 karolherbst: yeah.. it's fine
10:13 karolherbst: will link the patch shortly
10:13 pabs3: thanks
10:14 karolherbst: pabs3: https://github.com/karolherbst/linux/commit/33ab96bbe75a85c7fbf1de5b40f20be6e7766ba5.patch
10:15 pabs3: hmm, just forcibly turning on sync?
10:15 karolherbst: yeah
10:15 karolherbst: well, we don't know what's wrong
10:16 karolherbst: and we should look into the submission which caused the gpu to crash
10:16 karolherbst: the one in the log is one the GPU didn't even worked on
10:18 pabs3: the log is from linux 5.6.14-2 btw.
10:18 pabs3: hmm, ok. will build linux 5.7.6-1 with the change and reboot once done
10:20 karolherbst: pabs3: okay, and once you have a nother crash I'd like to have the stacktrace again
10:22 pabs3: ok, will do. probably will be some days before it happens
10:23 karolherbst: yeah.. no worries
13:47 VanackSabbadium: Ok, it's enough for me. My PC froze 4 times in 10 minutes, I'm sick and tired. I don't think this is gonna be solved shortly, so until that or until I have enough money to buy an amd gpu, I'm going with proprietary drivers. Sorry for that I still appreciate your support and work.
13:52 VanackSabbadium: However, this is my modinfo nouveau https://cloud.disroot.org/s/6rYApZxogY9sBrD
14:04 RSpliet: VanackSabbadium: Sorry mate. Thanks for giving it a serious try.
14:06 linkmauve: I still have to sell this GTX 1070 and this broken RX580, I’m sure I could buy a good AMD one with the money.
14:07 VanackSabbadium: Sorry RSpliet i swear I tried with all my strength
14:08 VanackSabbadium: linkmauve: go for a Vega
14:18 linkmauve: VanackSabbadium, why? Aside from being broken, the RX580 seems to be very good for my usecases.
14:18 linkmauve: These are my first non-Intel GPUs in decades, and I couldn’t use either of them. :(
14:18 VanackSabbadium: I heard great thing
14:19 VanackSabbadium: Sorry. I heard great things about Vega64
14:19 linkmauve: I heard power consumption though.
14:19 linkmauve: But since I’ll only use the GPU when playing games, it can sleep the rest of the time.
20:52 Lyude: Vblank workers have been r-b'd! imirkin, karolherbst :) will send out a respin with skeggsb's nits addressed today and then it should be good to go for merging
22:07 imirkin_: Lyude: cool
22:07 imirkin_: Lyude: btw, dunno if you saw, but that person with DP AUX timeouts said the increased timeout didn't help
22:08 Lyude: imirkin_: yeah I saw-i'll try to get back to them soon, been trying to fix a lot of laptop backlights in i915 (plus I think we're still waiting on hearing back from nvidia on what the actual timeout there is supposed to be...)