#asahi-gpu · 2023-07-29 07:56
jannaumps claims the crashes are not reproducible with asahi-6.4-1 but asahi-6.4-2 with the same mesa version which seems implausible since that's just dart and dcp changes#asahi-gpu · 2023-07-29 07:38
jannauwhen you feel better. take care#asahi-gpu · 2023-07-29 07:37
jannaunot a good one, use firefox for hours/days. seemed to occur less often after updating asahi-20230725, asahi changes in asahi-6.4-3 might be related, see also discussion in #asahi-alt over the last copule of days#asahi-gpu · 2023-07-29 07:36
lina(Still feeling a bit sick so I'm going to lie down for a bit)#asahi-gpu · 2023-07-29 07:30
linaThat looks like a UAF...#asahi-gpu · 2023-07-29 07:30
linaDo you have a way to reproduce it?#asahi-gpu · 2023-07-29 07:21
jannauthe bo looks corrupt#asahi-gpu · 2023-07-29 07:17
jannaubacktrace https://gist.github.com/jannau/4ba4250fded46e41b6ebcc50fd6abedf#asahi-gpu · 2023-07-29 07:13
jannauapaprently an agx_bo with dev == NULL#asahi-gpu · 2023-07-29 07:12
linaI'm going to get lunch, but this is weird...#asahi-gpu · 2023-07-29 07:12
linaBut yes, reverting it fixes xwayland on mutter#asahi-gpu · 2023-07-29 07:11
linaNow I'm really confused...#asahi-gpu · 2023-07-29 07:11
linaWhich both has nothing to do with its own commit description (!?) and doesn't have anything to do with modifiers#asahi-gpu · 2023-07-29 07:11
linaI bisected the xwayland thing to 1d9d72b3fb905e#asahi-gpu · 2023-07-29 07:11
linaUmmmmmmmmmm#asahi-gpu · 2023-07-29 07:11
jannaufirefox crashes are agx related. finally one after enabling symbols:: SIGSEV in agx_bo_unreference() ../mesa-asahi-20230725/src/asahi/lib/agx_bo.c:190#asahi-gpu · 2023-07-29 07:00
linaYeah, it's a core Mesa regression. I have a bad and good commit now (with trivial Asahi changes on top), going to bisect#asahi-gpu · 2023-07-29 06:25
linaI think it might be a Mesa regression somewhere, it worked with some random older branch#asahi-gpu · 2023-07-29 06:25
jannauglxgears seems to kill xwayland here and gnome doesn't restart it#asahi-gpu · 2023-07-29 05:17
linaIt means somewhere along the line the modifier gets lost or set to zero#asahi-gpu · 2023-07-29 05:17
linaApps are rendering compressed and mutter is compositing them as linear#asahi-gpu · 2023-07-29 05:16
linaThat's the modifiers issue#asahi-gpu · 2023-07-29 05:02
jannauXwayland is even more broken for me under gnome. On login I get a broken KGpg window but after that some X11 applications are completly broken, for example glxgears#asahi-gpu · 2023-07-29 04:41
Retr0id(I had xorg-xwayland-23.1.1-1-aarch64.pkg.tar.xz in my local cache)#asahi-gpu · 2023-07-29 04:40
Retr0idjust downgraded to 23.1.1, no change in symptoms#asahi-gpu · 2023-07-29 04:38
jannaurelevant branch https://gitlab.freedesktop.org/xorg/xserver/-/commits/xwayland-23.1?ref_type=heads#asahi-gpu · 2023-07-29 04:37
jannauRetr0id: can you try to downgrade to Xwayland 23.1.1#asahi-gpu · 2023-07-29 04:36
jannauthere are related changes in Xwayland 23.1.2, for example https://gitlab.freedesktop.org/xorg/xserver/-/commit/73e0510dee175b5c834c1d4c2cb24071b7f84bbf#asahi-gpu · 2023-07-29 04:31
jannauXWayland has probably a lot less commits and there was iirc a release recently#asahi-gpu · 2023-07-29 04:30
jannauI would start with mutter (gnome's compositor) if you're reasonably sure that the gnome update broke it#asahi-gpu · 2023-07-29 04:24
Retr0id^ if I wanted to look through commits and speculate about what might've caused it, which repo(s) should I be looking in?#asahi-gpu · 2023-07-29 02:32
linaLooks like GNOME 44 / XWayland broke window modifiers or something like that...#asahi-gpu · 2023-07-27 12:00
alyssa(scratch buffer allocation)#asahi-gpu · 2023-07-27 12:00
alyssawhoops now i have a different new kernel thing needed .-.#asahi-gpu · 2023-07-27 11:59
alyssalina: i changed branches to work on something else to avoid kernel backup#asahi-gpu · 2023-07-26 14:09
alyssaalso after that, passing ./amd_vertex_shader_layer-layered-depth-texture-render which will be more kernel stuff (or ZLS control bits)#asahi-gpu · 2023-07-26 14:07
alyssaso I just need help getting it to pass with that commit reverted ^_^#asahi-gpu · 2023-07-26 14:07
alyssaI've added a patch which disables hw layered rendering and uses only the sw emulation path, and that passes the test#asahi-gpu · 2023-07-26 14:07
alyssalina: OK, it's almost definitely kernel#asahi-gpu · 2023-07-26 13:23
alyssapor que no los 2#asahi-gpu · 2023-07-26 13:23
alyssacould also be a mesa snafu#asahi-gpu · 2023-07-26 13:21
linaIt's probably a kernel snafu...#asahi-gpu · 2023-07-26 13:20
alyssaI hope you can figure out the layered rendering mystery because I'm p stuck :(#asahi-gpu · 2023-07-26 13:18
linaYeah, that was my guess too ^^#asahi-gpu · 2023-07-26 12:56
alyssamy guess is that's the stride of the second plane, with the appropriate shifting/biasing done to pack#asahi-gpu · 2023-07-26 12:56
alyssalina: ^^ (stride / 16) - 1#asahi-gpu · 2023-07-26 12:45
alyssagood luck :3#asahi-gpu · 2023-07-26 12:44
alyssaforcing pass type to TRANSLUCENT changes the failure slightly which is a smoking gun for that#asahi-gpu · 2023-07-26 12:44
alyssaor kernel bug with the layered bits#asahi-gpu · 2023-07-26 12:44
alyssawhich could mean either shader bug (possible but I've read the asm and it looks fine..)