19:12 <marex> austriancoder: hey, have you ever seen flickering artifacts on WL desktop shell with etnaviv ?

19:15 <marex> mntmn: ^

19:27 <marex> austriancoder: I sent you an email with the photo of the corruption, it happens after many hours of running the machine and then the Weston desktop shell has this reverse L-shaped black thing flickering every second frame

19:28 <marex> oh, and the "notch" (the ____|-----) is not always the same, sometimes it is lower, sometimes higher

19:28 <marex> it is not always in the middle of the screen

19:54 <mntmn> marex: what is wl desktop shell? but i've seen flickering artifacts often with etnaviv

19:56 <mntmn> i have nothing reproducible though

20:01 <marex> mntmn: oh ?

20:01 <marex> mntmn: weston desktop shell

20:02 <marex> mntmn: do you have any pictures of it ?

20:50 <mntmn> marex: i only know the normal weston, is that the same thing?

20:50 <mntmn> marex: i don't have pictures

20:50 <mntmn> marex: but i have found on sway and especially gnome with some more extreme usage (i think last time with chromium-ozone) i would run into situations where it would get into a flickery mess state

20:51 <mntmn> marex: also reproducible with plugging secondary display device out and in on sway i think

20:51 <mntmn> marex: i meant chromium --enable-features=UseOzonePlatform --ozone-platform=wayland

20:54 <mntmn> but otoh my experience has been extremely stable in the last weeks, i'm using this imx8mq based laptop as my primary laptop all day without trouble

21:27 <marex> mntmn: yes, the "normal weston" is the desktop-shell

21:28 <marex> mntmn: what does "flickery mess" look like ?

21:30 <marex> see email

21:32 <marex> its as if one of the weston desktop buffers got corrupted

21:35 <mntmn> seen

21:36 <mntmn> hmm hard to say when it starts for me it can be all kinds of missing/non-updating rectangles

21:36 <marex> mntmn: is that the failure you saw ?

21:36 <mntmn> it doesn't look the same every time

21:36 <mntmn> but it's always a double buffering kind of flicker, like one buffer goes out of sync or is not updated anymore

21:38 <marex> mntmn: thats what it looks like in here too

21:42 <mntmn> when i encounter it again i'll record it

21:42 <marex> mntmn: well there is something obviously broken

21:42 <marex> mntmn: and I suspect it might even trigger further corruption elsewhere

21:43 <marex> which shouldn't happen on MMU-capable vivante GPUs, so it shouldnt happen even on your GPU

21:43 <marex> on mine it might, the GCnano has MMU off

21:43 <marex> hmmmm

21:44 <mntmn> i am not using latest mesa because it wasn't stable for me, but i did so much fiddling in the last 2 weeks that i can't remember what exactly was wrong

21:44 <marex> Im on 20.3.4

21:44 <mntmn> oh i'm on 21.1.0

21:44 <marex> Im afraid this is the context sharing issue all over again