04:07 <lina> alyssa: Yes, it ignores writes

04:07 <lina> And yeah, I was thinking of just adding it to the device params so mesa can optimize based on it

04:09 <lina> jannau: I haven't stress tested 13.2 so it's totally possible that is broken in new and exciting ways ^^

04:10 <lina> Plasma going back usually means kwin_wayland crashed and restarted, the recovery from that is... not great

04:10 <lina> You'd have to get some stderr output from that to debug it

04:10 <lina> *black

04:52 <alyssa> lina: awesome :)

04:52 <alyssa> btw what's your plan for the next [short term]?

04:55 <lina> M2 Pro/Max support and keep the review emails going ^^

04:55 <alyssa> ah

04:55 <alyssa> no mesa review?

04:55 <lina> That too!

04:55 <alyssa> :-D

04:55 <lina> Mesa sends me emails ^^

04:56 <alyssa> ahaha

04:56 <alyssa> fair

04:57 <lina> Do you think it's fair to release this at this point even if there are some weirdo issues we can't easily reproduce? It's edge, so... at this point I think it's fair to let users figure out the consistent repros, I'm sure they can test a lot crazier stuff than we can ^^

04:57 <lina> There's probably some WSI thing missing still but... that's really hard to catch without a consistent repro...

04:59 <lina> (And at that point it could be app bugs even)

05:02 <alyssa> idk

05:02 <alyssa> and don't particularly have the bandwidth to burn on caring right now

05:03 <alyssa> so far it sounds like M1 + 12.3 is rock solid at any rate

05:03 <alyssa> M2 and 13.x is not (yet) my problem :-D

05:03 <lina> jannau mentioned pink squares on startup which I think I've seen too, and also those kwin crashes are undebugged...

05:03 <lina> I'm not worried about 13.2, that's for later

05:59 <jannau> yes, no need to worry about 13.2 for now. users are not able to run that yet and dcp on 13.2 needs testing as well

06:01 <jannau> I see the pinkl boxes only on m2 with the same kernel and mesa

06:02 <jannau> I'll look into the kwin crashes later

08:44 <jannau> lina: the black plasma/wayland desktop is not a kwin crash. plasmashell itself crashes

08:50 <jannau> looks like a race during startup, not reproducible with executing plasmashell manualy in the same session

08:50 <jannau> I'd say probably not gpu related

09:08 <jannau> it doesn't crash but terminates itself after "The Wayland connection experienced a fatal error: Bad file descriptor". I don't see how that can be directly caused by the gpu driver

09:39 <lina> I've seen plasmashell crashes related to GPU driver issues, but it could be unrelated I guess...

09:47 <jannau> do you see how EBADF on the wayland socket could be gpu related while kwin survives?

09:50 <jannau> I've just reproduced the issue with simpledrm

10:02 <lina> jannau: simpledrm+llvmpipe?

10:02 <lina> wouldn't you get EBADF if kwin crashes and restarts?

10:04 <jannau> yes, but kwin doesn't crash. I see no indication of it crashing and it still has the same PID as it was started with

16:41 <alyssa> lina: I rebased https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21674