06:45 <jhovold> robclark, HdkR: ok, thanks.

06:47 <jhovold> robclark: could you send a reply about this to bamse's a690 v3 series, so he can include that in a v4?

06:47 <jhovold> still hoping we can get this in merged for 6.5...

14:25 <robclark> hmm, gnome-shell crashed when I plugged in external display (both otherwise things worked)

14:27 <travmurav[m]> robclark: remember the planes pinning hack? I think mutter still doesn't like non-pinned planes

14:27 <robclark> oh.. sad

14:28 <travmurav[m]> there seems to be some movement in https://gitlab.gnome.org/GNOME/mutter/-/issues/2398

14:29 <robclark> yeah, that backtrace looks the same

14:44 <steev> yeah, i dropped the kernel hack that i was carrying in favor of the mutter patch from danielt

14:44 <steev> the other workaround is using xorg for gnome session

14:46 <robclark> hmm, ok.. I guess I can just carry the kernel hack for now.. apparently danielt's mutter patch isn't in fedora

14:51 <danielt> At the moment I'm not sure my patch is anywhere but the bug tracker. I think everyone (including me) accepts my patch is a band-aid more than a proper fix. However I did my best to make it a band-aid that shouldn't cause regressions (which doesn't mean I achieved it).

14:58 <robclark> I'm semi-tempted to write up some patch to conditionally restrict what kernel exposes based on process name

14:59 <robclark> it would be better if we somehow had some version info

14:59 <robclark> but I guess we could add a cap that userspace needs to use to opt out of the "training wheels"

15:00 <steev> it's only mutter

15:00 <robclark> and chrome ;-)

15:01 <steev> er, yeah, chrome... https://bugs.chromium.org/p/chromium/issues/detail?id=1354398 i still have this tab open :(

15:03 <steev> i *think* chromium works on gnome with the mutter patch, at least 113.0.5672.126 is working here for me

15:04 <steev> https://usercontent.irccloud-cdn.com/file/QjybbpcM/Screenshot%20from%202023-06-05%2010-04-04.png

15:34 <clover[m]> chrome based browsers work fine for me with mutter patch

15:40 <robclark> steev: I meant chrome compositor, ie. what CrOS uses

15:40 <steev> oh

15:41 <robclark> basically msm tends to be compositors first experience w/ hardware that can move planes between displays.. it isn't _un_common in mobile thing but not on desktop hw

15:42 <robclark> so I wouldn't completely be surprised if kwin or others had similar issue

15:43 <steev> my understanding is... kwin has been a mobile thing for a while (not trying to disparage gnome devs here but i have been begging for better arm support since 2012...) so they might actually do it right

15:44 <steev> back in the days of gnome's gl check being able to be hacked with a simple symlink to true