09:54 <lynxeye> austriancoder: Have the expected fails in the etnaviv CI been updated after the switch to the NIR compiler and exposing ARB shadow? I've tried running https://gitlab.freedesktop.org/mesa/mesa/-/pipelines/555412 to get a bit more confidence for the gc7000r6204 changes, but a lot of the fails don't seem to be related to my changes. Checking a few of them, they are also present without my changes applied.

11:29 <marex> lynxeye: I also noticed that NIR doesn't seem to work well on 6204

11:30 <marex> falling back to TGSI does at least permit starting e.g. weston

11:58 <lynxeye> marex: See the updated MR, 6204 has the one const limitation again. This broke weston and a few glmark scenes and is fixed now.

11:59 <marex> oh, nice

12:04 <lynxeye> marex: GC7000 is also not covered by the CI pipeline currently. I just wanted to make sure I didn't break the covered GC2000, but that doesn't work well when there are lots of preexisting fails that aren't marked as expected.

12:05 <marex> lynxeye: I am mostly ignoring the gc7000 for now, I'm still trying to get to the "ultra-super complex locking" at least

12:11 <lynxeye> marex: Yea, some testing for the locking context multithread stuff would be nice. Then I kind of hoped to be able to land the GC7000 stuff before the branch point. Might not work out, as I was only able to figure out all required bits last week.

12:13 <marex> lynxeye: busy ? :)

12:19 <lynxeye> marex: Yep, jumping between all the unfinished i.MX8M* stuff. Only good thing about it is that there is sufficient time for other people to provide feedback before I manage to cycle through all the WIP items once. :)

12:40 <marex> lynxeye: I am still about to provide disagreement/feedback on the lcdif stuff, but ... I need to think about that a bit more

12:40 <marex> lynxeye: can you check the two patches I asked about in dri-devel ? You think now is about good time to apply them to drm-misc-next ?

12:41 <marex> they have AB/RB, I'm just a bit sensitive about applying my own patches