04:11airlied: dj-death: does vk-gl-cts pass on dg2? just seeing some gpu crashes here, wondering if that's normal
04:14bluetail: airlied what kind of crashes? real driver crashes? wondering if I can contribute with my NAVI setup (6950XT)
04:15Sachiel: airlied: there are some known failures, but crashes? Unless they are the modifiers asserts, I know of none
04:15airlied: Sachiel: like kernel dies and doesn't reset
04:15airlied: [ 242.112462] i915 0000:03:00.0: [drm] *ERROR* GuC engine reset request failed on 0:0 (rcs0) because 0x00000000
04:15Sachiel: that does sound bad. Can't say I've had that happen in a while
04:16airlied: starts there, and then degrades
04:16airlied:will try a linear cts run instead of deqp-runner
04:16Sachiel: I can try a full run on recent stuff tomorrow, not really in the mood of killing my desktop at the moment
04:17airlied: bluetail: nah this is intel dg2
04:18airlied: Sachiel: I'll try a run on the integrated coffeelake
05:01kurufu: if you happen to have a graphical session active, and can test again without a graphical session active it might be useful to note in https://gitlab.freedesktop.org/drm/intel/-/issues/7573 if it also affects cts.
05:04Sachiel: I haven't had the cts hit that one, but I guess it can't be entirely ruled out either
05:09kurufu: I did see thats GuC related in the crash which I cant say I remember seeing, so maybe not likely to be related...
07:17hakzsam: Sachiel: I need to update it and test against latest CTS I think
09:18tzimmermann: jfalempe, apologies for not being very resonsive to the ast bug. i'm fairly busy this week
09:19tzimmermann: 'responsive'
09:20jfalempe: tzimmermann, no problem, I think I made some wrong assumption when reading the code, so my first patch was completely wrong.
09:21jfalempe: I hope the v2 makes more sense ;)
09:56tzimmermann: jfalempe, your comment on the cursor address makes sense to me. thanks a lot for taking care of this bug
09:57jfalempe: tzimmermann, you're welcome.
10:22mlankhorst: 1
10:22mlankhorst: oops
14:32hakzsam: is 23.0 going to be released today?
14:33dcbaker: hakzsam: yes, that's the plan
14:34dcbaker: I had sick kids for the first part of the week, so I was only pulling stuff the applied cleanly, and wanted to get a bit more done before I made an actual release
14:35dcbaker: zmike: "zink: fix pointcoord y inversion" doesn't apply cleanly to 23.0, and the backport attempt I made yesterday led to a bunch of regressions, do you want to manually backport that or drop it?
14:35zmike: uhhhhhh
14:36zmike: gimme like 30s to preempt...
14:37dcbaker: Venemo: "ac/nir/ngg: Include culled primitives in query" doesn't apply cleanly to 23.0, and the backport I attempted yesterday led to a bunch of regressions, do you want to manually backport that or drop it?
14:37zmike: dcbaker: do you have an example of a job you ran with your attempt?
14:38dcbaker:goes to see
14:40dcbaker: zmike: https://gitlab.freedesktop.org/mesa/mesa/-/commit/c0592be5973c2d6e9aea923cadeb1f0355db4788
14:41zmike: dcbaker: the ci jobs tho
14:41zmike: I think zmike/backport should do it
14:41zmike: based off staging/23.0
14:42dcbaker: https://gitlab.freedesktop.org/mesa/mesa/-/jobs/36100233
14:42dcbaker: zmike: ^
14:42zmike: ah ok
14:42zmike: I think my branch should fix that, looks like you just missed a ci fail in the conflict
14:43dcbaker: ... I guess I should have looked at the diff more closely.
14:43dcbaker: Thanks for backporting :)
14:44zmike: 👍
16:09jani: airlied: danvet: hey, I could use a little help in deciding what to try to do with https://lore.kernel.org/r/Y+JLQfuSAS6xLPIS@ideak-desk.fi.intel.com - it's basically fixes that would be great to get backported to stable 6.1, but we're at -rc7, it touches three drivers (though amd and nouveau changes should be non-functional), and it's not exactly within stable kernel rules size limits
16:11jani: I'm leaning towards aiming for the merge window and 6.3-rc1, feels too risky for even considering v6.2
16:12jani: that would mean drm-misc-next-fixes or drm-intel-next-fixes, any thoughts on that?
16:12jani: imre: ^
16:13jani: well, I guess just too late for v6.2 anyway, unless there's -rc8
16:14vsyrjala: i thought linus promised an -rc8
16:17jani: I've missed that part
16:30danvet: jani, I guess nouveau/amd would need something like 4/4 too?
16:31danvet: imo this is fine for stable, but also it's mst, which has gone boom in fantastic ways rather much
16:31danvet: so -next-fixes sounds ok, other drivers should be able to backmerge fairly quickly after the next pull or so
16:48jani: danvet: thoughts on drm-misc-next-fixes vs drm-intel-next-fixes?
16:48danvet: *shrug*
16:49danvet: -next-fixes tends to be weekly so really doesn't matter, it's equally just a backmerge away for the other drivers
16:49danvet: I mean unless agd5f or Lyude or someone already wants to drop these patches somewhere, then maybe misc
16:50danvet: so a-b: me for intel if you feel that way
16:50danvet: I momentarily forgot what my job is here :-)
16:50danvet: mislaid the a-b stamper
16:51jani: danvet: thanks. I'll need acks from agd5f et al too
16:52danvet: sounds good
17:34agd5f: jani, probably best to get an ack from Wayne or Harry
18:33jani: agd5f: ack :)
19:34gfxstrand: hakzsam: Ahy update on !17922 with private fossil-db?
19:35gfxstrand: hakzsam: cmarcelo and I would both really like to get that landed.
19:58Venemo: dcbaker, zmike do you need help backporting that, or is that solved now?
21:53dcbaker: Venemo: I could still use help with the ac/nir/ngg patch; there were two patches, and I was lazy and copy-pasted my message, maybe a mistake in retrospect
23:34airlied: dj-death: I filed an internal khronos issue for the format test
23:35Venemo: dcbaker: okay, what is the best way I can help you? should I open a MR against 22.3?
23:36dj-death: airlied: thanks
23:36dcbaker: Venemo: yes please :)
23:36Venemo: okay, I can probably do that early mext week
23:37airlied: dj-death: on my coffeelake that's the only regression I have with cts
23:43dj-death: CTS doesn't test much tbh :)