12:10 <alyssa> Alright, starting up some scheduler stuff to get back into the swing of this

12:11 <alyssa> Taking a much more conservative approach: the goal is to do no scheduling yet

12:11 <alyssa> Rather, I'd like to get the dep graph/worklist code working and solid and then seemlessly transition

12:11 <alyssa> It should be possible to do in-order scheduling even with the worklist, so this will help debug that

12:12 <alyssa> (And then we can push that with no regressions, and concentrate on real scheduling)

12:49 <bbrezillon> alyssa: should I drop "panfrost: Take draw call order into account" in my v3?

13:27 <alyssa> bbrezillon: Yes, I think so, there's enough shuffling as it is :)

13:27 <alyssa> Nothing we can't cherrypick back later

13:30 <bbrezillon> alyssa: it's part of the series, but I'll leave them on the side

15:36 <shadeslayer> could someone have a look at https://gitlab.freedesktop.org/shadeslayer/linux/commit/05725ebfc9aafaa05d9ad0b89a181252a1522eee

15:37 <shadeslayer> I think I implemented the changes that were requested on the ML, just want to make sure that I understood correctl

16:06 <warpme_> dears: FYI: I'm building every 2..3 days mesa and testing it on s905 & s912. since last few days mesa gives me white screen on s912 (t820). is this wip or regression?

16:07 <warpme_> kernel is 5.3.0 with https://github.com/superna9999/linux/commits/amlogic/v5.3/panfrost applied

16:43 <CrystalGamma[m]> hi everyone! I'd like to contribute to panfrost but I don't have experience with the mesa code base yet … any suggestions for useful tasks to get me familiar with the code?

16:49 <bbrezillon> warpme_: can you bisect?

16:51 <bbrezillon> I've pushed quite a few changes recently, it's very possible that one of them regressed something on t820 :-/

17:05 <warpme_> bbrezillon: maybe at weekend I'll find some time. Is it worth also to test on h6 (t720)?

19:31 <warpme_> bbrezillon: bisect tell me: "c9bebae2877e55cdcd94f9f9f3f6805238caeb28 is the first bad commit"

19:48 <bbrezillon> warpme_: can you try with http://code.bulix.org/ox84jz-874439

19:48 <bbrezillon> ?

19:49 <warpme_> should I do this on current master?

19:52 <bbrezillon> warpme_: yep

19:53 <warpme_> ok building...

20:00 <warpme_> hmm: 99c186fbbe3e2da0f176eadb1db0a8a640786384 + this one line changer: still white screen :-(

20:18 <bbrezillon> warpme_: can you try http://code.bulix.org/n9u4q2-874463 ?

20:20 <bbrezillon> warpme_: or http://code.bulix.org/sh7ehd-874467

20:24 <warpme_> nope: both are still NOK

20:30 <bbrezillon> warpme_: hm, ok, I think I'll just revert the commit tomorrow if I can't find the issue

20:32 <bbrezillon> warpme_: 2 more things to test => http://code.bulix.org/kfwfrc-874472 and http://code.bulix.org/264jfb-874474

20:37 <warpme_> nope. still nok

20:37 <alyssa> Does reverting fix?

20:38 <warpme_> this is exactly what i will try now :-)

20:40 <alyssa> (Sorry for randomly popping in)

20:44 <warpme_> yes. reverting c9bebae2877e55cdcd94f9f9f3f6805238caeb28 helps!

20:47 <warpme_> btw: playing video on t820 gives me picture with randomly inverted luma&chroma. on mali450 all is ok. Should we look on this in next days or rather lets return to this when we will be near to milestone on pan frost?

20:47 <alyssa> With Kodi?

20:47 <warpme_> no. with mythtv

20:48 <warpme_> this is with sw decode so cpu is 100%. maybe this is issue?

20:51 <anarsoul> warpme_: mali450 is totally different GPU

20:51 <warpme_> yes. i know. i put this comment to tell that sw is exactly the same. just runs on different hw.

20:53 <warpme_> sw is exactly the same->from distro maker perspective (me)

21:22 <Lyude> warpme_: it's not though, mali450 uses a different driver

21:22 <Lyude> mali450 is lima, midgard/bifrost GPUs are panfrost