07:46 <rellla> anarsoul: i guess, we can add the missing whitespace later :p

07:47 <anarsoul> ?

07:47 <anarsoul> what's missing?

07:47 <rellla> fprintf(fp, "*/\n\t\t\t\t\t\t/* %s(3)", render_state_infos[i].info);

07:47 <rellla> should be fprintf(fp, " */\n\t\t\t\t\t\t/* %s(3)", render_state_infos[i].info);

07:47 <rellla> but that's a nitpick

07:49 <anarsoul> fixed

07:49 <rellla> :)

07:49 <anarsoul> thanks for review!

07:50 <rellla> yw

07:50 <rellla> how is covid-19 going overseas?

07:50 <anarsoul> depressing

07:51 <anarsoul> I'm working from home since Monday

07:53 <anarsoul> grocery stores are crowded, everyone is trying to stock up

07:54 <rellla> we have some toilet-paper-mania here :(

07:54 <anarsoul> yeah, same here

07:54 <ente> same

07:54 <ente> sucks for the people who actually need it

07:56 <anarsoul> there're rumors that they're closing border with US tomorrow

07:56 <anarsoul> so looks like Canada is in lockdown

07:56 <anarsoul> you can't get in, you can't get out

07:56 <rellla> i should have taken a picture from a woman on bicycle which had one packet toilet-paper on the left steering gear and one on the right.

07:57 <ente> Austria closed most of its borders too

07:57 <rellla> it's like you can see it on youtube, but i didn't believe, that we have that kind of people in our town ...

07:57 <rellla> germany also is in shutdown-mode

07:59 <ente> on the other hand now that I'm in home office, I'm doing lots of non-job-related stuff that I've been putting off for way too long

07:59 <ente> if there was ever a right time to exclusively sit at home and write code, it's probably now

08:12 <anarsoul|c> hehe

08:12 <anarsoul|c> We still have some tests to fix :)

08:13 <rellla> my home office is in the office :)

17:09 <anarsoul> enunes: hi, how's your ppir scheduler rework is going?

17:12 <enunes> anarsoul: it's still going, I got a bit of a slow down with the current situation and didnt work on it for some days, but I'm still on it

17:16 <anarsoul> I see

17:16 <anarsoul> are you expecting any deqp fixes due to scheduler rework?

17:18 <enunes> I don't expect anything to change, if it does it is more by luck

17:18 <enunes> so far only looking at instruction count

17:20 <enunes> well, shader-db results overall

17:20 <enunes> I do run deqp run sometimes but mostly to catch the bugs that I have introduced with the refactors

17:23 <enunes> I think the first thing I will submit in a separate series as a pre/incremental step is to merge spilling with scheduling, with my experiments so far its not possible to try either smart or dumb combining with spilling happening standalone at regalloc

17:23 <enunes> so that has to happen anyway and not break anything

17:37 <anarsoul> we have some deqp failures for fragment shaders

17:37 <anarsoul> I haven't looked into these yet though

17:37 <anarsoul> and probably it doesn't make sense until your changes land

17:47 <anarsoul> I'll look into shared_depthbuffer failures

17:50 <anarsoul> it looks like it's not doing depth reload when it really should

18:25 <anarsoul> enunes: btw you can run your changes through CI without submitting an MR

18:26 <anarsoul> just push your changes into a branch with name starting with 'ci-' and go to gitlab.freedesktop.com/enunes/mesa/pipelines to run the jobs manually

19:18 <anarsoul> warpme_: hey, have you had a chance to look into buffer importing issue with lima?

19:29 <warpme_> anarsoul: unfortuntelly not. but for sure I can do quick test on s905. Is this mean that it my be related with #2377 ?

19:30 <anarsoul> yep, but I doubt it's fixed

19:30 <anarsoul> there were no changes in that area for quite a while

19:31 <anarsoul> if you have sample code that would work on cedrus I can give it a try

19:32 <anarsoul> I don't think that apitrace would help since buffers won't be coming from vpu

19:32 <anarsoul> I wonder if vpu produces buffers that can't be imported due to some restriction

19:34 <warpme_> ok give sec. i'll compile current master and report... re: sample code&cedrus - for me this means H5 or lower. For AW I have only H6 HW :-( As software - i'm doing tests with mythtv and i'm afraid to reproduce issue - working mythtv is needed :-(

19:37 <anarsoul> would mythtv work with cedrus?

19:48 <anarsoul> dEQP-GLES2.functional.shaders.random.trigonometric.fragment.65 fails due to insufficient precision

19:48 <anarsoul> it uses atan() which doesn't exist in nir and thus lowered

19:49 <anarsoul> but lowering trigonometric fns with fp16 is not a good idea :)

19:52 <warpme_> more general issue I have is finding root cause of black screen playback on lima: is this because mythtv not correctly implements https://gitlab.freedesktop.org/lima/mesa/issues/130#note_391668 or rather issue is within lima. Ideally will be to verifiy apitraces from mythtv on 100% working 3D driver - to see where is root cause

19:53 <warpme_> fyi: I just done tests on current master and basically all works (i mean GLES UI rendering and also all 6 variants of GLSL deinterlacers). only video playback gives still black screen - like: https://gitlab.freedesktop.org/mesa/mesa/issues/2377#note_394546

19:54 <warpme_> s905=lima; s912=panfrost.

19:56 <warpme_> but I still don't know where root casue is: a\mythtv conversion from amlogic vdec output to lima required input ( like suggested in https://gitlab.freedesktop.org/lima/mesa/issues/130#note_391668) or b\this conversion is OK and issue is within lima....

19:59 <warpme_> re: mythtv on cedrus: we are adding support for v4l2_request in mythtv - but currently a\ team is busy with v31 release; b\main OpenGL developer have issue to het working his H6 tanix box (ough - this is another huge, frustrating story...)

20:01 <anarsoul> I see

20:02 <warpme_> oops - sorry: s905 gives me not black screen. It gives me flickering part of UI. Interesting that playback OSD works OK (i see progress bar and I can browse OSD menus...)

20:10 <anarsoul> warpme_: anything in dmesg?

20:25 <warpme_> nope. dmesg seems to be clean.

20:26 <anarsoul> could it be some error that you're not handling?

20:26 <anarsoul> note that midgard supports nv12

20:26 <anarsoul> (i.e. panfrost)

20:27 <anarsoul> utgard doesn't support any yuv formats

20:37 <warpme_> let me check this with our dev doing opengl....

20:56 <anarsoul> however it should be handled internally in mesa

21:20 <anarsoul> warpme_: I think it should be able to import yuv unless planes are not aligned to 16 bytes

21:20 <anarsoul> (that's hw requirement)

21:52 <anarsoul> warpme_: fwiw kmscube -M nv12-1img nor -M nv12-2img doesn't work on lima

21:52 <anarsoul> it fails in eglCreateImageKHR

22:16 <anarsoul> warpme_: lowering external samplers is broken if driver doesn't support R8 and R8G8 :(

22:24 <anarsoul> I'll look into it