15:40 imirkin_: karolherbst: welcome back
15:40 imirkin_: i had a bit of a go at the GL CTS
15:40 karolherbst: imirkin_: hey! Yeah I saw
15:41 karolherbst: I am still more or less o PTO, but I plan to look into the patch and see how that works for me here
15:41 imirkin_: ok
15:42 imirkin_: there's another patch to fake reset stuff
15:42 karolherbst: imirkin_: but that also fixed a bunch of those weirdo piglit tests, no?
15:42 imirkin_: some of the EXT_framebuffer_multisample ones? maybe
15:42 imirkin_: i didn't do the actual comparison
15:42 karolherbst: k
15:42 imirkin_: it also varies whether i force through 3d path or not
15:42 karolherbst: what fake reset?
15:42 imirkin_: e.g. forcing upsample through the 3d path makes it work
15:43 imirkin_: but we don't do that on nvc0
15:43 imirkin_: maybe we should
15:44 imirkin_: anyways, i'm currently stuck on a GTF test with XFB
15:44 imirkin_: also the non-GTF tests, when run in sequence, fail with a hang somewhere around ssbo
15:44 karolherbst: yeah...
15:44 imirkin_: (like a read error is reported)
15:44 karolherbst: but we can ignore the gtf ones
15:44 imirkin_: not really.
15:44 karolherbst: why not?
15:44 imirkin_: those are required for conformance
15:45 karolherbst: they aren't
15:45 imirkin_: i'm told they are.
15:45 karolherbst: then you got told wrongly
15:45 imirkin_: or you did :)
15:45 karolherbst: impossible :p
15:45 imirkin_: but one thing's for sure - we can't both be right
15:45 imirkin_: afaik the GTF tests are gone for ES, but are still there for GL
15:45 karolherbst: the open source one is the full test, we don't need anything else
15:45 karolherbst: might be worth looking into the old submissions, but we don't need to run those
15:46 imirkin_: the submission test docs are pretty clear about it being required.
15:46 karolherbst: maybe the older version of the docs
15:46 imirkin_: anyways ... the GTF test is simple enough, and appears to be pointing out a real issue
15:46 imirkin_: nope, the most recent docs.
15:46 karolherbst: in doubt, we can write khronos and they clarify it
15:47 imirkin_: i haven't gone through and simplified it all yet, but it's on my todo list... there are 3 failing tests, i'll try to work through them
16:37 karolherbst: imirkin_: "5. For OpenGL CTS releases, and OpenGL ES CTS releases prior to opengl-es-cts-3.2.4.0 test cases included in Khronos Confidential CTS must be run and present in the test logs. For opengl-es-cts-3.2.4.0 and later OpenGL ES CTS releases Khronos Confidential CTS results must not be included in the submission package."
16:37 karolherbst: https://github.com/KhronosGroup/VK-GL-CTS/wiki/OpenGL-and-OpenGL-ES-Conformance-Submission-Passing-Criteria
16:37 karolherbst: mhhh
16:37 karolherbst: the wording is not clear enough
16:38 karolherbst: but yeah.. maybe the gtf indeed needs to be ran, could be brought up in khronos to see what they think
16:38 imirkin_: the wording is pretty clear
16:38 imirkin_: it says what i said
16:38 imirkin_: the kc-cts isn't needed for gles
16:39 imirkin_: but is required for GL
16:39 imirkin_: anyways, there are 3 failing tests (i don't have them on me, but easy to determine obviously)
16:39 imirkin_: the first one is with xfb of gl_Layer and gl_PrimitiveID
16:40 imirkin_: but it's weird, since those do work fine in simpler tests
16:40 imirkin_: i went down a lot of wrong paths before determining that the "core" functionality was ok
16:40 imirkin_: so it's basically something weird about what the test is doing
16:40 imirkin_: which is messing things up - unclear what at this point
17:10 karolherbst: imirkin_: it's always worth looking at submissions and see what patches they used on the CTS
17:15 imirkin_: yeah, i had to apply some local patches
17:16 imirkin_: but the GTF failure is real
17:16 imirkin_: or real enough
21:44 imirkin_: karolherbst: btw, are you good with the blitter fix becoming permanent? hopefully my comments explain sufficiently why it's correct.
21:44 karolherbst: as long as it works and it doesn't break stuff I am fine with it :p
21:44 imirkin_: hehe
21:44 karolherbst: not like I had any plans when coming up with the initial idea
21:45 imirkin_: yeah, i'm guessing you were just trying stuff
21:45 imirkin_: which is fine ... but explanations are nice.
21:45 karolherbst: well.. the scaling was off, that was obvious
21:45 imirkin_: right