04:05 <SolidHal> I'm getting some graphical glitches on xfce with panfrost + mesa and swrast, I'm trying master now but was wondering if anyone has seen something like this?

04:51 <HdkR> With panfrost and swrast? So kernel panfrost and userspace swrast?

04:51 <tomeu> SolidHal: if the glitches are seen with swrast, then I guess panfrost has nothing to do?

04:51 <tomeu> then it could be the rockchip kms, if that's the soc you are using

04:55 <SolidHal> kernel 5.2 panfrost, swrast and mesa built from the git repo

04:56 <SolidHal> Just tried master instead of the 19.2 tag, now when I login to xfce it black screens and kicks me back to the login window

04:57 <tomeu> SolidHal: I don't understand how swrast is used here

04:57 <tomeu> what hw? anything in dmesg?

04:59 <SolidHal> Im compiling with the following command meson .. . -Dprefix=/usr -Ddri-drivers= -Dvulkan-drivers= -Dgallium-drivers=panfrost,kmsro,swrast -Dlibunwind=false

04:59 <urjaman> ah blah

04:59 <SolidHal> asus c201

04:59 <SolidHal> nothing in dmesg

05:00 <urjaman> no you're not using swrast unless panfrost failed to load (or glamor failed to initialize it) or you disabled glamor

05:00 <SolidHal> ah, my mistake

05:00 <urjaman> so dont say that, it's just good to have in a mesa build as a fallback

05:00 <urjaman> what kind of glitches?

05:01 <urjaman> with the 19.2

05:02 <SolidHal> on 19.2 most but not all of the toolbar icons are missing

05:02 <urjaman> oh bleh that old thing

05:02 <SolidHal> known issue?

05:02 <urjaman> on master the shader compilation (spilling) of some glamor shader will fail pretty quickly

05:03 <urjaman> yeah i havent used 19.2, but i do remember that the toolbar thing happened a long time ago

05:03 <urjaman> i can point you a commit hash to build that should be a bit better, wait a moment

05:04 <SolidHal> ah, so it fail and x crashes? That would be awesome, thanks

05:05 <urjaman> try git checkout 6c4b97011b209fb9c034208b2b2f7e261feaf17c

05:06 <SolidHal> thanks urjaman!

05:15 <urjaman> currently known issues for me in Xorg (with a commit that doesnt do the spill failure immediately) are some glitches when trying to scroll gtk things beyond the end, and a memory leak

05:17 <urjaman> plus there are kernel level issues that bbrezillon has been working on that will cause hangs (of some seconds) in Xorg

05:17 <urjaman> so alyssa how about fixing the spill failure so i can get to complaining about the smaller problems ;P

05:18 <SolidHal> I'll be on the lookout for those, I second the request to fix the spill failure :D

05:19 <dhewg> I won't complain today, thanks guys for pushing T720 over the line!

05:21 <urjaman> yeah i know she already knows about most of these things, i just wanted to tell what to expect for now

05:22 <SolidHal> I appreciate it!

05:26 <urjaman> the memory leak for me is that Xorg after starting had about 250MB of VIRT and 90 MB of RSS (after starting my applications) and now 12 hours later is at 714 MB of VIRT and 533 MB of RSS

05:27 <urjaman> so like ... usable for hours but not for days

05:29 <SolidHal> are you just using top to track that, or have you found a more clever way?

05:30 <urjaman> just top for now (i've only just noticed it...)

05:32 <urjaman> noticed it after Xorg got oom-killed a few times while i was away

05:51 <SolidHal> just tried that commit, much much more stable. Thanks again!

07:37 <bbrezillon> alyssa: sorry, won't have time to look at it this week :-/

13:59 <alyssa> bbrezillon: No worries, not urgent

14:00 <alyssa> urjaman: Is it wrong that I'm just used to OOMs as a fact of life now

14:00 <alyssa> tomeu: ^^ maybe we should fix some leaks yo ;P

14:01 <tomeu> do we have leaks, or is it the cache?

14:01 <alyssa> tomeu: Either way.

14:02 <urjaman> can the cache exceed 1GB? :P

14:03 <urjaman> either way that'd be wrong :P

14:03 <alyssa> tomeu: Also, woaaaaah, T720 merged and T820 CI? Busy night!

14:04 <urjaman> but nah i think it's not the cache ... or atleast the cache shouldnt be getting more and more items while i just sit watching youtube

14:04 <alyssa> ...It can, thanks to greedy behaviour

14:05 <alyssa> But new kernels and new userspace are s'posed to fix that (either one would, they both have cache taming stuff so if you're on old kernel + new user or new kernel + old user, it should be okay ... old kernel + old user will leak)

14:20 <bbrezillon> would anyone have time to work on the IGT tests danvet was asking for (to validate the patch 7 of my 5.4 fixes branch)?

14:21 <bbrezillon> s/branch/series/

14:23 <bbrezillon> I'm not so familiar with IGT, and also don't have much time to work on this in a near future

14:23 <bbrezillon> robher, robmur01, tomeu ^

18:10 <robher> bbrezillon: I'm familiar enough with IGT to be annoyed when I do look at it. I don't think a test is a blocker though.

18:34 <alyssa> Well I've been trying random things to debug MRT

18:34 <alyssa> and accidentally fixed a test on T760

18:34 <alyssa> confused.

18:36 <anarsoul> :)