02:18Plagman: anyone know offhand what ENOSPC means when returned from drmModeAtomicCommit?
02:18Plagman: there's free vram, this is amdgpu
02:20okias[m]: Anyone tried compare Mesa performance with LLVM compiled in Release vs MinSizeRel?
02:20imirkin: Plagman: a few reasons. e.g. /* Make sure source coordinates are inside the fb. */
02:21okias[m]: Since I'm heavy user of LLVMPIPE, I wonder hoe much it could impact performancenon armv7a
02:22imirkin: Plagman: also if you run out of bandwidth on DP
02:26Plagman: ah, coordinates might be it
02:27Plagman: thanks, will look around there
02:28imirkin: Plagman: https://hastebin.com/orutucuxac.php
02:28imirkin: make sure you turn on drm debug messages
02:28imirkin: i like drm.debug=0x1e
02:28Plagman: oh cool, usually i use 0xf and get spammed by a bunch of stuff
02:28Plagman: usually that's a bit better
02:29Plagman: will try it
02:29imirkin: the bottom bit is a LOT of noise
02:29imirkin: i think also diff drivers produce a diff amount of messages for these
02:35imirkin: anarsoul: dunno if your gpu has clip_halfz support, but you may want to plan ahead by using util_viewport_zmin_zmax instead of manually computing those. just a thought.
02:49HdkR: okias[m]: Give it a compile and report back? :)
02:50HdkR: Presume it would take an hour to compile on some ARMv7a device
03:03imirkin: anyone seen this error with building libdrm? https://hastebin.com/secamuhuqo.coffeescript
03:05imirkin: oh, i think it's just missing -lm?
03:10imirkin: and that only happens because i have some local patches. ok. all figured out -- add dep_m to the modetest deps
04:01anarsoul: imirkin: I don't know either. I haven't gotten to it yet
08:20MrCooper: Plagman: lack of memory would give you ENOMEM
08:35mlankhorst_: git diff
08:36mlankhorst_: oops
10:50okias[m]: HdkR I except it takes much more on my tegra3 tablet (Nexus 7 2012). Anyway, I could also compile on my ultrabook, but... I guess it will take a long time
18:10storgance: this channel is *nix specific, right?
18:14storgance: any specific ways you can prevent apps from flashing the taskbar back up? my taskbar is auto-hid but every time i get a notification in a messaging app it pops my taskbar up un-hiding it, then refuses to re-hide it without me opening the application or closing the application.
18:14storgance: mb, wrong channel on that
18:16imirkin_: this channel is for developers of the direct rendering infrastructure on linux (and other places it's been ported to)
18:18storgance: ok, thank you imirkin.
20:13anarsoul: daniels: narmstrong: any comments on https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3163 ?
20:15daniels: anarsoul: can you shard the mali400 job across two devices?
20:16anarsoul: any examples on how to do that?
20:29daniels: anarsoul: yeah, look at the a6xx jobs - anholt can tell you more. splitting it across two devices should bring the 400 duration down to about the average of the rest of the test jobs
20:30anarsoul: OK, will do
20:45anarsoul: daniels: well, it takes ~7mins for lava-sun8i-h3-libretech-all-h3-cc to boot, so it doesn't look like deqp is a bottleneck here
20:46jekstrand: bnieuwenhuizen: I'm 5 minutes from a WIP mutable_format MR
20:56daniels: anarsoul: the mali400 job took 12min?
20:58anarsoul: daniels: https://gitlab.freedesktop.org/anarsoul/mesa/-/jobs/1382041 13min 19s
20:59anarsoul: also something's wrong with attributes on 32-bit :\
21:05anarsoul: daniels: tomeu: narmstrong: how do I debug why lava CI jobs don't appear? I made this change to pass CI_NODE_INDEX to lava but now there's no lava jobs in pipeline: https://gitlab.freedesktop.org/anarsoul/mesa/commit/7580a7c465a252318eeef6267abd782f742d8aee
21:11anarsoul: oh, nevermind
21:19jekstrand: bnieuwenhuizen: Posted a MR. I think it's in pretty good shape
21:19jekstrand: chadv1: It also includes some of the bits we need for modifiers
21:47anarsoul: OK, it seems to work, mali400 job now takes 7mins 24secs