ChanServ changed the topic of #lima to: Development channel for open source lima driver for ARM Mali4** GPUs - Kernel has landed in mainline, userspace driver is part of mesa - Logs at https://people.freedesktop.org/~cbrill/dri-log/index.php?channel=lima and https://freenode.irclog.whitequark.org/lima - Contact ARM for binary driver support!
yuq825 has joined #lima
jernej has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
jernej has joined #lima
anarsoul|2 has joined #lima
buzzmarshall has joined #lima
anarsoul has quit [Ping timeout: 272 seconds]
maciejjo has quit [Ping timeout: 265 seconds]
maciejjo has joined #lima
dddddd has quit [Remote host closed the connection]
buzzmarshall has quit [Quit: Leaving]
megi has quit [Ping timeout: 272 seconds]
anarsoul|2 is now known as anarsoul
monstr has joined #lima
yann has quit [Ping timeout: 265 seconds]
yann has joined #lima
yuq825 has quit [Remote host closed the connection]
megi has joined #lima
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo has joined #lima
dddddd has joined #lima
austriancoder has quit [Read error: Connection reset by peer]
austriancoder has joined #lima
dddddd has quit [Ping timeout: 240 seconds]
deesix has quit [Ping timeout: 258 seconds]
deesix has joined #lima
dddddd has joined #lima
<enunes> rellla: btw, I have a fix/workaround for the regression you mentioned in the random deqp test, but I think it's something we could fix in ppir scheduler or even in nir
<enunes> I think it's kind of an existing bug that was just exposed
<enunes> my patches also fix a few other deqp tests that were failing even before the new liveness analysis implementation
<rellla> enunes: nice, i'll give it a try once you pushed them
<enunes> I'll give it a quick shot in seeing how hard it would be to improve nir to not emit that code
chewitt has joined #lima
jernej has quit [Quit: Free ZNC ~ Powered by LunarBNC: https://LunarBNC.net]
jernej has joined #lima
Elpaulo1 has joined #lima
Elpaulo has quit [Ping timeout: 265 seconds]
Elpaulo1 is now known as Elpaulo
monstr has quit [Remote host closed the connection]
yann has quit [Ping timeout: 265 seconds]
paulk-leonov has quit [Excess Flood]
paulk-leonov has joined #lima
Elpaulo has quit [Quit: Elpaulo]
ric96 has quit [Ping timeout: 265 seconds]
ric96 has joined #lima
yann has joined #lima
<anarsoul> enunes: looks like there's some regression in ppir compiler, ubports folks are seeing weird color artefacts
<UnivrslSuprBox> Hello I'm here, and I have a trace that almost plays
<anarsoul> almost?
<anarsoul> white screen with green "uhoh" button here
<anarsoul> enunes: ^^
<UnivrslSuprBox> Retracer on my desktop doesn't like the 'eglGetPlatformDisplayEXT' call, or the eglQueryString response
<UnivrslSuprBox> This is how it looks on display
<UnivrslSuprBox> I can't decide if it's more 60s or 90s
<anarsoul> :)
<UnivrslSuprBox> brb, switching client
UnivrslSuprBox has quit [Quit: ZNC 1.6.6+deb1ubuntu0.2 - http://znc.in]
UnivrslSuprBox has joined #lima
<UnivrslSuprBox> There
<anarsoul> UnivrslSuprBox: please open new bug on https://gitlab.freedesktop.org/mesa/mesa/issues and attach your apitrace there
<UnivrslSuprBox> On it
<anarsoul> thanks
<enunes> anarsoul: yeah rellla noticed one regression as well, did they confirm it was the liveness patch?
<anarsoul> nope
<enunes> the issue doesn't mention if it is a regression or not
<anarsoul> we haven't had any other compiler changes recently though
<enunes> anyway, I have a local branch with some fixes
<enunes> let me push it then
<enunes> yeah that one, doesn't mention if it is a regression
<UnivrslSuprBox> Apologies for taking pictures of the screen, but here's another problem. I'm not sure if it's the same or if it should be a new trace and bug report?
<enunes> UnivrslSuprBox: pictures don't help much; whether it is a regression helps, and bisect would be better
<enunes> but it's likely it's the liveness patch
<UnivrslSuprBox> Sure, let's get that building
mariogrip has joined #lima
<anarsoul> we can use *any* register if it's undef
<anarsoul> it shouldn't matter whether it's used for some live value or not
<enunes> anarsoul: which is why I didn't push that yet; though it fixes 5 deqp tests, 4 of which weren't even regressions
<anarsoul> that's weird
<anarsoul> I'd assume it masks some other bug
<enunes> yeah we definitely have bugs related to that
<enunes> not regressions due to liveness, though
<anarsoul> if you need some help please let me know and I'll look into it
<UnivrslSuprBox> No more funky fresh colors, so we're getting somewhere!
<enunes> UnivrslSuprBox: does that mean it fixes the issue? did the thing you are reporting ever work before?
<UnivrslSuprBox> It *does* fix the strange red and blue shader, yes
<UnivrslSuprBox> shading*
<UnivrslSuprBox> enunes, what was the original ppir liveness change?
<UnivrslSuprBox> We can try to build without it and get a full regression list
<UnivrslSuprBox> if any
<enunes> UnivrslSuprBox: https://gitlab.freedesktop.org/enunes/mesa/commits/ppir-new-regalloc , it fixes glamor shaders required for xorg though, so if you're on xorg, you may run into other problems without it
<UnivrslSuprBox> No xorg here, so it should be okay
<UnivrslSuprBox> Just 7e276 and 9bf21 then?
<enunes> pretty much only 9bf210ba, the other one shouldn't have a big impact
<anarsoul> enunes: this one sounds like missing dependencies: https://gitlab.freedesktop.org/enunes/mesa/commit/d01ffcd944bf87b77acd3150a48b1a7dd664ca3a
<enunes> anarsoul: it's a read write emitted by nir
<enunes> dead write*
<anarsoul> I don't really understand why scheduler reorders it though
<anarsoul> let's say we have: r1.xyz = ssa1.xyz
<anarsoul> and then only r1.x is used
<enunes> x and z are used in this case
<enunes> y is dead
<anarsoul> enunes: we need DCE
<enunes> nir should do it, not us
<anarsoul> it doesn't do per-component analysis
<anarsoul> anyway
<anarsoul> even if it's dead ppir compiler should be able to handle it
<enunes> yeah in this case it's a bug in scheduler
<anarsoul> why not to fix it there?
<UnivrslSuprBox> Well, on the dim side, reverting 9bf210ba causes stack smashing
<enunes> I didn't create a MR for that branch because I am looking into nir and ppir scheduler solutions
<anarsoul> enunes: oh, OK
<anarsoul> sorry for the noise then :)
<anarsoul> enunes: it's still sounds like a bug in nir->ppir translation though, likely we're missing a dependency somewhere
<enunes> nit works fine until there is a vec4->per-component translation, and nir dce doesn't catch that really
<enunes> I think a good solution would be to improve nir dce to catch it
<anarsoul> yes, but in addition to bug fix in ppir
<enunes> it might be difficult because it might not be trivial to prove it's actually dead since it's not ssa
<anarsoul> enunes: that's why we need DCE pass in ppir :)
<enunes> DCE in ppir might be sort-of trivial with the liveness algorithm merged
<enunes> just need a set to store dead things, which is pretty easy to detect
<anarsoul> enunes: sounds good
<enunes> and run a pass of that before regalloc
<enunes> but we can only get rid of an instruction if all ops are dead
<enunes> or need to remove an of and reschedule... then it starts getting bad
<anarsoul> yes
<enunes> an op*
<anarsoul> that's why we need to rewrite scheduler :)
<enunes> my plan for next task was to attempt a simpler scheduler
<anarsoul> basically we should start with one op per instruction
<anarsoul> then do liveness analysis
<anarsoul> then optimization passes
<anarsoul> then merging instructions
<anarsoul> or rather
<anarsoul> merge and/or order ops
<anarsoul> merging and reordering may require doing liveness analysis several times
<enunes> I wish I had the time to try all of that :) and possibly include spilling in the loop
<enunes> hopefully there is a way to do this redesign one thing at a time, which seems more feasible
<anarsoul> yeah
<anarsoul> good thing is that our data structures don't need major redesign
<UnivrslSuprBox> I have good news for you and bad news for myself. Reverting both commits at the tip of the liveness branch shows that indeed caused the weird colors
<UnivrslSuprBox> Bad news is SVG rendering and rounded corners across the board are broken both without liveness and your new liveness patches
<UnivrslSuprBox> So that regressed somewhere else I suppose...
<anarsoul> do you have apitrace for rounded corners issue?
<UnivrslSuprBox> Bah, tried to start it and stack smashing... Installing another build
<enunes> UnivrslSuprBox: so with the fixed branch I sent earlier the result is the same as before the liveness patches?
<UnivrslSuprBox> Correct
<enunes> well it's good to know though those are likely not the final patches, I'll look for a proper solution over this week
<anarsoul> I'm not sure why it broke gpir compiler, likely there's some bug in it
<UnivrslSuprBox> Sure, that's building now. In the mean time, I'm back with the liveness patches and I'll get you a trace of something that should have rounded corners
<anarsoul> let me look into it
<UnivrslSuprBox> Here's a trace with an image which should have rounded corners
<anarsoul> again, please open a bug for it on fdo gitlab and attach it here
<UnivrslSuprBox> Sure
<enunes> UnivrslSuprBox: btw, how upstream is that github tree? any patches not from fdo mesa/mesa?
<mariogrip> enunes: its a clean master tree
<mariogrip> from yesterday iirc
buzzmarshall has joined #lima
<mariogrip> Im working togheter with UnivrslSuprBox on this btw :)
<enunes> ok, good
<UnivrslSuprBox> anarsoul, build finished. Still getting stack smashes with all of those reversions
<mariogrip> still smashing indeed
<anarsoul> well, I've never seen any stack smashes
<anarsoul> try debugging it?
<anarsoul> I've ran lima through valgrind last week and it was fine
<mariogrip> anarsoul: it seem to have started fine after some tries
<mariogrip> seems to be random when it hit that
<anarsoul> well, you'll need to debug it
<mariogrip> race condition somewhere?
<mariogrip> anarsoul: will do, need to install some symbols first
<mariogrip> anarsoul: got this https://paste.ubuntu.com/p/5hgDWprdKZ/
<UnivrslSuprBox> Reminder, this is in the branch with the ppir changes *reverted*
<UnivrslSuprBox> Not the new fixes sent here today
<anarsoul> mariogrip: try with valgrind?
<anarsoul> this backtrace isn't very helpful
<anarsoul> rellla: so I checked hsl_to_rgb.vert with 6f394343b1f704f8b98a24add7f4106e72e2db7b and without it
<anarsoul> without it we get more optimal nir code
<anarsoul> with it there's some duplications
<anarsoul> oh sorry
<anarsoul> ah, it's actually correct
<anarsoul> so this optimization results in suboptimal nir code
<anarsoul> but yet it's correct
<anarsoul> so it's a bug in gpir compiler
<anarsoul> :(
<mariogrip> So I hit a valgrind bug, seems like i never valgrind, as this version does not support ldpsw opcodes. but I need to contune that tomarrow, i need to get some sleep
<anarsoul> you need newer valgrind
<mariogrip> sorry i ment s/never/newer :P