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!
jrmuizel has joined #lima
Da_Coynul has joined #lima
Da_Coynul has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Da_Coynul has joined #lima
Da_Coynul has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
kaspter has joined #lima
jrmuizel has quit [Remote host closed the connection]
Barada has joined #lima
yuq825 has joined #lima
raimo has joined #lima
mardestan has quit [Read error: Connection reset by peer]
hoijui has joined #lima
raimo has quit [Ping timeout: 264 seconds]
suprothunderbolt has joined #lima
<suprothunderbolt> I've got the Lima drivers running! I've got a fe issues with QML though, I'm not sure what the current state is but circles don't render...
<suprothunderbolt> if i have a square it'll render but if I specify radius: 2 nothing will be drawn.
yuq825 has quit [Remote host closed the connection]
<suprothunderbolt> the only error i'm getting is : gpir: if nir_cf_node not support
<rellla> suprothunderbolt: control flow in vertex shader is missing. you can try this MR though: https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1906#note_221892
<suprothunderbolt> oh neat, thanks
<suprothunderbolt> oh that fixes a ton of stuff
hoijui has quit [Ping timeout: 264 seconds]
hoijui has joined #lima
hoijui has quit [Ping timeout: 264 seconds]
hoijui has joined #lima
suprothunderbolt has quit [Ping timeout: 244 seconds]
mrueg has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
mrueg has joined #lima
hoijui has quit [Ping timeout: 264 seconds]
Da_Coynul has joined #lima
Da_Coynul has quit [Client Quit]
Da_Coynul has joined #lima
Da_Coynul has quit [Client Quit]
Da_Coynul has joined #lima
Da_Coynul has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
<rellla> huh, wanted to give piglit a chance with all the latest MRs pulled, but i get an enormous count on dmesg-fails regarding lima_sched_timedout_job ... so it seems i should sort out a bit :p
<enunes> rellla: did you also pull the kernel patch on the dri-devel mailing list
<rellla> enunes: no, but i did not use the bo cache commit. seems lima driver crashed. i did a rmmod/modprobe and it seems to work again :)
<rellla> running piglit on top of master now, including the following MRs: undef, vs-vf-flattening, missing gpir-functions, gpir-lowering-fixes... can't await the fixes :)
<rellla> ...and regressions :p
<enunes> have you been running with gbm or switched to something like glx/egl_x11 ?
<enunes> btw I have an automated system running it with master every day, maybe I should push the results somewhere if it's useful for you to compare
<rellla> only gbm all the time.
<rellla> i also hacked some scripts to run the tests, but do manual pushes to my webspace for now. could we use any place where we can push the html output?
<rellla> enunes: it's no problem for me to also set up an automation running it with master, so i can produce feature-related html summaries.
<enunes> can't think of anything good and easy to automate now, I'd just push to a public server of mine
<rellla> enunes: so i'm doing right now with imkreisrum.de/piglit/
<enunes> yes something like that
<enunes> I'm also compiling the latest drm-misc-next and loading on it every day
<enunes> not really feature oriented but to track regressions
adjtm has quit [Ping timeout: 245 seconds]
<rellla> hm. some test is still wrapping up lima... have to find out, what the reason is.
dddddd has joined #lima
hoijui has joined #lima
adjtm has joined #lima
deesix has quit [Ping timeout: 245 seconds]
dddddd has quit [Ping timeout: 244 seconds]
hoijui has quit [Ping timeout: 264 seconds]
dddddd has joined #lima
deesix has joined #lima
libv_ has joined #lima
libv has quit [Ping timeout: 258 seconds]
Barada has quit [Quit: Barada]
<MoeIcenowy> anarsoul: strange... lima-serialize seems to be making some glamor rendering jobs finish after X server is terminated
hoijui has joined #lima
adjtm has quit [Ping timeout: 258 seconds]
hoijui has quit [Ping timeout: 268 seconds]
adjtm has joined #lima
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
MoeIcenowy has joined #lima
jrmuizel has joined #lima
hoijui has joined #lima
<anarsoul> MoeIcenowy: likely it just reveals some hidden bug
Barada has joined #lima
Barada has quit [Quit: Barada]
Barada has joined #lima
<anarsoul> rellla: how does it fail?
jrmuizel has quit [Remote host closed the connection]
jrmuizel has joined #lima
<rellla> i'll push the results of master .. master+undef+gpir-funcs in a minute and then test MoeIcenowy's MR again. it's definitely that one.
<rellla> i'lll keep an eye on it then to see which test is the first one.
libv_ is now known as libv
Barada has quit [Quit: Barada]
<rellla> no regressions, 30 fixes.
libv_ has joined #lima
libv has quit [Ping timeout: 245 seconds]
libv_ is now known as libv
<rellla> anarsoul: ok, so dmesg-fails start with glsl-fs-convolution-2 https://pastebin.com/raw/tuXPifq0
<rellla> let me see the logs ...
<rellla> the "[drm:lima_sched_timedout_job [lima]] *ERROR* lima job timeout" should be the corresponding ones.
adjtm has quit [Remote host closed the connection]
adjtm has joined #lima
<anarsoul> rellla: probably it never exits this loop
<anarsoul> try inspecting disassembly?
<anarsoul> also try to confirm that this test is culprit, rmmod/insmod lima and run only this test
hoijui has quit [Ping timeout: 252 seconds]
<anarsoul> rellla: I noticed that it stores counter into temporary
<anarsoul> so -30 is counter, -26 is counter + 1 (yeah, I know that looks stupid but that's what we get from nir)
<anarsoul> so if spilling works fine it this shader should work
<anarsoul> enunes: ^^
<enunes> yeah let's blame spilling :)
<enunes> which huge shader is that that goes to -30?
<anarsoul> could be an issue with cf as well
<bshah|matrix> FTR: https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1906 causes the infinite loop and eventually oom for me
<anarsoul> you need to rebase it onto master
<anarsoul> enunes: it's glsl-fs-convolution-2.shader_test
<bshah|matrix> anarsoul: any specific thing in master that might be fixed there to not cause infinite loop?
<enunes> is this one that passed before or one of the MRs makes it run?
<anarsoul> bshah|matrix: yes, cwabbott merged this fixes to gpir last night
<bshah|matrix> I had reported it to cwabbott about that loop earlier FTR
hoijui has joined #lima
<anarsoul> enunes: IIRC it failed before due to missing undef handling
<bshah|matrix> Aha
<anarsoul> bshah|matrix: yes, and he fixed it
<anarsoul> darn, it's pretty simple shader
<anarsoul> but it gets unrolled because we don't support arrays
<anarsoul> we can add support for indirect uniforms/varyings loads
<anarsoul> and we probably should
<anarsoul> yeah, we should enable PIPE_SHADER_CAP_INDIRECT_CONST_ADDR and PIPE_SHADER_CAP_INDIRECT_INPUT_ADDR and add support for uniform and varyings offsets
<anarsoul> rellla: if you're looking for another task that one ^^ should be pretty easy
<anarsoul> just a matter of checking if load index is constant and if it's not add a source, and then add it to codegen
<anarsoul> if no one's planning to do that I'll look into it at the end of week
<anarsoul> and we'll have to call nir_lower_indirect_derefs() for gpir since implementing indirects is pretty tricky there
drod has joined #lima
<anarsoul> cwabbott: do you understand how indirect uniform load works for GP?
drod has quit [Ping timeout: 258 seconds]
<anarsoul> do I understand correctly that we have to deal with register store latencies to use register as an index?
chewitt has joined #lima
drod has joined #lima
adjtm has quit [Ping timeout: 245 seconds]
adjtm has joined #lima
hoijui has quit [Ping timeout: 245 seconds]
jrmuizel has quit [Ping timeout: 268 seconds]
jrmuizel has joined #lima
jrmuizel has quit [Remote host closed the connection]
drod has quit [Read error: Connection reset by peer]
drod has joined #lima
Da_Coynul has joined #lima
drod has quit [Remote host closed the connection]
Da_Coynul has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
Da_Coynul has joined #lima
drod has joined #lima
jrmuizel has joined #lima
drod has quit [Remote host closed the connection]
Elpaulo has quit [Ping timeout: 268 seconds]
Da_Coynul has quit [Quit: My MacBook Air has gone to sleep. ZZZzzz…]
suprothunderbolt has joined #lima