ChanServ changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - Logs https://oftc.irclog.whitequark.org/panfrost - <macc24> i have been here before it was popular
<cphealy> cool, tnx
<alyssa> cphealy: It was backported
<cphealy> I wasn't aware of the 'Cc: mesa-stable' tag, tnx.
nlhowell has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
chewitt has quit [Quit: Zzz..]
davidlt has joined #panfrost
warpme_ has quit [Quit: Connection closed for inactivity]
davidlt has quit [Ping timeout: 480 seconds]
alpernebbi has joined #panfrost
davidlt has joined #panfrost
rasterman has joined #panfrost
user92 has joined #panfrost
user92 has left #panfrost [#panfrost]
<macc24> tomeu: what's up with g72 ci? is it working now?
<tomeu> macc24: there's a bug in panfrost that causes unreliability
<macc24> tomeu: ack
<icecream95> It's the cases where the 2457-clause shader has a reason to exist that I'm worrried about...A
<icecream95> I've got a fairly good idea of how the blob handles geometry shaders now, and writing an implementation that works for the subset of usage tested by dEQP shouldn't be too hard
<tomeu> icecream95: cool, I don't have a g72 here myself
<macc24> btw, xorg runs better than i thought it would on duet
<macc24> with hexdump0815's xorg.conf
<icecream95> tomeu: My statement about the "2457-clause shader" was meant to come after the one about geometry shaders, it's not related to the g72 ci issues
<icecream95> I think the CI failures are caused by the kernel, because running dEQP tests locally with the same setup (but different kernel) as used in CI doesn't have any failures
<tomeu> ah, ok
<tomeu> hmm, really?
<macc24> icecream95: what kernel are you running?
<icecream95> macc24: Still 5.10.1 with the Cadmium patches
<macc24> icecream95: can you try on 5.14?
<macc24> with cadmium patches
camus has joined #panfrost
camus1 has quit [Read error: Connection reset by peer]
atler is now known as Guest3376
atler has joined #panfrost
Guest3376 has quit [Ping timeout: 480 seconds]
macc24 has quit [Ping timeout: 480 seconds]
<rasterman> when in doubt... always blame the kernel :)
nlhowell has joined #panfrost
nlhowell has quit [Ping timeout: 480 seconds]
<bbrezillon> alyssa, icecream95: have any of you started working on a panfrost -> kbase drm_shim ?
<alyssa> bbrezillon: I think icecream95 has code for that somewhere, or maybe it was the other way around
stepri01 has quit [Quit: leaving]
karolherbst has quit [Remote host closed the connection]
karolherbst has joined #panfrost
nlhowell has joined #panfrost
karolherbst has quit [Quit: Konversation terminated!]
karolherbst has joined #panfrost
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
davidlt has quit [Ping timeout: 480 seconds]
camus has joined #panfrost
camus1 has quit [Read error: Connection reset by peer]
alpernebbi has quit [Quit: alpernebbi]
narmstrong has quit [Ping timeout: 480 seconds]
cwabbott has quit [Ping timeout: 480 seconds]
warpme_ has joined #panfrost
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
nlhowell has quit [Ping timeout: 480 seconds]
<alyssa> https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12208 needs testing (it touches a /lot/ of code)
<alyssa> (Probably won't be merged until close to the end of the month, so no rush.)
<icecream95> Recently I've written three different methods for serialising BO contents / GPU jobs:
<icecream95> - logging to a file in my "dump-hacks" branch
<icecream95> - sending over the network with checksumming to only send changed BOs, and zstd compression ("net-shim")
<icecream95> - hexdump and pandecode before/after execution to a file ("decode")
<icecream95> Maybe I should try unifying these so I can write tools that work for all of them
warpme_ has quit [Quit: Connection closed for inactivity]
rasterman has quit [Quit: Gettin' stinky!]