alyssa changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - Logs https://freenode.irclog.whitequark.org/panfrost - <daniels> avoiding X is a huge feature
remexre has quit [Quit: WeeChat 2.9]
embed-3d has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
embed-3d has joined #panfrost
vstehle has quit [Ping timeout: 240 seconds]
stikonas has quit [Remote host closed the connection]
kaspter has joined #panfrost
stollejocke has quit [Ping timeout: 272 seconds]
kaspter has quit [Remote host closed the connection]
kaspter has joined #panfrost
icecrea105 has joined #panfrost
icecream95 has quit [Ping timeout: 260 seconds]
icecrea105 has quit [Client Quit]
icecream95 has joined #panfrost
davidlt has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 240 seconds]
camus1 is now known as kaspter
karolherbst has quit [Read error: Connection reset by peer]
karolherbst has joined #panfrost
icecream95 has quit [Ping timeout: 260 seconds]
davidlt has quit [Ping timeout: 264 seconds]
icecream95 has joined #panfrost
vstehle has joined #panfrost
_whitelogger has joined #panfrost
davidlt has joined #panfrost
rando25892 has joined #panfrost
mixfix41 has quit [Ping timeout: 240 seconds]
yann has joined #panfrost
<chewitt> [ 792.102523] panfrost ffe40000.gpu: AS_ACTIVE bit stuck
<chewitt> ^ that's a new one
<chewitt> from an idle VIM3 .. haven't touched the device (made the Kodi do anything) in ~12 hours
<chewitt> although kernel timer is closer to boot
remexre_ has joined #panfrost
stikonas has joined #panfrost
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 272 seconds]
stikonas_ has quit [Ping timeout: 272 seconds]
cyrozap has quit [Ping timeout: 260 seconds]
stikonas has joined #panfrost
mixfix41 has joined #panfrost
rando25892 has quit [Ping timeout: 240 seconds]
stikonas_ has joined #panfrost
raster has joined #panfrost
stikonas has quit [Ping timeout: 272 seconds]
stikonas has joined #panfrost
stikonas_ has quit [Ping timeout: 256 seconds]
<icecream95> TIL you can use =all for Mesa debug variables. All that time I've wasted typing MIDGARD_MESA_DEBUG=shaders,shaderdb when =all would have done the job…
yann has quit [Ping timeout: 240 seconds]
icecream95 has quit [Ping timeout: 240 seconds]
yann has joined #panfrost
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 272 seconds]
raster has quit [Remote host closed the connection]
stikonas_ has quit [Read error: Connection reset by peer]
raster has joined #panfrost
stikonas_ has joined #panfrost
yann has quit [Ping timeout: 256 seconds]
yann has joined #panfrost
stikonas has joined #panfrost
stikonas_ has quit [Quit: Konversation terminated!]
clementp[m] has quit [Quit: killed]
l-as has quit [Quit: killed]
wiizzard has quit [Quit: killed]
Ke has quit [Quit: killed]
nhp[m] has quit [Quit: killed]
tomboy64 has quit [Ping timeout: 240 seconds]
tomboy64 has joined #panfrost
l-as has joined #panfrost
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 240 seconds]
Ke has joined #panfrost
clementp[m] has joined #panfrost
nhp[m] has joined #panfrost
wiizzard has joined #panfrost
raster has quit [Remote host closed the connection]
<macc24> tomeu: is this commit present in mainline linux?
<macc24> mainline as in kernel.org mainline
<macc24> oh i'm missing it
<macc24> thanks
_whitelogger has joined #panfrost
yann has joined #panfrost
tgall_fo_ has joined #panfrost
tgall_foo has quit [Ping timeout: 256 seconds]
<alyssa> icecream95: *gasp*
<alyssa> I didn't know that either. Thanks :)
<alyssa> PAN_MESA_DEBUG=bifrost,nofp16,trace is pretty comoon here
<macc24> alyssa: why nofp16?
<alyssa> macc24: there are some fp16 bugs on bifrost I haven't sorted and trying to catalogue failures better for working through confomrnace stuff
<macc24> oh that's good to hear before ~playing video games~ testing bifrost
popolon has joined #panfrost
tomboy64 has quit [Remote host closed the connection]
tomboy64 has joined #panfrost
nlhowell has quit [Ping timeout: 256 seconds]
<macc24> is there any reason why panfrost has no dt bindings for arm,mali-bifrost?
<tomeu> macc24: that should be in mainline already
<macc24> when i add { .compatible = "arm,mali-bifrost", .data = &default_data, }, at end of this block it works fine
<tomeu> macc24: then that commit is still in drm-misc I guess
<tomeu> macc24: I think I would use the branch we use in CI: https://gitlab.freedesktop.org/tomeu/linux/-/tree/v5.9-rc5-for-mesa-ci
<tomeu> that should ahve all of it
<tomeu> we are now testing bifrost there
<tomeu> alyssa: btw, should we run on CI with nofp16 on bifrost?
tgall_fo_ is now known as tgall_foo
<macc24> tomeu: compiling it...
nlhowell has joined #panfrost
<alyssa> tomeu: i'm strongly considering making nofp16 default for bifrost for a little while while things are in progress
<macc24> supertuxkart does the segmentation fault thing on my bifrost board
<macc24> no idea what it can be
<macc24> tried nofp16
TheMojoMan has joined #panfrost
TheMojoMan has quit [Client Quit]
stikonas_ has quit [Remote host closed the connection]
stikonas has joined #panfrost
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 244 seconds]
camus1 is now known as kaspter
<macc24> tomeu: on your fork i have just white screen on internal mipi/dsi display
<macc24> on my odroid go2
stikonas_ has joined #panfrost
stikonas has quit [Ping timeout: 272 seconds]
stikonas_ has quit [Ping timeout: 272 seconds]
stikonas has joined #panfrost
stepri01 has quit [Quit: leaving]
warpme_ has quit [Quit: Connection closed for inactivity]
davidlt_ has quit [Ping timeout: 240 seconds]
archetech has joined #panfrost
nlhowell has quit [Ping timeout: 272 seconds]
<raster> macc24: yup. thats a new kernel feature on newer kernels - at least from drm-next :)
<alyssa> ever improving
<macc24> raster: wow!
<macc24> i can't even thing of what's next
<macc24> maybe it won't even boot? ;)
<alyssa> macc24: I thought they already released that feature?
<alyssa> Maybe it hasn't hit mainline yet
<macc24> alyssa: oh i try to use mainline wherever i can, so maybe it didn't hit my debian derivative
<alyssa> Kernel upgrades are scary :(
<raster> macc24: i did start bisecting kernel but... that proved to not work well and had other thigns to do...
nlhowell has joined #panfrost
raster has quit [Quit: Gettin' stinky!]
<daniels> alyssa: more scary than writing a new driver from scratch with no specs? c'mon ...
<alyssa> daniels: Yes.
<alyssa> Easily, yes.
<alyssa> Source: have done both, still scared of one.
<daniels> if I can do it, you definitely can ;)
* macc24 likes being on the *bleeding* edge
<macc24> i try to make it so my go2 always runs newest kernel and mesa;)
* alyssa hugs her debian kernel
<alyssa> macc24: bloody hell?
<macc24> alyssa: i got scripts for building both, kernel from newest mainline release and mesa from git master ;)
<alyssa> I meant the bleeding edge
<alyssa> sounds like bloody hell 🙃
* alyssa doesn't even run mesa from master most of the time
<macc24> there is fine line between them
<alyssa> (I used to. But eventually I got bored of fixing the build in the middle of class.)
<macc24> oh, you don't run mesa master on your daily driver?
<macc24> just don't install it to /usr and you can fall back onto debian mesa 20.1.8
<alyssa> Both machines run mesa master, just not... today's master :)
<macc24> i build whenever i have 1 hour free time
<alyssa> I care about my machines working continuously too much to run bleeding edge stuff in production
<HdkR> As someone that hates kernel development. Upgrading kernel is infinitely more scary than RE driver writing :)
<macc24> well, by running newest kernel and newest mesa i can catch some bugs
<alyssa> macc24: yeah, that's admirable
<alyssa> stresses me out too much, I figure out I give back enough panfrost code I can get a
<alyssa> way with picking something a bit more stable :')
kaspter has quit [Ping timeout: 260 seconds]
<macc24> alyssa: oh, and the other reason why i run everything as new as posibble is totally not that it draws triangles faster
kaspter has joined #panfrost
<alyssa> well, yes :)
<alyssa> new code, new bugs.
<alyssa> but everyone knows it's impossible for Rust code to have bugs so let's rewrite 15 million lines of mesa in Rust /s
<HdkR> I'm sure we can make a clang based tool to help us translate the C to rust safely
<macc24> at least it's better than 3D driver in javascript
<alyssa> HdkR: no it won't compile since there are bugs in Panfrost but Rust can't have bugs
<alyssa> 🙃
<alyssa> macc24: Don't be ridiculous, all the cool kids use typescript.
<macc24> alyssa: it will auto-fix the bugs
<HdkR> yea, just pass in --auto-heal to the tool
<alyssa> oh, ok
<macc24> HdkR: but a rust to c compiler will reintroduce the bugs?
<HdkR> I had zero doubt that this was a real tool :P
<macc24> or will it make fixed C code?
<HdkR> Once you've fixed the C code, you can't just unfix it
<macc24> sed 's/a/e/g' will unfix it
<alyssa> macc24: Rust to C won't work
<alyssa> since there's undefined behaviour
<alyssa> so you can't do it
<alyssa> 🙃
<macc24> hold my beer
<alyssa> 🍺
<HdkR> 🤲
<macc24> what a nice reason to fix unicode
<chrisf> beer.. javascript.. ive heard some serious proposals to do 3d drivers in wasm
<alyssa> chrisf: hold my apple juice
remexre_ is now known as remexre
<alyssa> this is fine
* alyssa sips
raster has joined #panfrost
<Lyude> alyssa: exactly, code correctness doesn't exist. only memory safety.
<alyssa> Lyude: It doesn't matter if it's correct as long as it's fast ⏩
<Lyude> hehehehe
<alyssa> Fun fact: piping customer data to /dev/null is actually faster than writing to MongoDB.
<macc24> /dev/null is fastest hdd every
<macc24> ever*
<macc24> it's just write-only ;)
<HdkR> Lyude: Memory isn't safe when I'm doing a race
<HdkR> :P
<macc24> HdkR: supertuxkart?
<Lyude> HdkR: impossible, everyone knows your memory is guaranteed safe if you don't use pointer
* macc24 points at memory
<HdkR> haha
<HdkR> Just claim the variable is volatile, that'll resolve the problems
<alyssa> Lyude: I take it you read my mail? :p
<Lyude> alyssa: no, but I like making jokes about people making spit takes about things using C because they're not in rust lol
<Lyude> alyssa: ahhh, i like your thinking
<Lyude> seeing some rust code in mesa would be nice, it'd be a good influence on the rest of the community :)
<Lyude> and yeah, c++ kinda sucks
<alyssa> see? If an XDC board member agrees, we have to do it 😸
<HdkR> I'm a heavy C++ user and I still agree that rust is a god idea
<HdkR> good idea even. Gods have nothing to do with it
<alyssa> 👼
<alyssa> match
<HdkR> match two games?
icecream95 has joined #panfrost
archetech has quit [Quit: Konversation terminated!]
camus1 has joined #panfrost
kaspter has quit [Ping timeout: 264 seconds]
camus1 is now known as kaspter
raster has quit [Quit: Gettin' stinky!]
yann has quit [Ping timeout: 240 seconds]
<daniels> the best argument for Rust is that LLVM isn't written in it
<daniels> also tbf, unless you're doing some quite specific things, memory safety only just scrapes into the top 3 benefits of using Rust
<HdkR> So what you're saying is, if you're a rust only programmer, you get the benefit of never needing to touch LLVM? :P
<daniels> correct!
<HdkR> Sounds like a win. I should switch FEX over to Rust right now