alyssa changed the topic of #panfrost to: Panfrost - FLOSS Mali Midgard & Bifrost - - Logs - Transientification is terminating. Memory reductions in progress.
<hanetzer> wewt
<alyssa> Lyude: Hooray!
marcodiego has quit [Ping timeout: 268 seconds]
marcodiego has joined #panfrost
<alyssa> My scissor's upside-down!
<alyssa> ..among other things
stikonas has quit [Remote host closed the connection]
marcodiego has quit [Quit: Leaving]
<cyrozap> Lyude: What was the baud rate?
<cyrozap> I once had to write a script to set up a custom baudrate on an FTDI dongle to talk to some hardware I was poking around on (HDMI to DP adapter) and... Well, let's just say I learned a _lot_ about how serial ports worked on Linux, and I was really surprised that there isn't better tooling for using serial devices (or maybe it exists and it's just hard to find).
_whitelogger has joined #panfrost
<Lyude> cyrozap: yeah the tooling is bad and I've been thinking about that a lot
Elpaulo has joined #panfrost
<Lyude> cyrozap: basically it looks like conserver by default doesn't enable baud rates higher then 115200 due to compile time assertions, and errors out if you try to specify higher baud rates in your file (turns out my roc-rk3399-pc, aka "bismuth" uses a baud rate of 1500000)
<Lyude> except of course
<Lyude> if you are compiling for the classic cyclades TS console server from 2001
<Lyude> so I have some patches for that, but I want to see if goes through first before trying to file a pull request because ~conflicts~
_whitelogger has joined #panfrost
stikonas has joined #panfrost
raster has joined #panfrost
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 -]
MoeIcenowy has joined #panfrost
Moe_Icenowy has joined #panfrost
MoeIcenowy has quit [Ping timeout: 240 seconds]
Moe_Icenowy is now known as MoeIcenowy
Elpaulo1 has joined #panfrost
Elpaulo has quit [Ping timeout: 246 seconds]
Elpaulo1 is now known as Elpaulo
Elpaulo1 has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo1 is now known as Elpaulo
Elpaulo1 has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo1 is now known as Elpaulo
Elpaulo1 has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo1 is now known as Elpaulo
Elpaulo1 has joined #panfrost
Elpaulo has quit [Read error: Connection reset by peer]
Elpaulo1 is now known as Elpaulo
belgin has joined #panfrost
belgin has quit [Quit: leaving]
raster has quit [Remote host closed the connection]
rhyskidd has quit [Ping timeout: 240 seconds]
<Lyude> alright, time to see if I can get topaz, peridot and bismuth on fedora
<Lyude> cwabbott: btw ^ if I end up getting to peridot today (the hikey 960) you might not be able to use it today, but I'm going to try to keep the android installation somewhere so that it can be easily booted into from my server remotely
<Lyude> in case we need it for some more re
rhyskidd has joined #panfrost
rhyskidd has quit [Quit: rhyskidd]
<alyssa> Lyude: I think there are bifrost blobs floating around?
<Lyude> alyssa: there should be, it's more of a backup just in case
<alyssa> Lyude: Makes sense
<Lyude> hopefully I get some time to work on bf today as well
<alyssa> ^^
<Lyude> hanetzer: poke, you around? wondering if you might be able to answer some uboot questions w/ the rk3399
<hanetzer> Lyude: am, whassup?
<Lyude> hanetzer: do you know if it's possible to get uboot on say, an roc-rk3399-pc using just the mainline uboot repo?
<Lyude> I'm mostly curious if you really need the SDK that every single manufacturer seems to have their own version of just to get a uboot image on an sdcard
<hanetzer> unsure offhand, but you may need to add some glue code. what's it boot with right now?
<Lyude> hanetzer: right now I've just got one of their ubuntu lts images on an SD card so I could make sure the board worked
<Lyude> seemed to be normal uboot
<Lyude> even had efi support!
<Lyude> but now I'm trying to make my own fedora SD\
<alyssa> Lyude: I'd advise against tinkering with bootloaders for now, since it's going to take longer than just fighting whataver the default is and moving on to play with Bifrost :p
* hanetzer is about to try and port u-boot to kevin now that the bob framework is done :P
<hanetzer> though, apparently the cros-ec-keyboard does not work atm on it
<mmind00> Lyude: that board of yours seems to use lpddr4 for which open initcode is still wip for mainline uboot
<Lyude> alyssa: it's not really going to take much time on this system tbqh
<Lyude> or at least getting the bootloader setup
<Lyude> a lo9t of that time was just me getting familiar with how booting these systems is even supposed to work
<Lyude> mmind00: np then
<Lyude> i'll just build the uboot from their website since that looks to have everything I need already
<mmind00> Lyude: yep, but also still a binary 1st stage loader I guess ... but that really might be the shortest route for now
<Lyude> yeah, i'm not too worried about that
<Lyude> as long as it does uefi that's all I really need
<Lyude> (alyssa: also it should be noted the vim2 boots really weirdly compared to the other systems I've got here :p)
<Lyude> oh yay, this tool looks easy as heck
<alyssa> Lyude: That's how it starts...
<alyssa> Soon enough you're sending 1kloc patches to the u-boot lists...
<Lyude> alyssa: lol, i'm definitely not planning on developing uboot
<Lyude> just getting an image built that works
<alyssa> That's the problem, yes
<mmind00> alyssa: I don't think you need to worry much ... vendor uboot and mainline diverged quite a bit again, so Lyude cannot "accidentially" create uboot patches ;-)
<hanetzer> in related news, my veyron-speedy patches finally hit
raster has joined #panfrost
<alyssa> mmind00: There's always mainlining
<mmind00> alyssa: not with a lpddr4 board right now ... but yeah, there is the possibility of taking the wip lpddr4 code and trying to mainline it ... but hopefully Lyude won't get _that_ distracted :-D
<Lyude> again, not planning on working on uboot :p
<rtp> alyssa: by any chance, do you happen to have some t6xx panwrap logs somewhere ? I fail to generate some on my device
<Lyude> ( ^ things like that are why I'm trying to get my systems setup today :)
<alyssa> rtp: Really old ones, maybe
<rtp> alyssa: well, old logs are better than no logs :)
* alyssa fetches
<alyssa> Aside: wonder how hard it would be to backport an overlay for the ooold kernel I have here
<alyssa> So I can use the old panwrap to help y'all with porting
<alyssa> Having troubles getting complete logs, um
<alyssa> ^ That's a partial from test-triangle-smoothed run on panfrost (working version) in t6xx-compat-mode on 32-bit
<alyssa> Hope there's something salvageable in there
<alyssa> But... the headers have changed so much that that is almost unrecognizable unfortunately
<rtp> ok. will see if I find something interesting in it. thanks
<alyssa> Although tbqh the values look very similar. I'm betting I can work this out just by doing a wrap on a modern frost
<alyssa> (Did you have logs from that?)
<alyssa> HdkR: ^^ You got a panwrap on a modern Panfrost on a t6xx, yeah?
<rtp> the only logs I have are with panfrost. So far, I seem to be able to run mutter with mali blob but panwrap gives no log :(
<alyssa> rtp: Could you send the logs you got with Panfrost?
<alyssa> (The ones that yielded a DATA_INVALID_FAULT)
<rtp> I think I've uploaded some on the bug report
<alyssa> :eyes:
<alyssa> Promised myself I wouldn't dedicate too much time to frosting today, but I'm getting good vibes about this :p
<alyssa> Alright, got it
<alyssa> (the Panwrap log, that is)
<hanetzer> mmm, frosting
<Lyude> I think the most tedious and time consuming part of embedded development is the part where you drop a micro sd card and spend like 20 minutes trying to find it on the ground
<alyssa> Yupyup
<hanetzer> that or misplacing your debug cable
<alyssa> So nothing jumps out in the panwrap as obviously wrong, hrm
raster has quit [Read error: Connection reset by peer]
<alyssa> (I'm trying to get a sane dev environment setup on my rk3288)
<hanetzer> :P
<alyssa> Not helpful :P
<hanetzer> sure it is, insanity is the perfect sanity :P
* hanetzer needs to do the things to make gentu emerge all the things cross-ways onto a sdcard
<stikonas> gentoo mesa-9999 still has no panfrost :(
<hanetzer> eh?
<hanetzer> as in, ::gentoo ?
<stikonas> well, I mean ::gentoo
<hanetzer> I have an overlay with it
<stikonas> yes, not your ebuild
<stikonas> send a patch upstream :)
<hanetzer> I was like, I thought I fixed that
<alyssa> stikonas: Hm?
<stikonas> not mesa/mesa
<alyssa> Due to kernel shenanigans people should probably not be packaging panfrost at this point but.
<hanetzer> wait, did that get merged to mesa/mesa ?
<hanetzer> yeah
<alyssa> Yes, save for the overlay
<stikonas> yeah, maybe...
<stikonas> hanetzer: then fix overlay :D
<hanetzer> k. gimme a bit, I recently accidentally nuked my gentoo install so had to do again :P
<alyssa> Again, stuff related to the non-DRM kernel really shouldn't be packaged
<hanetzer> alyssa: its not meant to be packaged per se but to help 'install' it for dev purposes without using the prefix thing/slinging binaries into your root :)
<alyssa> Glmph
<hanetzer> wat?
<alyssa> "Alignment trap: not handling instruction...." Uhhh
<stikonas> well, those portage overlays are like PPAs...
<stikonas> without binary packages :)
<alyssa> I rest my case :P
<alyssa> Gah I forgot how many shenanigans are involved in bringing up a new machine
<alyssa> Oh, rtp fixed this, right
<stikonas> you can always automate some stuff :)
<alyssa> Oh, delightful
<stikonas> I'm trying to convert some of the manual setup into ansible scripts
<alyssa> rtp: I can repro your setup (mainlineish kernel, mali_kbase, latest Panfrost, 32-bit t6xx-style)
<alyssa> FRAGMENT jobs work so that's good (screen clears), VERTEX/TILER will need some work it looks like
<alyssa> But I'm in a good place to poke at this / help in any way I can
<memeka> Yay alyssa :)
<memeka> alyssa: btw both HdkR and myself got to job submission error
<memeka> I think this is past the invalid data error?
<alyssa> What? No, that's worse off :P
<alyssa> So, if I do vertex alone, it still has issues. That narrows it a little
<memeka> alyssa: hmmm how then to get past that?
<alyssa> ? Not sure how you *got* there :p
<memeka> Running 5.0 + mali kbase r28p0 + latest panfrost
<memeka> alyssa: last comment - that’s how
<memeka> alyssa: but if you push something or have some patches ready I can test :)
<memeka> Only if I’m not driving tho. I don’t test and drive
<alyssa> I'm thinking field sizes got messed up somewhere
<rtp> that or wrong padding or similar stuff
Elpaulo has quit [Ping timeout: 272 seconds]
TheCycoONE has quit [Quit: ZNC 1.7.2 -]
TheCycoONE has joined #panfrost
<Lyude> yessss
<Lyude> think I've figured out how to get bismuth booting now :)
<alyssa> Delightful
<hanetzer> so, are we aiming to use the same ioctl definitions within mali_kbase or is that free for rewriting purposes?
<Lyude> hanetzer: the ladder
<Lyude> afaik
* hanetzer pulls out the folding chair
<urjaman> ha
<hanetzer> someone watches WWE ne WWF
<hanetzer> I wonder if they got a new Undertaker
marcodiego has joined #panfrost
BenG83 has joined #panfrost