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
<macc24> yes
<macc24> i call this "fix by yeet"
<macc24> gnome 3 is underrated tablet ui
<robmur01> OK, but you having made OPPs usable on your machine hardly disproves that the CI machines might be flaking because *they* aren't at a valid OPP
<macc24> robmur01: bruh just use my patched in ci ;)
<macc24> or add opp that would match boot clock
<robmur01> I'd be curious to see what the default regulator voltages are without any hacks to disturb them
<robmur01> if they *do* look high enough then that's unlikely to be the issue and I can stop speculating
<macc24> as a hack you can just try lowering the clock and not touching regulator
<robmur01> Sure, or one could do the trivially easy thing and gather evidence before changing things at random based on wild guesses
<robmur01> believe me, if I could SSH into Collabora's LAVA farm and cat a thing from debugfs I would have done so already ;)
<robmur01> but I work for the wrong employer so all I have are useless piles of documentation, Mali RTL and DDK source code :P
<macc24> *shrugs*
<macc24> "[918.260979 power state 7 = S0->S3, in 0x0003]" after this it resumed perfectly fine... so guess suspend is working perfectly on duet
<icecream95> macc24: Suspend is working somewhat reliably on 5.10
<macc24> icecream95: i mean with display
<icecream95> macc24: It works if I select "Suspend" from the GDM menu
* macc24 shrugs
camus has joined #panfrost
camus1 has quit [Remote host closed the connection]
<alyssa> robmur01: O:)
<alyssa> robmur01: "Any hint of what the magic numbers mean?"
<alyssa> "identity remapping"
<alyssa> 0f0e0d0c 0b0a0908 07060504 03020100
<alyssa> Does this help? :)
<macc24> Numbers? Those are strictly forbidden in Ponyville!
<alyssa> Talk about eye candy...
atler is now known as Guest716
atler has joined #panfrost
Guest716 has quit [Ping timeout: 480 seconds]
pendingchaos has quit [Ping timeout: 480 seconds]
pendingchaos has joined #panfrost
camus has quit [Remote host closed the connection]
camus has joined #panfrost
<icecream95> alyssa: "\text{src} \stackrel{?}{=} 0 \\ PC + 8 &amp; \text{otherwise} \end{cases}$$". Thanks, that really cleared things up for me. /s
<icecream95> Speaking of BRANCHZ, I note that on Bifrost there are several "reserved" cases that the blob occasionally emits
* macc24 misspelled 'ucm' as 'cum' more times than she likes to admit
<alyssa> icecream95: https://rosenzweig.io/a.pdf
<alyssa> Yeah.. the general BRANCH encoding on Bifrost is one of the most ridiculous parts of the ISA, and the complexity is almost entirely unwarranted ...
<macc24> bifrost is the x86 of arm
<alyssa> IIRC handling it correctly would require special casing BRANCH in the disassembler
<alyssa> The encodings aren't actually reserved (duh), and they should all be in ISA.xml
<alyssa> Anyway, I never got around to fusing conditions into branches in the bifrost compiler, and I spend very little time looking at DDK-generated Bifrost assembly these days .. so it's been a known bug but out-of-sight, out-of-mind
ids1024 has quit [Ping timeout: 480 seconds]
macc24 has quit [Read error: No route to host]
macc24 has joined #panfrost
<macc24> my computer itself started procrastinating and crashed
ids1024 has joined #panfrost
<macc24> umm
<macc24> supertuxkart broken again ,_,
<macc24> on panfrost from mesa main, with advanced rendering pipeline thing enabled
<macc24> can anyone reproduce?
<macc24> on g72
davidlt has joined #panfrost
* alyssa sighs
vstehle has quit [Ping timeout: 480 seconds]
davidlt has quit [Ping timeout: 480 seconds]
vstehle has joined #panfrost
pendingchaos has quit []
warpme_ has joined #panfrost
pendingchaos has joined #panfrost
pendingchaos has quit [Read error: No route to host]
rasterman has joined #panfrost
<icecream95> macc24: There were some fixes for SuperTuxKart which were merged yesterday, try pulling again?
camus has quit [Remote host closed the connection]
camus has joined #panfrost
floof58 has quit [Ping timeout: 480 seconds]
floof58 has joined #panfrost
floof58_ has joined #panfrost
floof58 has quit [Ping timeout: 480 seconds]
<tomeu> robmur01: will look at getting those numbers
pendingchaos has joined #panfrost
<tomeu> tell me if you want an account in our lava lab to do these things yourself :)
camus has quit [Read error: Connection reset by peer]
camus has joined #panfrost
floof58_ has quit [Ping timeout: 480 seconds]
<robmur01> heh, I've got more than enough to be busy with already
<robmur01> but thanks for digging that up - looks like those voltages are high enough for all but max clock freq, so if it's staying stuck at ~512MHz then it's probably not an undervoltage issue
floof58_ has joined #panfrost
<tomeu> seems to be the case, from clk_summary:
<tomeu> mfgpll 1 1 0 511999970 0 0 50000 Y
davidlt has joined #panfrost
floof58_ has quit [Ping timeout: 480 seconds]
floof58_ has joined #panfrost
camus1 has joined #panfrost
camus has quit [Ping timeout: 480 seconds]
br has quit []
br has joined #panfrost
floof58_ has quit [Ping timeout: 480 seconds]
floof58_ has joined #panfrost
hanetzer1 has joined #panfrost
hanetzer has quit [Ping timeout: 480 seconds]
hanetzer1 is now known as hanetzer
<macc24> icecream95: uh i just installed supertuxkart from debian repos
jernej_ has joined #panfrost
jernej has quit [Remote host closed the connection]
robmur01_ has joined #panfrost
robmur01 has quit [Ping timeout: 480 seconds]
robmur01_ is now known as robmur01
<robmur01> about 4 levels deep into recursive procrastination, I seem to have done this:
<robmur01> I'm not entirely sure why :/
<alyssa> robmur01: incredinle
<robmur01> no idea if it works, I just have this weird curiosity for devfreq...
<alyssa> it is pretty frequing cool
* robmur01 groans and leaves the room
<daniels> robmur01: what do you need cat'ed?
<daniels> ah, scrollback just caught up with tomeu already sorting it
<macc24> robmur01: hippity hoppity your code is my property
<HdkR> Looks like macc24 is going to upstream it
* macc24 was about to ask robmur01 if they're going to
<alyssa> macc24: is this witchcraft
<macc24> alyssa: everything mediatek is witchcraft
<alyssa> buy some apples
<macc24> cant
warpme_ has quit [Quit: Connection closed for inactivity]
warpme_ has joined #panfrost
jernej_ is now known as jernej
Danct12 has quit [Remote host closed the connection]
Danct12 has joined #panfrost
rasterman has quit [Quit: Gettin' stinky!]
<alyssa> tomeu: I just pushed through CI a Mesa tree from early May, right before G72 CI was disabled
<alyssa> It faults in the same way
<alyssa> 2021-07-14T19:55:06 [ 44.031490] panfrost 13040000.gpu: gpu sched timeout, js=0, config=0x7302, status=0x8, head=0x310c640, tail=0x310c640, sched_job=000000000d9c4f99
<alyssa> s/faults/timeouts/
<alyssa> Only material difference is that PAN_MESA_DEBUG=sync didn't exist back then so the timeout were silent
<alyssa> (Well. There was a PAN_MESA_DEBUG=sync but it doesn't mean what it does now.)
<alyssa> Without PAN_MESA_DEBUG=sync, the job passes but with flakes caught by deqp-runner
<alyssa> So AFAICT this isn't exactly a G72 regression, as much as it never working.
<macc24> and now the only things that are broken on duet are usb-c display, hardware video decoding and cameras
<alyssa> what's wrong with display + cameras?
<macc24> alyssa: it6505 driver is still not in mainline, and i'm not even sure if dpi works
<macc24> and there is literally nothing for mipi cameras on mt8183 in mainline kernel
<alyssa> macc24: Thanks for volunteering to help mainline those drivers
<macc24> alyssa: bruh i am not doing anything besides tracking down the patches and gluing distros to kernels
<alyssa> don't @ me
<macc24> also it's not really "volunteering"
<alyssa> Bifrost seems like a parody of Valhall.
<alyssa> No, scratch that
<alyssa> Bifrost seems like Midgard doing a parody of Valhall.
<macc24> bifrost seems like parody of a gpu
<macc24> alyssa: does this mean that valhall will be quicker supported in panfrost?
davidlt has quit [Remote host closed the connection]
davidlt has joined #panfrost
<alyssa> In principle yes
<alyssa> In practice, TBD
<alyssa> Valhall is technically much simpler than Bifrost, but there are non-technical factors in play.
<alyssa> Of course I'm a fundamentally technical person so I can't say I pay much attention to the latter :-p
<macc24> hmmmmmmmmmmmm
davidlt has quit [Ping timeout: 480 seconds]
bluebugs has quit [Quit: Leaving]
* icecream95 has spent far too much time trying to find the demo of Gallium Nine running natively without wine again...
<icecream95> (For next time I'm looking for it: https://github.com/axeldavy/Xnine )
camus has joined #panfrost
camus1 has quit [Ping timeout: 480 seconds]