#
…

2022-10-12

00:00swampertx[m] has joined #asahi-gpu
00:04xorly[m] has joined #asahi-gpu
00:06bluetail0 has joined #asahi-gpu
00:11braindeadmosquito[m] has joined #asahi-gpu
00:17AndrewLee[m] has joined #asahi-gpu
00:17nulltemp[m] has joined #asahi-gpu
00:25handofstand[m] has joined #asahi-gpu
00:27ella-0[m] has joined #asahi-gpu
00:30AdityaJS[m] has joined #asahi-gpu
00:35Makusu[m] has joined #asahi-gpu
00:38Andy[m] has joined #asahi-gpu
00:39TellowKrinkle[m] has joined #asahi-gpu
00:42The_DarkFire_[m] has joined #asahi-gpu
00:49boulabiar[m] has joined #asahi-gpu
00:58catinahatisback[m] has joined #asahi-gpu
01:00DragoonAethis has quit [Quit: hej-hej!]
01:00DragoonAethis has joined #asahi-gpu
01:20dgarciagarcia[m] has joined #asahi-gpu
01:44TristenNollman[m] has joined #asahi-gpu
01:55ofir8989[m] has joined #asahi-gpu
02:03ryanmitts[m] has joined #asahi-gpu
02:04SuchirMishra[m] has joined #asahi-gpu
02:04Sobek[m] has joined #asahi-gpu
02:05ryanhrob[m] has joined #asahi-gpu
02:06pedrohos[m] has joined #asahi-gpu
02:08katatafjsh[m] has joined #asahi-gpu
02:12hpux735[m] has joined #asahi-gpu
02:20Kevy[m] has joined #asahi-gpu
02:30mofux[m] has joined #asahi-gpu
02:34madpro[m] has joined #asahi-gpu
02:39HuayuZhangLevi[m] has joined #asahi-gpu
02:45princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
02:46princesszoey has joined #asahi-gpu
02:49waytrue[m] has joined #asahi-gpu
02:54Mary has quit [Quit: The Lounge - https://thelounge.chat]
02:54Mary has joined #asahi-gpu
03:02threerik[m] has joined #asahi-gpu
03:05icepie[m] has joined #asahi-gpu
03:07pashencija[m] has joined #asahi-gpu
03:22sunyiynus[m] has joined #asahi-gpu
03:24ognju[m] has joined #asahi-gpu
03:32defolos[m] has joined #asahi-gpu
03:32pazyleon[m] has joined #asahi-gpu
03:32pg12_ has quit []
03:33pg12 has joined #asahi-gpu
03:44princesszoey5 has joined #asahi-gpu
03:50kl22[m] has joined #asahi-gpu
03:51l1npengtul[m] has joined #asahi-gpu
03:53princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
03:53princesszoey5 is now known as princesszoey
03:54Soroush has joined #asahi-gpu
03:58princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
03:59princesszoey has joined #asahi-gpu
04:12XProtocol[m] has joined #asahi-gpu
04:20daftfrog[m] has joined #asahi-gpu
04:20nehbyua[m] has joined #asahi-gpu
04:21firox263[m] has joined #asahi-gpu
04:26larabee[m] has joined #asahi-gpu
04:33HauXs[m] has joined #asahi-gpu
04:33badlydrawnface[m] has joined #asahi-gpu
04:34polarn[m] has joined #asahi-gpu
04:38zhxchen17[m] has joined #asahi-gpu
04:59balrog has quit [Ping timeout: 480 seconds]
05:01sajattack[m] has joined #asahi-gpu
05:13bluetail03 has joined #asahi-gpu
05:15latosca[m] has joined #asahi-gpu
05:21bluetail0 has quit [Ping timeout: 480 seconds]
05:21bluetail03 is now known as bluetail0
05:27Tsic[m] has joined #asahi-gpu
05:28dantedrac[m] has joined #asahi-gpu
05:31espo has quit [Ping timeout: 480 seconds]
05:33skoobasteeve has quit [Ping timeout: 480 seconds]
06:23balrog has joined #asahi-gpu
06:37princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
06:38princesszoey has joined #asahi-gpu
06:39pjakobsson has joined #asahi-gpu
06:49princesszoey has quit [Read error: Connection reset by peer]
06:50princesszoey has joined #asahi-gpu
06:59SSJ_GZ has joined #asahi-gpu
07:00princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
07:02princesszoey has joined #asahi-gpu
07:15goldsoultheory has joined #asahi-gpu
07:44princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
07:45princesszoey has joined #asahi-gpu
08:19princesszoey has quit [Read error: Connection reset by peer]
08:30Major_Biscuit has joined #asahi-gpu
08:32cr1901_ has joined #asahi-gpu
08:33princesszoey has joined #asahi-gpu
08:35cr1901_ has quit [Read error: Connection reset by peer]
08:36princesszoey has quit [Remote host closed the connection]
08:40DarkShadow44 has quit [Quit: ZNC - https://znc.in]
08:40cr1901 has quit [Ping timeout: 480 seconds]
08:44DarkShadow44 has joined #asahi-gpu
08:45princesszoey has joined #asahi-gpu
08:45DarkShadow44 has quit []
08:47DarkShadow44 has joined #asahi-gpu
09:02princesszoey has quit [Remote host closed the connection]
09:07princesszoey has joined #asahi-gpu
10:34princesszoey has quit [Quit: The Lounge - https://thelounge.chat]
10:35princesszoey has joined #asahi-gpu
10:40kov has quit [Quit: Coyote finally caught me]
10:43bluetail0 has quit [Ping timeout: 480 seconds]
10:50ferrypie[m] has joined #asahi-gpu
11:17kov has joined #asahi-gpu
11:23vup has joined #asahi-gpu
11:27mkurz has quit [Quit: Leaving]
12:13bluetail0 has joined #asahi-gpu
15:03cr1901 has joined #asahi-gpu
15:22yamii_ has joined #asahi-gpu
15:22yamii has quit [Read error: Connection reset by peer]
15:38yamii_ has quit [Ping timeout: 480 seconds]
15:48yamii_ has joined #asahi-gpu
15:55goldsoultheory has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
15:56jlco_ has joined #asahi-gpu
15:58jlco has quit [Ping timeout: 480 seconds]
16:24goldsoultheory has joined #asahi-gpu
16:36jannaulina: is the dcp issue my or your problem? just started watching the last ~20 minutes
16:38Major_Biscuit has quit [Ping timeout: 480 seconds]
16:39linaI think it's your problem, but it's not a blocker for what I'm trying to do ^^
16:39alyssawhat's the issue?
16:41zzywysm has quit [Quit: Textual IRC Client: www.textualapp.com]
16:42linaglmark2 makes it go explodey
16:42alyssaoh, yeah, ok
16:42jannauI didn't stop the video, I spotted an atomic_disable
16:43alyssajannau: even without agx in the loop, closing compositors (eg quitting sway to return to the tty) crashes dcp... I don't think that happened on my original prototype but IDK what would've changed. I think you mentioned you know about this issue?
16:43alyssamight that be the same bug as glmark2 is hitting?
16:43jannaualyssa: I think I've fixed that
16:44alyssai haven't uprevved in a while, will need to for agx though ;)
16:44jannauand lina should have the fix
16:46jannauit makes sense to wait at least a few days, we're going to merge iommu mapping handling in m1n1 soon
16:48jannauthat unfortunately requires devicetree changes compared to what lina merged so it might take a few days until everything is sorted out
16:50alyssai've got macOS side work to do, not actually looking forward to fixing glamor anyway :p
16:52jannauI might have a workaround for the most glaring glamor bug. I hooked the cursor plane up and copy its framebuffer if it's too much off-screen
16:53alyssaI am missing a lot of context I think
16:53jannaunot sure if I should publish that though
16:53alyssaI assumed glamor was blocked on mesa bugs
16:53alyssaWhat do cursor planes have to do with it?
16:53alyssaor is there a bug drawing the cursor in GL and using a plane offloads from AGX to DCP to workaround?
16:54alyssa(but of course cursors are differently broken in DCP hence the copy?)
16:54jannauthe cursor triggered the most glaring glamor rendering bugs
16:54alyssadelightful
17:02linaI think it was drawing 512x512 squares for the cursor or something like that ^^
17:03alyssawhat a delight
17:04boulabiar[m]So did you get over the gpu restart each frame issue today?
17:05linaNow it crashes even more! (Which is probably a good thing for debugging)
17:06goldsoultheory has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
17:07goldsoultheory has joined #asahi-gpu
17:08goldsoultheory has quit []
17:18goldsoultheory has joined #asahi-gpu
17:24eric_engestromlina: I just randomly saw ARM64_ERRATUM_2441009 "Cortex-A510: Completion of affected memory accesses might not be guaranteed by completion of a TLBI"; might be relevant to your TLB issue?
17:24eric_engestrom(in kernel config)
17:25eric_engestrom(sorry if this is irrelevant, and/or if you've already seen it :)
17:25linaI doubt Apple's cores have Cortex errata... ^^;
17:25eric_engestromno, but perhaps the same kind of thing applies?
17:25eric_engestromI don't really know anything about all this so I'm really just guessing
17:26linaI actually don't think I have a TLB issue at this point, at least not with the GPU. As far as I can tell from the m1n1 experiment, TLBI actually works fine
17:27linaI might still have a TLB issue with the coprocessor, which was being hidden by the GPU shutdown thing beacuse I think the coprocessor shuts down shortly after the GPU (so it mostly worked in practice)
17:27linaBut there's something else wonky going on with the GPU. If I force it to stay on on purpose, sometimes it crashes on the second frame, sometimes a bit later.
17:28linaFor the coprocessor I'm just never unmapping/freeing anything for now, to eliminate variables (besides, I plan to do that in my allocator later on anyway, there's no good reason to thrash those mappings)
17:51chipxxx has joined #asahi-gpu
17:51chipxxx has quit [Remote host closed the connection]
17:51chipxxx has joined #asahi-gpu
18:23DarkShadow44 has quit [Quit: ZNC - https://znc.in]
18:23DarkShadow44 has joined #asahi-gpu
18:24jannauset_digital_out_mode(color=-1, timing=43) doesn't look healthy (from dcp error from the stream at 10:04:29)
18:25jannaudrm complains that the driver didn't call drm_crtc_vblank_off
18:29DarkShadow4444 has joined #asahi-gpu
18:32DarkShadow44 has quit [Ping timeout: 480 seconds]
18:59alyssajannau: ...no, that does not indeed.
18:59alyssamaybe we need a WARN_ON(mode < 0) in there
18:59alyssaor more precisely WARN_ON(IS_ERR(mode))
19:00alyssa(I don't remember how the parser worked ..)
19:02* alyssareads the code
19:02alyssathis thing really wants to be rust doesn't it.
19:02jannauI don't think it's the parser's fault. this happens when starting glmark2-es2-drm after it already booted
19:03svenDCP needs less WARN_ON and nit
19:03sven*not more
19:03svenwarn_on spams dmesg and most of the time the stack isn’t all that useful because it will always be the same
19:04jannauI guess glmark selects an an unusual mode with the capture card and dcp doesn't verify it
19:04alyssasven: what is the right way to do asserts then
19:04alyssaasserts are the best tool we have to write correct code
19:04svendev_err or something
19:04alyssaI thought kernel C assert() is just WARN_ON(!)
19:04sventhere are also people who run kernels with panic on warn
19:05alyssaThat is the right behaviour yes
19:05svenso if DCP does something slightly unexpected you just panicked the entire kernel
19:05alyssaAn assertion failure means a bug in my code.
19:05alyssaNOT a bug in the firmware or the device or the userspace
19:05sveniirc there are at least a few warns inside DCP that are more like "dcp did something strange"
19:05alyssaThose should all be errors that become dev_err and should never be asserted it
19:06alyssaOkay, well, that's wrong, complain to Alyssa about that.
19:06svenmaybe i'm wrong, i just skipped over the code fwiw
19:06alyssano i'm looking now
19:06svenbut some of them looked like they want to be dev_err instead
19:06alyssaI see a bunch of legitimate asserions and a bunch of bogus ones that should indeed be dev_err, bah
19:06alyssastatic u8 dcp_pop_depth(u8 *depth)
19:06alyssa{ WARN_ON((*depth) == 0);
19:07alyssathat is a legitimate assertion, it can only ever fail due to a bug in the dcp kernel driver
19:07alyssa if (WARN(dcp_channel_busy(&dcp->ch_cmd), "unexpected busy channel") ||
19:07alyssa WARN(!dcp->connector->connected, "can't flush if disconnected")) {
19:07alyssathese are bogus assertions that should be dev_err instead
19:07alyssa WARN_ON(endpoint != DCP_ENDPOINT);
19:08alyssathis one is bogus
19:09alyssaactual bug though: if no modes match, the mode parser returns 0 (success) with best mode = -1
19:10alyssabut dcpep.c assumes 0 means there's a valid mode
19:10alyssaone of those needs to change, idk what colour the bikeshed should be
19:10jannauI'll fix it and look at the WARN_ONs
19:11jannaudcpep.c is now iomfb.c btw
19:12alyssaseems legit
19:12jannaunot yet what is used in the agx tree
19:12svenand i have some bastard tree forked off a few commits ago that implement yet another dumb parser for these EPIC things :(
19:13svenlet me see if I can at least still rebase
19:13alyssais dcp getting RiiR'd or not I can't remember
19:14svenlol, i also have a commit in my tree to move some dev_dbg to tracepoints
19:14alyssaI'd rather it weren't but otoh maybe the mode parser should be.
19:14svenoh well..
19:15jannaulol, with a clash in the filename? I looked at tps6598x as template
19:15jannaure-implemented in rust
19:15svenyup, because I also used tps6598x as template :D
19:16svenmeh, i'll take care of that later this week
19:16alyssaRiiR is a very important algorithm for being friends with lina~ ;-p
19:19svenah, don't really care. if you want to have dcp rust I'll gladly learn it to implement the EPIC crap there *shrug*
19:21jannauI won't mind Rust and depending on how much Apple changed the DCP rpc in ventura I think it could be useful
19:21jannauon the other hand judging by the amount of changes they did between macos 11 and 12 it should be manageable in C
19:28* alyssashrugs
19:28alyssayou're the dcp maintainer now, it's your call :p
19:29svennerd sniping successful :>
19:55chipxxx has quit [Read error: Connection reset by peer]
19:57chipxxx has joined #asahi-gpu
19:59chipxxx has quit [Read error: Connection reset by peer]
21:01goldsoultheory has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
21:09goldsoultheory has joined #asahi-gpu
21:23mactroon65[m] has joined #asahi-gpu
21:23mactroon65[m] was banned on #asahi-gpu by ChanServ [*!*@2001:470:1af1:101::f0c2]
21:23mactroon65[m] was kicked from #asahi-gpu by ChanServ [You are not permitted on this channel]
21:25mactroon66[m] has joined #asahi-gpu
21:25mactroon66[m]hello trannies
21:26mactroon66[m]have you dilated your stinkditches today yet
21:29mactroon66[m]rust is the biggest disgrace that happened to linux kernl
21:29mactroon66[m]they should have adopted c++ instead of the tranny sjw language
21:29mactroon66[m] was kicked from #asahi-gpu by ChanServ [You are not permitted on this channel]
21:36alyssahello fruit farmer
21:36alyssaquery sven
21:40jannaulina: https://github.com/jannau/linux/commit/60babbde481e945a979b2c1c368d93d7cbdce1f5 should fix the dcp issue. newer tree than what you merged but that commit should be cherry-pickable
21:40jannaustarnge that dcp reports timing modes without color modes
21:41jannaunot sure why glmark2 hits that. For some reason it must chose a different mode than all other software
21:43jannaumaybe it tries to select the mode with the highest refresh rate and you capture card additional unusual modes besides the 64Hz FullHD mode
21:43alyssathat sounds dcp
22:02goldsoultheory has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
22:09SSJ_GZ has quit [Ping timeout: 480 seconds]
23:17bisko has quit [Ping timeout: 480 seconds]
…