2022-03-14
02:04kgarrington has joined #asahi-dev
02:05kgarrington has quit [Remote host closed the connection]
02:31PaterTemporalis has quit [Ping timeout: 480 seconds]
02:35RevHelix has joined #asahi-dev
02:59riker77_ has joined #asahi-dev
03:04riker77 has quit [Ping timeout: 480 seconds]
03:04riker77_ is now known as riker77
03:20PhilippvK has joined #asahi-dev
03:23phiologe has quit [Ping timeout: 480 seconds]
03:25kov has quit [Quit: Coyote finally caught me]
03:25kov has joined #asahi-dev
04:08chadmeddamn so you cant actually hide devices with pipewire/wireplumber, only disable them completely
04:28bluetail[m]I wish Linux audio was less of a hassle
04:42chadmedand the worst part is it turns out that wireplumber was the problem causing the filter chain to not route automatically to the output, so ive had to fall back on good ol pipewire-media-session
05:36chadmedpovik, marcan, anyone else brave enough to test their j31x speakers: https://drive.google.com/file/d/1r--1iLggAO2JII-5UFJYw7_fTrD5dvKl/view?usp=sharing
05:37chadmedim pretty happy with this for a release tbh, only thing im still not happy with is that the bass completely disappears at low volumes, but we'd need some sort of dynamic loudness/expander filter to do that and i've no faith that pipewire could do that properly
05:46rabbitz has joined #asahi-dev
05:48* rabbitzhttps://docs.google.com/document/d/16xzfOLHkuaKWL0yg2evzdJLXKjuN8Gi_/edit?usp=sharing&ouid=100976977371543538488&rtpof=true&sd=true , 0day book on going , any suggestion ? ... .
05:50* rabbitzhttps://docs.google.com/document/d/16xzfOLHkuaKWL0yg2evzdJLXKjuN8Gi_/edit?usp=sharing&ouid=100976977371543538488&rtpof=true&sd=true any suggestion ?
05:53rabbitz has quit []
05:56gladiac has joined #asahi-dev
06:13rabbitz has joined #asahi-dev
06:14* rabbitzany suggestion https://docs.google.com/document/d/1T5rOhHGQvTlfj7zMHWImHALQ9hJPlyTS/edit?usp=sharing&ouid=100976977371543538488&rtpof=true&sd=true , Teaser Finished ... .
06:32rabbitz was kicked from #asahi-dev by ChanServ [You are not permitted on this channel]
06:36rabbitz has quit []
06:49hizon has joined #asahi-dev
06:52joske has joined #asahi-dev
06:53PaterTemporalis has joined #asahi-dev
07:04kita99 has joined #asahi-dev
07:09PaterTemporalis has quit [Ping timeout: 480 seconds]
07:26PaterTemporalis has joined #asahi-dev
07:29kita99 has quit [Ping timeout: 480 seconds]
07:52MajorBiscuit has joined #asahi-dev
08:05ChaosPrincess has joined #asahi-dev
08:13jannauno, m1n1 is expected to fix that. We need it for cases where the hypervisor doesn't run with all CPUs
08:14j`eyok then I think the affinities issue will also affect the 8core version, once the dtb has that in
08:16DarkShadow4444 has joined #asahi-dev
08:16DarkShadow44 has quit [Read error: Connection reset by peer]
08:23marcanI still don't know why we need to *remove* CPU nodes but apparently disabling them doesn't work :-(
08:27j`eyI started to make an attempt here but it isnt finished or tested or compilable yet https://termbin.com/upul
08:50PaterTemporalis has quit [Ping timeout: 480 seconds]
08:55chadmednow that i have the workflow down and have wrapped my head around playing with alsa and pipewire, i could probably manage a turnaround of t+1d from the kernel driver being made available if i had access to the machines
08:56Chol has joined #asahi-dev
08:58CholDid any of the devs see this? https://lore.kernel.org/lkml/32EA0FE1-5254-4A41-B684-AA2DEC021110@adelielinux.org/
09:07the_lanetly_052 has joined #asahi-dev
09:24chadmedi was devastated when i first learnt that like 90% of what makes tiny speakers like this sound good is the DSP in the background
09:25chadmednot because its a bad thing or whatever, just because i thought it was always super impressive that they could "make" such small speakers that put out a decent sound. turns out they cant! :P
09:27bluetail[m]sometimes those solutions are sufficient. especially when usually you use headphones
09:28joske has quit [Remote host closed the connection]
09:28chadmedmy background in terms of audio stuff is coming from professional recording gear and hifi stuff where the engineering is in the physical hardware, and your goal is to apply as little DSP as is absolutely neccessary to correct for the room's characteristics (if at all)
09:29chadmedso its been interesting seeing the other side of it, where its used as basically a cost optimisation strategy
09:29mpssorry for clueless question from someone who don't know much about modern audio, but why adding filters or equalizers to kernel driver, isn't this job of userspace
09:31kameks has joined #asahi-dev
09:31Chol has quit [Quit: Leaving]
09:32chadmedwell its a bit of both. modern loudspeakers are usually made of pretty exotic materials and have some decently advanced passive (sometimes even active) electronics in them to control things like crossover frequencies and such
09:33chadmedif you can reliably do all that in software, you can use a very basic paper/cardboard cone, ditch the LC/RLC filters
09:34chadmedit adds up when youre a company like QSC or Yamaha or Peavey and producing thousands of units a month
09:37chadmedsmaller speaker = smaller cost, not to mention you can use jellybean parts from whatever factory is next door to foxconn instead of trying to source actually good speakers
09:47chadmedoh good small cameras havent been around long enough for me to be naive enough to be fooled by them
09:48marcanthe iLoud Micro Monitors sound better than quite a few significantely larger (uncalibrated) monitors thanks to the onboard DSP
09:49marcan(the disadvantage is they're extremely directional as a result, but if you *are* in the sweet spot, they kick ass)
09:50chadmedoh yeah i remember when we added a DSP unit into our FOH rack. until we got a behringer x32 in, that thing saved my bacon on multiple productions
09:51chadmedkind of became redundant when i could do it all from the desk and save it as a preset but hey, it was nice for about a year
09:51marcanyeah, at the end of the day, it comes down to there not being any reason not do use DSP in this day and age
09:52marcanyou want to optimize the drivers and enclosures mostly for spatial uniformity (or not, depending on use case) and lack of notches/problems in the frequency response
09:53marcanthe main thing DSP can't fix is spatial issues, since you only have so many channels anyway
09:53chadmedyep exactly, even most good high end hifi receivers do it by default now, and turning it it off is an extra thing you have to press a button for
09:55marcanbut my first run-in with this stuff was in a *point and shoot*, Canon S100, which I bought a good 9 years ago now
09:55marcanraw output (this is before libdcraw supported it properly so it ended up B&W, ignore that): https://marcan.st/transf/s100-raw.jpg
09:56marcanturns out Canon also figured out it's better to give up on making the lens rectilinear and just cram more zoom in there, and fix it in software
09:56marcan(OTOH it's fun just how much more you can see around the corners in the uncorrected version!)
09:57chadmedi have nothing but support for any DSP that is put to work on making sure mio honda is presented in the best possible light
09:58chadmedthats interesting that it was in such an old point and shoot though, my first run in was with a blackmagic pocket cinema in 2016
09:58marcanactually I think webcams have barely gotten in that game, if at all? mostly because they all kinda suck
10:00marcanmost webcams are a *really* dumb USB bridge chip and a jellybean sensor, they can't do much more than color correction and sharpness/trivial convolution
10:00marcan(webcams also don't have the same depth constraints as smartphones so they can also get away with much dumber lenses)
10:03marcanspeaking of, I wonder what the ISP interface to the cameras in the M1s is like... probably also just a very high level thing
10:04marcankinda funny that Apple didn't throw another Chinook in there to deal with speaker DSP to be honest, it's the kind of thing they'd do
10:10bluetail[m]Had the pleasure of owning the Neumann KH80. Even with my treated room I wasn’t well enough prepared. Probably they are so professional that you need to be an audio engineer and spend tons in room optimization before
10:11bluetail[m]that said I prefer the tannoy live mini to those 1200 eur speakers
10:11chadmedinterestingly none of their machines have done it that way. im sure if it occurred to them to do it, they would have but afaik macos has handled audio DSP in some fashion or another for ages so the infrastructure was all there
10:11chadmedyou dont need to spend tons on any room optimisation. especially if its just a room in your home, something as simple as a rug or thick curtains can make an enormous difference
10:13marcanI actually guessed the T2 macs might do it in the T2? if nothing else so Windows doesn't have to deal with it
10:13bluetail[m]but room corners… I spent 3 grand and my room is so much quieter. I love it. Yea. Carpet, ceiling absorbers and ikea billy
10:14bluetail[m]also it helps if ur bed is in the same room and not much glass
10:15bluetail[m]use big plants before tall windows to disperse sound waves. Doesn’t always have to be a thick wall of wood art
10:15marcanmy studio/bedroom has a few nasty resonances in certain spots from parallel walls, but the weird bed corner helps make it not too bad; I think a few strategically placed foam panels will take care of it
10:16bluetail[m]my problem is 50hz
10:16bluetail[m]but I think it might be coming from another room… idk
10:16bluetail[m]theres uneven wall filling
10:17chadmedcould be a ground loop too. i have a shortwave radio that has such a cheap and nasty mains transformer in it that the whole unit vibrates at 50hz so i just run it off batteries
10:20joske has joined #asahi-dev
10:21marcanI have a problem with elevator VFDs, which I've experienced in *two* apartment complexes already
10:21marcanI don't know if it's normal for them to be so nasty, but I can hear a constant high pitched whine whenever the elevator is powered on, it gets into *everything* audio
10:22marcanyou can even touch the connector on bare headphones to a ground terminal or so and hear it unamplified
10:23marcanat least it seems to spare balanced connections, so mics survive (and it's also trivial to notch out if it comes down to that)
10:23marcaneven with the input disconnected or shorted, it gets into the internal amplifier stages :(
10:24matthewayers[m]chadmed: what part of the world are you located in?
10:25matthewayers[m]chadmed: Nice! Where I am it’s 6:25 AM
10:25marcan"let's make a differential digital connection that isn't actually differential! and who needs a dedicated drain wire for the data path, just share the power ground!"
10:28chadmedremember when early usb3 devices would just tank their connections because somehow usb-if ended up certifying non-compliant devices anyway
10:28matthewayers[m]Why is USB such a horrible standard?
10:28bluetail[m]Don’t manufacturer’s use filters to make those problems go away? Owning an rme adi 2 sac fs headphone amp+dad
10:29bluetail[m]usb 3.2 gen2 is actually usb 3.1
10:29chadmedyeah lmao, wasnt the hack for that to open the thing up and ground the outside of the connector with a bodge wire?
10:30chadmedi wonder if its even at all physically possible for the PC industry to do something properly ever
10:31kloenk has joined #asahi-dev
10:32bluetail[m]What about galvanic isolation? Isn’t that how dehummers work? Using now 4 instead of 2 cables, making the connection balanced and canceling stuff out
10:32marcanyou can certainly galvanically isolate the analog side after a USB audio device and that will fix the problem, of course
10:33bluetail[m]how do I know my audio is kaputt cause using usb ?
10:33bluetail[m]I never had that issue with rme
10:34marcanpovik: because it's only balanced *during* data transmission - between transmissions it goes single ended
10:34joske has quit [Remote host closed the connection]
10:35bluetail[m]We should only use spdif /optical
10:35povikright, but for EMI considerations, that current should occur only at start/stop transitions, right?
10:37marcananother fun thing is that USB is broadcast-downstream, unicast-upstream, so you get data-dependent noise even going to devices which aren't involved
10:37marcane.g. have a USB audio interface and a hard disk on the same hub? the USB audio interface's USB cable is carrying currents that are a function of your disk reads/writes
10:39chadmedwhen they designed it, they were relying on interference from GSM to drown out any interference caused by usb :P
10:40marcanback in the day I could point my iPhone 2G (yes I had one at the time...) at my Yamaha keyboard at just the right angle from across the room, and clearly hear it out the speakers
10:41chadmedwe had a bunch of 700MHz band wireless shure mics that we had to bin in 2014, the government licensed the entire band they were on to mobile carriers for LTE
10:52svenspeaking of the horror that is USB, at least usb3 on the typec ports works *sometimes* now. other times it still pretends the thing is high-speed only or that the port is just broken...
11:02hizon has quit [Read error: Connection reset by peer]
11:03PaterTemporalis has joined #asahi-dev
11:09svenohhh… it always seems to work in one orientation. I guess there’s still a bug in the lane swapping
11:10svenI also still need to implement the tear down. Otherwise we’re back to “you only get to use this USB port once unless you reboot” :D
11:36gabuscus_ has joined #asahi-dev
11:39gabuscus has quit [Ping timeout: 480 seconds]
12:11marcanhonestly makes me want to rewrite kboot.c in Rust just so I can use a real programming language with real collections
12:18joske has joined #asahi-dev
12:37joske has quit [Remote host closed the connection]
12:54kovpovik, \o you said you had a few changes to tooling to build linux under mac, any chance I could bother you for the patch? =)
12:54joske has joined #asahi-dev
12:59kameks has quit [Ping timeout: 480 seconds]
13:17kgarrington has joined #asahi-dev
13:17kgarringtonmarcan: will you eventually support FileVault, or are you avoiding it completely?
13:18kgarringtonOh, the wiki explicitly says not to turn it on
13:20kgarringtonSo when you boot the Mac, it asks for your FileVault password (via the recovery partition instead of the Elle EFI system), and then boots to Asahi Linux?
13:20kgarringtonold EFI system, that is
13:21marcanI haven't tested the new step2.sh process but given how it goes through the volume scan, it should ask you for your password just like the old terminal method did
13:22marcanI don't think it has to be your FileVault password per se, we don't care about the encrypted volume
13:22kgarringtonBut the FileVault full disk encryption applies to the Linux partition? Or only to the Mac partition?
13:23marcanFileVault is very specifically encryption of the Data subvolume of the APFS container for macOS
13:24kgarringtonWell, if the storage system is based off of iPhone hardware, everything on the drive gets a base level of encryption (what Apple sometimes refers to as “class-D”
13:25marcanthe low-level full disk encryption should be there as far as I know, and should be transparent to us, since the NVMe controller ought to do that
13:25kgarringtonYou’re saying that on Apple Silicon Macs, FileVault works like Data Protection works on iOS does, only adding additional encryption to class A,B, and C data?
13:27kgarrington has quit [Remote host closed the connection]
13:32kgarrington has joined #asahi-dev
13:32marcankgarrington: you're confusing the Data Protection stuff with how the encryption actually works
13:33kgarringtonI’m not gone, I primarily read this via whitequark’s log and open Palaver when I (rarely) want to ask a question
13:34marcanalso https://support.apple.com/guide/security/volume-encryption-with-filevault-sec4c6dc1b6e/web
13:35sventhere might be additional layer of encryption macos uses with additional hw support from ANS. there are some fields inside that TCB structure related to some "aes key" that we just don't use for now
13:36marcanthat's how the protection classes are implemented, and presumably also the general FileVault wrapping for that subvolume
13:36svenask SEP or whatever for the key and just forward it to ANS when reading/writing blocks to offload the actual encryption
13:37yuyichao has quit [Ping timeout: 480 seconds]
13:37kgarrington has quit [Remote host closed the connection]
13:39marcanand then there should be a flag that tells ANS whether to use the xART key generation process or not
13:39svenI just guessed SEP wrapped the key somehow and passed it along in that field then. but I haven't looked at this at all fwiw
13:40kgarrington has joined #asahi-dev
13:41kgarringtonI should also say: awesome work everybody, and a special thanks to tpw_rules for the NixOS configuration
13:43kgarrington has quit [Remote host closed the connection]
13:43marcaneither way, from the point of Linux's NVMe driver, everything on disk is plaintext except for the Data subvolume for macOS
13:44marcanall that FileVault does is change whether the user login is part of the key unwrapping for that volume key (as protected by the xART mechanism) or not
13:44marcanand then individual files can have their own tweak or whatever and the data protection classes are implemented like that
13:46marcan(which is a bit different since no such volume-wide class C mechanism actually exists on iOS)
13:48marcanthe difference is on iOS you never need the passcode to mount it overall, on macOS you do if FileVault is enabled
13:49marcanI'm guessing the underlying NVMe storage is encrypted with some other device key anyway before it hits NAND, and that would apply to the entire namespace/device, but there's no real way to verify that from Linux
13:50marcaneventually I want to support that whole SEP-backed/xARTs/transparent encryption stuff from Linux, but that will take time
13:52yuyichao has joined #asahi-dev
14:20kameks has joined #asahi-dev
14:34klaus_ has joined #asahi-dev
14:40klaus has quit [Ping timeout: 480 seconds]
14:46joske_ has joined #asahi-dev
14:46joske has quit [Read error: Connection reset by peer]
14:49joske__ has joined #asahi-dev
14:49joske_ has quit [Read error: Connection reset by peer]
14:52joske has joined #asahi-dev
14:58joske__ has quit [Ping timeout: 480 seconds]
14:58kameks has quit [Ping timeout: 480 seconds]
15:09joske has quit [Quit: Leaving]
15:09bluetail[m]Will APFS require custom drivers ?
15:42WhyNotHugo has quit [Remote host closed the connection]
15:42gpanders_ has quit [Remote host closed the connection]
15:42coder_kalyan has quit [Remote host closed the connection]
15:42WhyNotHugo has joined #asahi-dev
15:42coder_kalyan has joined #asahi-dev
15:42gpanders_ has joined #asahi-dev
15:45loki_val has joined #asahi-dev
15:49crabbedhaloablut has quit [Ping timeout: 480 seconds]
16:17marcankettenis: any bright ideas for passing down the keyboard layout from ADT to DT for the laptops? particularly what mapping to use
16:17marcanlinux console keymaps are a different thing, so that'd need a mapping; what does BSD use for the console?
16:19jannaumarcan: I would have hoped that https://github.com/AsahiLinux/linux/blob/a90a4472ece3167409168aab7c5010fe0eba8bb9/drivers/hid/spi-hid/spi-hid-apple-core.c#L866 is sufficient
16:21jannauit certainly reports itself not sure if that is enough, I don't have the best keyboard layout to test this
16:21dottedmagmarcan: Isn't there a mapping from XKB layouts to console keymaps (I think Debian has introduced that)?
16:22jannauiirc I've checked if there a DT binding to report keyboard layout but I didn't find anything
16:23arndThere is a binding for the chromebook keyboard at include/dt-bindings/input/cros-ec-keyboard.h, but I think this is just to establish which keys are where, not what is printed on them
16:24marcan[root@asdasd3-pc ~]# cat /sys/devices/platform/soc/39b10c000.spi/spi_master/spi0/spi0.0/001C:05AC:0342.0001/country
17:25trouter has quit [Read error: Connection reset by peer]
17:28trouter has joined #asahi-dev
17:59the_lanetly_052 has quit [Ping timeout: 480 seconds]
18:18klaus_ has quit [Quit: leaving]
18:32ketteniswich means that the actual layout doesn't matter all that much unless you actually want to remap the keys
18:34kettenisas opposed to xt and at keyboards which report a keycode which you then have to map to what's actually printed on the keys
18:35marcankettenis: USB-HID keyboard report the keys by physical position, not what's printed on them
18:35m6wiq has joined #asahi-dev
18:43kettenismaybe because I didn't pay attention to the footnotes in the relevant section of the HID Usage Tables document ;)
18:46ketteniscombined with the fact that while there is an official .nl keyboard layout, all the letters are in the same position as with the .us layout
18:47kettenisand nobody actually uses that layout anyway, and what apple sells as a .nl keyboard layout doesn't actually match the official .nl keyboard layout anyway
18:51crabbedhaloablut has joined #asahi-dev
18:54loki_val has quit [Ping timeout: 480 seconds]
19:34Guest1274 is now known as foxtrot
20:01gladiac has quit [Quit: k thx bye]
20:06___nick___ has joined #asahi-dev
20:06___nick___ has quit []
20:08___nick___ has joined #asahi-dev
20:19MajorBiscuit has quit [Ping timeout: 480 seconds]
20:37riker77 has quit [Quit: Quitting IRC - gone for good...]
20:50m6wiq1 has joined #asahi-dev
20:51riker77 has joined #asahi-dev
20:51m6wiq1 has quit [Remote host closed the connection]
20:56m6wiq has quit [Ping timeout: 480 seconds]
21:07___nick___ has quit [Ping timeout: 480 seconds]
21:17m6wiq has joined #asahi-dev
21:33m6wiq has quit []
21:34conradev has quit [Quit: -]
22:52nyx_o has quit [Quit: WeeChat 3.4]
22:55nyx_o has joined #asahi-dev
23:23ChaosPrincess has quit [Remote host closed the connection]
23:46PaterTemporalis has quit [Ping timeout: 480 seconds]