#
…

2022-03-09

00:07amarioguy has quit [Ping timeout: 480 seconds]
00:11PthariensFlame[m]So what does the M1 Ultra officially having "32 Neural Engine core" mean for the ANE@1 on each die?
00:11PthariensFlame[m]Does the fact that having two ANEs hooked up at once is apparently not a problem tell us anything about why one (or two) of them are still disabled?
00:28user982492I noticed that Alyssa is developing the GPU drivers on macOS. Will that mean that macOS will get Vulkan support? If so, will it be good enough to run VKD3D-Proton?
00:32off^ has joined #asahi
00:55amarioguy has joined #asahi
01:02bluetail[m]<user982492> "I noticed that Alyssa is..." <- Yea, but I doubt it will happen within the next 12 months
01:02bluetail[m]My personal thought on this, perhaps in 2 years
01:03user982492Will the driver for M1 GPU work for M2, M3, ...? Or will it have to be redone each time?
01:07off^ has quit [Ping timeout: 480 seconds]
01:08Gu____________________________ has joined #asahi
01:15Gues__________________________ has quit [Ping timeout: 480 seconds]
01:21branon has joined #asahi
01:22branonm1 ultra support when
01:23chadmedit should Just Work(tm) with a devicetree
01:31user982492Are their Vulkan features or extensions that M1 cannot efficiently support?
01:31pwg_ has joined #asahi
01:31chadmedthere'll probably be a few things, yeah. we will have to emulate them in software
01:32chadmedthat's putting the horse before the cart though. the current focus is on GLES compliance as a baseline and to understand what the hardware can and cannot do
01:33chadmedand that's before even thinking about making GLES fast :P
01:34chadmedgood news is once we understand the hardware, writing a vulkan driver means we get zink support, so we only have to worry about making vulkan work nicely after that
01:36pwg has quit [Ping timeout: 480 seconds]
01:37user982492Why write GLES at all if zink can be used?
01:38chadmedits a (relative to vulkan) small API, but is enough to give us accelerated desktop environments and web browsers so its a good starting point for reverse engineering and understanding the hardware while also implementing must-have features
01:39user982492That makes sense
01:53user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
01:58csileeeeeeeeeeeeeeeeeeeeeeeeoe has joined #asahi
02:00user982492 has joined #asahi
02:05riker77_ has joined #asahi
02:10riker77 has quit [Ping timeout: 480 seconds]
02:10riker77_ is now known as riker77
02:14user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
02:26pfdub[m] has joined #asahi
02:35pfdub[m]Hey There, wanting to help with checking notes and creating documentation. I currently have a MacMini M1, know a bit about networking, good at Java, haven't programmed C in a while. How do I get permissions to see earlier messages so I can see what is going on?
02:37csileeeeeeeeeeeeeeeeeeeeeeeeoe has quit [Remote host closed the connection]
02:38Jamie[m]1not sure in matrix, but you can view logs here https://oftc.irclog.whitequark.org/asahi/2022-03-09
02:38pfdub[m]Thanks Jaime!
02:39pfdub[m]I am hoping to get permission from the mods.
03:07csileeeeeeeeeeeeeeeeeeeeeeeeoe has joined #asahi
03:18user982492 has joined #asahi
03:26phiologe has joined #asahi
03:28amarioguy has quit [Ping timeout: 480 seconds]
03:29PhilippvK has quit [Ping timeout: 480 seconds]
03:51artemist has quit [Ping timeout: 480 seconds]
03:58kov has quit [Quit: Coyote finally caught me]
03:59kov has joined #asahi
04:18psykose has quit [Remote host closed the connection]
04:18psykose has joined #asahi
04:30marvin24_ has joined #asahi
04:33marvin24 has quit [Ping timeout: 480 seconds]
05:00hizon has joined #asahi
05:22colemancda has joined #asahi
05:49GlanzmannSome posted a geekscore benchmark on the new m1 ultra: https://browser.geekbench.com/v5/cpu/13330272 on hacker news.
06:00Aaron[m]1<pfdub[m]> "I am hoping to get permission..." <- Matrix doesn't have any function to grant specific users access to history
06:02csileeeeeeeeeeeeeeeeeeeeeeeeoe has quit [Remote host closed the connection]
06:04chadmedmatrix is just a bridge to the IRC channels on OFTC
06:04chadmedthis is an IRC channel
06:11the_lanetly_052__ has joined #asahi
06:13colemancda has quit []
06:26gladiac has joined #asahi
06:48sheepgoose has quit [Ping timeout: 480 seconds]
06:54sheepgoose has joined #asahi
07:06sheepgoose has quit [Ping timeout: 480 seconds]
07:23sheepgoose has joined #asahi
07:27user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
07:43jannauapple's order system is strange, the same M1 Ultra / 128GB config has an earlier delivery date when configured from the m1 max mac studio
07:46eroux has joined #asahi
07:51memoryleak has joined #asahi
07:54arndjannau: good to know. I don't expect it to actually make a difference. It seems another difference is that doing it this way you can pick a 512GB SSD, while the other one starts at 1TB for an extra €230. For the difference, I would probably get a 2TB m.2 drive and an external TB enclosure.
07:55sheepgoose has quit [Ping timeout: 480 seconds]
08:04jannauas far as I'm aware the store delivery estimates are somewhat reliable. The difference is not a few days but a month: 30th of march to 6th of april vs 4th to 11th of may
08:06jannauit's the other way for the 64GB ram config
08:09sheepgoose has joined #asahi
08:18memoryleak has quit [Ping timeout: 480 seconds]
08:23sheepgoose has quit [Ping timeout: 480 seconds]
08:25ciggi has joined #asahi
08:31sheepgoose has joined #asahi
08:45memoryleak has joined #asahi
09:29marcanjannau: weird, maybe I should've tried that...
09:30marcanhuh, indeed, it's earlier, though still later than what I got (at least now)
09:33rednaks[m]Been reading that Linux scheduler wasn't ready for 12th gen was there any specific dev on the scheduler made for the M1 ?
09:33chadmedarm chips have had heterogeneous multiprocessing for a long time so the infrastructure was all there
09:34rednaks[m]oh nice then :) thanks
09:35chadmedintels problem aiui was getting the linux scheduler to actually play nicely with their hardware scheduler block, which they call thread director
09:36mpssecure boot, https://xkcd.com/538/ (sorry for OT)
09:38mpsif I lost or machine is stolen I consider all my data are open
09:39mpsand nowadays with these secret CPUs in SoCs even if I'm online
09:39marcanthe M1 does not have any secret CPUs with full system access
09:39marcannevermind network access either
09:40marcanthe only proprietary firmware with full access as far as we can tell is the GPU ASC, because it manages page tables
09:40marcanand it obviously does not talk to the network or anything else
09:40mpsmarcan: but what with these 'chips' around in machine
09:40marcanthose chips have even less access
09:40marcanI have a fairly exhaustive list of all the blobs in these machines already
09:41mpshmm, so the M1 is safest than intels, iiuc
09:41marcanyes, I've been saying that all along :p
09:41mpsdidn't know this, thanks
09:41eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
09:42mpsbut still I don't care about secureboot and FS encryption
09:42marcanthat's something for each person to decide
09:43pwg_ is now known as pwg
09:43pwg has quit [Quit: WeeChat 3.4]
09:43pwg has joined #asahi
09:43marcanI personally do not consider secureboot critical, but I do FS encryption at rest at the very least
09:43marcanand secureboot is nice, especially for more sensitive use cases or unattended boot
09:45eroux has joined #asahi
09:45mpsI see them just as a layers in security not as final solution
09:56marcanstreaming shortly
10:03eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
10:06MajorBiscuit has joined #asahi
10:53dianshi has joined #asahi
11:14eroux has joined #asahi
11:26Gu____________________________ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
11:26hizon has quit [Remote host closed the connection]
11:57eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
12:00amwGlanzmann: Thanks for the postings. I'm still running debian bullseye - so I might try the live stick with testing
12:01amwFunny thing is that the older kernel works perfectly fine with the debian bullseye - ...
12:19Glanzmannamw: Did you update the libinput? Because I still don't get how the touchpad works for you.
12:20os has quit [Quit: The Lounge - https://thelounge.chat]
12:21os has joined #asahi
12:21fridtjof[m]they mention the NVMe issues, but excluded/missed the detail where it's a firmware bug that might (or might not tbh) get fixed
12:23amwGlanzmann: libinput-bin 1.16.4-3, xserver-xorg-input-libinput 0.30.0-1 - stick bullseye I believe
12:25Glanzmannamw: Yes, thats bullseye. Because someone else is also running Debian stable, but did not have a trackpad, so I told him about libinput. He took the one from testing and than it was working for him.
12:25Glanzmannamw: The exact same x error I had before, but I don't know what I did to get rid of it.
12:26amwGlanzmann: Your live-stick only has an initrd (big one) - has that got X in it?
12:27Glanzmannamw: No, it has not. But you can do apt update; apt install xinit blackbox xterm
12:27Glanzmannor any other package for that matter.
12:28GlanzmannSo as long as you don't run out of ram, you can install whatever you want including gnome.
12:28amwok - assume it has the wifi stuff in it?
12:28GlanzmannIt uses the new location.
12:28GlanzmannSo make sure that you have a vendorfw/firmware.tar on your efi partition, than it works out of the box.
12:29GlanzmannYou just need to edit /etc/wpa/wpa_supplicant.conf and put your essid and psk and do a
12:29Glanzmannifdown <wifi interface> ; ifup <wifi interface>
12:30Glanzmannamw: The current version uses 4k pagesize.
12:31Glanzmannj`ey: Is the m1n1 chainloading from disk already working for kernel junior developers?
12:31j`eyGlanzmann: looks like it's working now, just not pushed yet
12:32j`eywell, pushed to the rust branch, so you could try build it at least.. if you wanted
12:32GlanzmannI see, I did a git pull and there was a new branch 'rust' that is why I was asking.
12:32Glanzmannj`ey: Is there a quickstart howto what needs to be done to test it?
12:33j`eyGlanzmann: no
12:33j`eyGlanzmann: installing rustup is probably a good first step
12:34GlanzmannI see, I'll try in 1,5 hours or so and ask if I don't get it running.
12:34GlanzmannThank you.
12:35amwGlanzmann: You initrd contains the .deb files (in var/cache/apt/archives) that were used to build it :-)
12:38amwGlanzmann: Looks like my initrd is missing the wifi firmware - probably didn't put it in the right spot - thanks will play with it tomorrow - should be fun
12:41eroux has joined #asahi
12:48euxine has joined #asahi
12:52amarioguy has joined #asahi
12:59yuyichao_ has quit [Ping timeout: 480 seconds]
13:01mbilker has joined #asahi
13:03Glanzmannamw: You can also manually extract it and than run the commands from /etc/rc.local
13:03euxine has quit [Remote host closed the connection]
13:07Glanzmannamw: I had the line in, but forgot the chroot .... https://pbot.rmdir.de/hJ6lmw4QjyTyIUF-7PO6mQ
13:09mpsGlanzmann: iiuc from backlog you built kernel with povik's patches for speakers, did you used some patches or build from povik's fork?
13:11j`eyGlanzmann: rust is now in the main branch
13:12mbilker has quit [Quit: Arch Linux Rules]
13:12mbilker has joined #asahi
13:14yuyichao_ has joined #asahi
13:25Glanzmannmps: I never built it, but someone else did and it worked for him.
13:25GlanzmannBut than only the spakers work IIRC.
13:26j`eyGlanzmann: repull, it's fixed
13:26j`eymake CHAINLOADING=1 to get the nvme chainloadin code
13:27GlanzmannThanks. I see.
13:27Glanzmannj`ey: So m a r c a n currently puts it in the installer?
13:27GlanzmannThan I only need to put my m1n1/u-boot into right directory?
13:27mpsGlanzmann: ah, I misunderstood then, sorry
13:28j`eyGlanzmann: exactly
13:29GlanzmannHas m a r c a n mentioned in the stream how he plan to distribute the to be chainloaded object? Will he put in the kernel package or as a separate package?
13:29mpsj`ey: where u-boot should be put? '/' on ESP
13:29j`eynot sure
13:29Glanzmannmps: I'll find soon enough and tell you.
13:30povikconcatenated to m1n1 maybe
13:30j`eyit might not be fixed, you might be able to specify it
13:30mpsj`ey: ok, will look source later
13:30povikm1n1 the second, to be specific
13:31mpsGlanzmann: no, you don't have to look this for me, I'm not in hurry
13:31Glanzmannmps: m. is working on the installer as soon as he is finised, I'll get the efi zip extract it and than I know the filename.
13:31Glanzmannmps: I'll look it up for myself to adopt the debian installer and than tell you. :-)
13:31j`eypovik: mIInII
13:31mpsGlanzmann: ok, and thanks :)
13:32GlanzmannI'll call it studio.
13:36bisko has joined #asahi
13:37kovGlanzmann, can I haz dm-crypt on that one =D
13:38Glanzmannkov: Do you have the options, than I put them in.
13:40Glanzmannj`ey: It is building.
13:41kovGlanzmann, CONFIG_DM_CRYPT and CONFIG_CRYPTO_AES should probably be enough
13:42kovthere are a bunch of hardware-specific acceleration options CONFIG_CRYPTO_AES_ARM64_CE CONFIG_CRYPTO_AES_ARM64_CE_CCM CONFIG_CRYPTO_AES_ARM64_CE_BLK CONFIG_CRYPTO_AES_ARM64_NEON_BLK CONFIG_CRYPTO_AES_ARM64_BS, I don't know which are useful but I think none are required
13:43kovhopefully with those two only we can use cryptsetup to prepare or fix the file systems from debian live
13:46mbilker has quit [Quit: Arch Linux Rules]
13:46mbilker has joined #asahi
13:52GlanzmannLet me rebuild with them then.
13:55mpskov: I enabled all of them as modules, as I do for alpine linux-edge kernel
13:56Glanzmannkov: CONFIG_CRYPTO_AES was already in, CONFIG_DM_CRYPT was missing. I'm rebuilding. Should be done in 5 minutes.
13:57Glanzmannmps: I will try to take the Debian config and enable the asahi specific options and put that in a script, than this should be once and for all working, hopefully. I tried that before but the kernel was not booting for whatever reason.
13:57GlanzmannAnd in the future I'll probably boot two different variants: 4k and 16k pagesize so that everyone has what he wants/needs.
13:58j`eywhat *they* want :)
13:59Glanzmannjannaus says it is more reasonable to ship a 4k kernel because that is what distros probably will do and more stuff works. m a r c a n says he wants to ship 16k so that more stuff breaks and gets fixed.
13:59mpsGlanzmann: strange it is not booting if the needed parts for asahi is enabled
13:59GlanzmannI'm also more a 16k person. But I noticed that most users want 4k for the time being.
14:00kovmps, same
14:00Glanzmannmps: Maybe it was just me doing something stupid. I'll try again. And if it does not boot, I'll boot it in the hypervisor and see what is actually breaking.
14:01Glanzmannj`ey: Got it. :-)
14:01mpsGlanzmann: you should get error on console if you remove 'quiet' from kernel cmdline and add loglevel=7
14:01GlanzmannLast time no console came up ...
14:02Glanzmannkov: You're build artefacts are ready.
14:02mpsoh, that is strange then, could be clash with CONFIG_FB somewhere
14:03GlanzmannOr me doing something stupid.
14:03GlanzmannI'll try now.
14:03mpsearlier (unrelated to asahi) I had this problem with simpledrm and FB on one test machine
14:04mpsbut forgot details
14:04_jannau_I would not expect that distros off multiple kernel images or multiple installer images for an efi distro installer. This only matters when enough HW support is upstream (or as patches in the distro kernel)
14:05mpsbtw, I reported f2fs problem with 16K pages to upstream, waiting for answer
14:06Glanzmannjannau: The Debian installer guy also says that debian will add an additional flavor regarding the kernel for m1.
14:15arndmps: the problem that a lot of users reported with CONFIG_FB is that DRM_FBDEV_EMULATION no longer forces CONFIG_FB=y, but instead gets turned off when you have DRM_KMS_HELPER=y and FB!=y
14:16arndso you need to manually enable CONFIG_FB=y in order to use the FBDEV emulation
14:19_jannau_Glanzmann: but will they add an installer/live image using that kernel flavor? I hope not
14:28mpsarnd: I looked here when I added simpledrm to some machines https://fedoraproject.org/wiki/Changes/ReplaceFbdevDrivers
14:37jmr2 has joined #asahi
14:39Glanzmannjannau: No, they will do that once it is upstream.
14:39Glanzmannarnd: That could be my problem, then.
14:40mpsjmr2: I don't have github account, please do this if you have it
14:44jmr2It might be easier for others to do it if you provided a description of the issue and a link to your bug report. But you know that github accounts are free, right?
14:45jmr2 has quit [Quit: Page closed]
14:50amarioguy has quit [Ping timeout: 480 seconds]
14:53mpshm, jmr2 left? I don't want to create account on microsoft owned platform
14:53emergenz[m]<jmr2> "It might be easier for others to..." <- Free as in beer, but not free as in speech
14:56amarioguy has joined #asahi
15:37ciggi has quit [Ping timeout: 480 seconds]
15:44kloenk has quit [Remote host closed the connection]
15:54amarioguy has quit [Ping timeout: 480 seconds]
16:03marcanpushed a new installer, OS kernel is probably broken (:p) but the UEFI only option should work and now does the whole chainloaded m1n1 magic
16:03j`eymarcan: can you push the insaller to git too?
16:04j`ey(np if you wanted to fix the commits up later)
16:04tpw_rulesmarcan: is the chainloaded m1n1 written down anywhere? want to know how to trigger it and stuff
16:05marcanit's in the commit message
16:05j`eytpw_rules: 4575b35479306e1942fb726454a8c892921548df in m1n1 shows an example
16:06_jannau_wasn't it just a PCIe module late loading issue? The keyboard was probed. So it could work on the laptops or with a keyboard connected to the usb-c ports
16:06tpw_rulesah so it is, thanks
16:06marcanyeah, probably works on the laptops?
16:06marcanbut also USB-C didn't work either so that's bork
16:06marcanalso wifi didn't load either I think
16:06marcananyway, I'll debug later
16:08mpsiiuc from example, make and install m1n1 in 1TR, put m1n1.bin on ESP
16:09FireFox317not sure if this is mentioned already, but marcan your tweets on the nvme storage of apple silicon are mentioned in LTT: https://www.youtube.com/watch?v=k8RkyaSHFLM&ab_channel=LinusTechTips
16:09mazmarcan: any update on AICv2? I'm hopeful that BHB will buy us an extra week, but I can't promise it either...\
16:10marcanmaz: going to send that out right after dinner
16:10mazmarcan: awesome, thanks.
16:20tpw_rulesit's not possible to chainload m1n1 from anything but m1n1, right? could it be done from u-boot or grub
16:26_jannau_no, m1n1 uses the xnu boot protocol
16:48ciggi has joined #asahi
16:50Telvana2 has joined #asahi
16:51eroux has quit [Remote host closed the connection]
16:53Telvana has quit [Ping timeout: 480 seconds]
16:53eroux has joined #asahi
16:56kovGlanzmann, \o/
16:57Telvana2 has left #asahi [Textual IRC Client: www.textualapp.com]
16:57Telvana has joined #asahi
17:04amarioguy has joined #asahi
17:37kovtpw_rules, _jannau_ I thought that was what marcan was working on in the last stream?
17:37kovor did I misunderstand what the chainload feature was
17:38j`eykov: what he worked chainloaded m1n1 off nvme, from m1n1
17:38kovoh I see
17:38j`eynot chainloading m1n1 from uboot/grub
17:39kovso you put a new m1n1 somewhere and it'll chainload from there
17:39j`eyexactly, so you do: m1n1 -> m1n1+u-boot (chainload from ESP)
17:39j`eyso that u-boot can be upgraded by the distro
17:40kovgreat, that way the dtb can be upgraded as well, is the protection against downgrades already a thing as well?
17:40j`eyno protection there
17:41GlanzmannSo I tried to modify a debian distro kernel and added the asahi options. I verified that all options are in. When I boot I see it reaching userland but it never starts x or offers me a login prompt. The keyboard appears to be _not_ working. But all the SPI options are in.
17:42GlanzmannLet me do a video of it.
17:43mpsj`ey: do you know how to create m1n1+u-boot+dtbs for chainloading? as earlier?
17:44FireFox317mps, im pretty sure its just catting m1n1.bin dtbs and then u-boot
17:44j`eybut to create a m1n1 that *can* perform chainloading, build with CHAINLOADING=1
17:44j`eythat bit requires rust
17:44mpsok, I think so but didn't tried yet, lets see
17:45mpsj`ey: yes, I see these instructions in m1n1 repo
17:46mpsoh my whatever, now need cargo to build
17:48mpshm ` failed to read `/work/m1/m1n1/rust/vendor/bitflags/Cargo.toml``
17:49j`eymps: git submodule update --init
17:49Glanzmann[ 4.278] (EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x188) [0xaaaad26e0398]
17:49Glanzmann[ 4.278] (EE) unw_get_proc_info failed: no unwind info found [-10]
17:49GlanzmannThis is the same error amw gets.
17:49j`eyGlanzmann: "Task dump for CPU 5:" is fishy, everything else looks ok :/
17:49GlanzmannIs this the colorspace mapping isue?
17:51marcanGlanzmann: I bet you're using efifb instead of simpledrm
17:52marcansimpledrm as a module is a bad idea, as I said
17:52GlanzmannMight be, yes.
17:52Glanzmannmarcan: But now I'm booted, I have x and I also seem to use 'efifb'.
17:53marcandidn't you say it crashes?
17:53mpsj`ey: now got `error[E0463]: can't find crate for `core``
17:53Glanzmannmarcan: Nope, it does not crash. It hangs.
17:53j`eymps: Im not sure how to get a cross compiler rust on alpine
17:54GlanzmannAnd if I press the power button it shutdowns cleanly.
17:54GlanzmannBut yes, I guess it is the simpledrm issue. So let me try that and report back.
17:54mpsj`ey: I'm building on aarch64
17:54mpsnot cross building
17:54marcanyou're cross building because the rust target is not linux
17:55j`eyits a 'bare metal' target
17:55psykosethey're not packaged afaik
17:55mpspsykose: right
17:55psykoseyou would need rustup, if it has those
17:55j`eyrustup does
17:55psykosewith rustup add `RUSTFLAGS="-C target-feature=-crt-static"` to be safe, and the rest should work fine
17:56mpspsykose: where to add these flags?
17:56psykoseexport, env
17:57psykosenot that it probably matters for bare metal anyway :p but i have no idea
17:57j`eyyeah it shouldnt need that
17:57csileeeeeeeeeeeeeeeeeeeeeeeeoe has joined #asahi
17:57psykosei don't think it would affect the build in that case at all, but in regular builds the rustup musl toolchains are broken
17:58mpspsykose: it doesn't :)
17:58mpsyou are right
17:58mps(looks like I have to relearn about rust build system again)
17:59psykoseyou can just run rustup-init, then after it's rustup target add or something
18:00mpsalready did, but didn't helped
18:01psykoseyou have to load the new env vars
18:01psykosenamely just adding ~/.cargo/bin to path
18:01psykoseand make sure that is before the system rust
18:03mpspsykose: ok, thanks
18:04arndjannau: I ordered mine now with expected delivery Apr 6-13, but now the product page says May 6-13 regardless of the model. Either they the order system it, or enough people did the same thing until the dates lined up
18:13mazarnd: what model did you go for?
18:16arndmaz: 20 cores, 48GPU/64RAM/512SSD
18:18Glanzmannmarcan: simpledrm=m improves the situation, but I still have no sound and no x. dmesg is here: https://pbot.rmdir.de/YTSJHigCoiiBsqtfbe_3Iw
18:18Glanzmannthe x error is still the same.
18:19pugguu[m]1Glanzmann: is the chain loading working in debian?
18:20Glanzmannpugguu[m]1: Not yet.
18:20Glanzmannpugguu[m]1: I have to put one file into esp.tar and than it works.
18:20GlanzmannAt least I hope so.
18:21Glanzmannmarcan: But it is simpledrm.
18:22GlanzmannBecause I now built in the module but I can tell from the dmesg that it is not initializing
18:22pugguu[m]1By the way I’m now making it a thing to ask for things to update every once in a while
18:22mazarnd: ah, there *is* a version with a smaller nvme?
18:23eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
18:23arndmaz: yes, you have to pick the 10-core model first, and then upgrade to the larger CPU but keep the small SSD, jannau first noticed that doing this also gives you four week delivery instead of eight weeks for identical hardware config
18:23j`eymaz: yeah you start by configuring the m1 max version
18:23mazah, nice!
18:23arndbut now it seems to all be up to eight weeks
18:24mazjust in time for my birthday!
18:24arnd(in apple.com/de/)
18:24eroux has joined #asahi
18:25arndmaz: oh, it still works in apple.com/uk/: 2-3 weeks vs 8-9 weeks
18:25eroux has quit []
18:25mazarnd: says 9 May - 13 May
18:27arndmaz: ok, I see it now, the 2-3 weeks delivery is only for 32GB RAM, which is not available with the Ultra
18:28eroux has joined #asahi
18:29pugguu[m]1<arnd> "maz: yes, you have to pick the 1..." <- Yea I did that trick to get a 16gb ram for Cheaper instead of getting the 16gb 512 I got a 16gb 256 and saved 200 quid at the time until they made the 512 model 8GB by defult
18:31pugguu[m]1Or something like that lol
18:32amarioguy has quit [Ping timeout: 480 seconds]
18:36MajorBiscuit has quit [Quit: WeeChat 3.4]
18:37amarioguy has joined #asahi
18:45the_lanetly_052__ has quit [Ping timeout: 480 seconds]
18:55Glanzmannmarcan: So I tried compiling SIMPLEDRM that does not solve my problem, than I tried to remove CONFIG_FB_EFI, than I have no framebuffer.
18:56GlanzmannFor me it appears we need both, but they need to be intialized in the right order so FB_EFI initialized the framebuffer but simpledrm is providing the /dev/dri/card0
18:58GlanzmannAnd efi fb seems to be fb0 and simpledrm seems to be fb1.
18:58jevinskie[m]marcan: did you eventually find IODTNVRAM::unescapeBytesToData()?
19:04mpsGlanzmann: I have FB_EFI, SYSFB and DRM_SIMPLEDRM, all =y and it works fine
19:06GlanzmannI have the same, but on the debian distro kernel it does not work but with my old config which is probably based on yours or jannaus it works.
19:10user982492 has joined #asahi
19:12mpsGlanzmann: do you have CONFIG_FB_SIMPLE, maybe it should be disabled
19:13Glanzmann# CONFIG_SYSFB_SIMPLEFB is not set
19:13mpshave no idea then
19:16Glanzmannmps: It was: CONFIG_BACKLIGHT_GPIO=m
19:17GlanzmannWhich was missing.
19:17GlanzmannSo now xorg works, but sound does not come up.
19:17Glanzmannpovik: Any idea? [ 2.236988] apple-pmgr-pwrstate 23b700000.power-management:power-controller@2c0: PS mca1: Failed to reach power state 0xf (now: 0x24f)
19:18GlanzmannI wonder why it was not a compile failure, but it is clear the new simpledrm has backlight binding, so simpledrm was not initializing for me.
19:18Glanzmannso efifb is fb0; simpledrm is fb1
19:20arndGlanzmann: what happens when the gpio backlight driver is not there? No backlight or probe failure?
19:21mpsGlanzmann: I don't have /dev/fb1
19:22Glanzmannarnd: It seems that I get a probe failure, because I get no simpledrm log message at all.
19:22Glanzmannmps: Can you post your dmesg?
19:22___nick___ has joined #asahi
19:24Glanzmannmps: I assume you have no efi support in your kernel, this is why efi fb is never initalized: [ 0.000000] efi: UEFI not found.
19:25GlanzmannSo now I'll try to throw out efi_fb.
19:25arndGlanzmann: I think that message is the opposite: efi support is in the kernel but not the boot loader
19:25mpsright, I boot with extlinux.conf
19:25Glanzmannarnd: I see, thanks.
19:26arndGlanzmann: if the lack of the backlight-gpio driver causes the probe to fail, that might be a bug in the simpledrm driver, but I don't see where it references the backlight at all
19:26mpsdidn't yet found time to switch back to EFI
19:26Glanzmannarnd: I'll thow it out again, recompile and verify my assumption.
19:26arndI would expect it to treat the backlight as optional: if no driver is found, it would continue without backlight
19:26mpsarnd: simpledrm worked without gpio earlier
19:27mpsso I think it is not related, but idk for sure
19:28Glanzmannmps: But that was before marcan did the patch.
19:28arndI think there should also be some handover code, which ensures that you can boot with a built-in simpledrm or simplefb or efifb but later load the real driver that then owns the device after the simple drivers give it up.
19:29GlanzmannAlso when I rmmod backlight_gpo my xorg hangs.
19:29arnd I can imagine that there is no handover between efifb and simpledrm because both expect to be the early drivers, not the real ones
19:29Glanzmannbut I can press the powerbutton and see the shutdown messages.
19:29marcanarnd: the DCP driver will do proper handover
19:29marcanwhich is why I expect simpledrm to be built in
19:29amarioguy has quit [Ping timeout: 480 seconds]
19:30j`eyarnd: the backlight code in simpledrm is only in the asahi linux repo
19:30jannauhandoff from simpledrm to dcp already works
19:30arndthere should also be handover from efi_earlycon to efifb, but I assume nobody here is actually using efi_earlycon (despite it being really useful for debugging problems in early boot)
19:30marcanbacklight can't really be "optional" because it could be a module, and we need the usual PROBE_DEFER stuff to handle init order properly
19:31arndah right, you can't have it both ways like this
19:31marcanbut in that case it doesn't actually do it properly
19:31marcanso it *should* be optional (and sometimes broken)
19:31arndj`ey: ok, that's why I didn't see that commit in mainline ;-)
19:32marcan(that needs a fix to handle PROBE_DEFER)
19:32marcaneither way that backlight stuff is a stopgap
19:32marcanit'll go away entirely once DCP is in
19:33marcanI don't plan on trying to have handoff/fallback for that
19:33arndso it works when CONFIG_BACKLIGHT_CLASS_DEVICE is disabled, and it works before that commit
19:33marcanit's going away from the DT once DCP is in
19:33Glanzmannarnd / marcan: I verified that the simpledrm does not initialize when I move the gpio_backlight module out of the way and rebuild the initrd
19:34* marcan-> sleep
19:34Glanzmannmarcan: Have a good night sleep.
19:35j`eymarcan: to idle or a deep sleep? ;)
19:35GlanzmannHe has not yet disabled nmis.
19:35bluetail[m]slightly off-topic: Does pastebot runs in docker or something, and how do you invocate a new upload? Through pipe?
19:35GlanzmannBut I can tell that he is hydrated due to the toilet breaks and that he is proper fed, because he interrupted a stream to get something to eat. :-)
19:36kaferka has joined #asahi
19:37GlanzmannSo now I need poviks expertise to get the sound up and than we have a 'proper' distro kernel for Debian with the asahi options.
19:38arndmarcan: I think we probably want to always ignore -EPROBE_DEFER from devm_of_find_backlight() in the simpledrm, but not ignore it in any of the real drivers, so only the handoff gets deferred but not the console messages
19:38arndthat's probably what you and in mind as well
19:41povikGlanzmann: fixed in wip tree (that error you posted ^^)
19:42Glanzmannpovik: Okay, so what is the best way to incorporate your wip tree in the current asahi branch?
19:42GlanzmannJust a 'git pull' or should i cherry pick a few commits?
19:43Glanzmannpovik: Also with your wip branch, do I have speakers or the jack on the macbook air?
19:43kaferkahey there - i'm no dev by any means but could anyone point me towards the detection timeout for pcie on m1? my wi-fi is missing and I want to poke around for fun.
19:44j`eybut I think someone (chadmed?) said that it didn't help, but of course it's fun tro trying thins
19:45kaferkaj`ey: yeah, i thought that was it but i guess i can confirm chadmed's finding
19:46kaferkabut yeah it is fun - adhd is a bitch but asahi linux gets me going lol
19:47povikGlanzmann: ehm, best is to wait
19:47poviksecond best is pull
19:47povikit's on top of asahi, don't know if still current asahi
19:49j`eykaferka: were you K-A on youtube?
19:49kaferkaj`ey: yeah haha
19:50j`eykaferka: I read your message :-)
19:50kaferkathanks :')
19:56tpw_rulesGlanzmann: what page size is your kernel?
19:56tpw_rulesthat x is crashing on
19:56Glanzmanntpw_rules: 4k
19:57Glanzmanntpw_rules: Issue resolved. Issue was that gpio-backlight was not in the kernel as a result simpledrm was not initializing as a result x did not start due to a colormapping issue.
19:57Glanzmannamw: PING. This is also the problem you had.
19:58Glanzmannj`ey: So about the new installer is it ready for junior kernel developers?
19:58GlanzmannIs it already on the usual location or do I need to build it myself?
19:58bluetail[m]<j`ey> "kaferka: https://github.com/..." <- You guys are merging with Torvalds? Just to put that into perspective, Linus Torvalds is the main developer of the Linux kernel.
19:58bluetail[m]I think that is kinda impressive.
19:59j`eybluetail[m]: yes all the code is going upstream
19:59j`eyGlanzmann: I think it's ready yeah
19:59GlanzmannThan lets deploy it.
20:00j`eyGlanzmann: you'll have to create the m1n1+uboot yourself, and double check where it's actually expected to be, I cant remember where
20:00j`eybluetail[m]: you can already boot on the m1 with mainline linux.. it just doesnt have many features
20:03Glanzmannj`ey: Will do so.
20:03povikha! turned out my speakers were being continuously throttled by the brownout detector
20:03povikthat explains two things: why they weren't too loud, why one of them died sometimes
20:04j`eysomething about the low frequencies?
20:04bluetail[m]povik: Brownout detector? Inside your home installation? That's... interesting...
20:04povikhm? home installation?
20:04povikj`ey: no, just bad (default) configuration
20:05j`eypovik: any progress on the fast audio?
20:05bluetail[m]povik: Thinking about M1 internal speakers, they don't have brownout detectors, do they?
20:05povikj`ey: fast? :D
20:05Glanzmannmps: That appears to be the path for the to be chainloaded object: esp/m1n1/boot.bin
20:05povikbluetail[m]: they do
20:05j`eypovik: you said the audio was sped up (yesterday or a few days ago)
20:05povikbluetail[m]: or rather the chip driving them has
20:06povikeach speaker has its own chip, so i consider them together
20:06povikj`ey: oh right, that was just bad NCO config, easy to fix
20:06povikone final thing i was fixing now was reliability
20:06bluetail[m]povik: What do you do to solve that? Would you just communicate to the brownout connector to consistently push a specific voltage?
20:07povikbluetail[m]: i configure the chip over i2c, copying what apple configures
20:07j`eypovik: ah cool
20:09poviknow let's slap a new compatible onto it and ship it!!!
20:10poviki am very excited i fixed both issues i had with it at once :D
20:10bluetail[m]How did you manage to understand those things ( shortly )
20:13poviki built robots when i was younger
20:13povikthe other end of that question would be answered by there being a datasheet for a part similar than is in the macbooks
20:14user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
20:15mpsGlanzmann: aha, but I can't built m1n1 now because of rust build crap^Wsystem on alpine
20:16Glanzmannmps: Let marcan build the 'rust' m1n1 and build it without rust.
20:16GlanzmannAnd than the spinning rust will chainload your m1n1.
20:17Glanzmannmps: This is what I'm about to do.
20:17j`eymps: can you paste your current m1n1 apkbuild (with rust attempt), maybe i will have a play later
20:18mpsj`ey: I can upgrade it alpine aports later this evening, but will not push to alpine because it will fail and block builders
20:18kaferka has quit [Read error: Connection reset by peer]
20:18___nick___ has quit []
20:18mpsI mean, upgrage in local branch
20:19mps(I'm out of house now)
20:21___nick___ has joined #asahi
20:30user982492 has joined #asahi
20:36vnogueira has quit [Remote host closed the connection]
20:36vnogueira has joined #asahi
20:42Glanzmannm a r c a n / j`ey / mps: Installer works, thanks. Chainloading, too. Where do I get the u-boot tree with the fancy asahi logo in the middle from?
20:43Glanzmannpovik: Sound works from speakers with your wip branch merged, for now I'll not put in the debian tree and wait until m a r c a n has merged it.
20:44povikGlanzmann: good
20:44povikjack should be there too, but we are missing ucm for switch to it to happen automatically
20:45Glanzmanntpw_rules: THanks.
20:45Glanzmannpovik: Let me test the jack.
20:46Glanzmannpovik: In pavucontrol I can only choose between Stereo output / off.
20:46povikwith pulse, you can get the jack by getting into pacmd
20:46povikand running: load-module module-alsa-sink device=hw:0,1
20:48GlanzmannThat works, but now how I expected it to be. :-)
20:49Glanzmannpovik: Also on the jack the audio volume was to high.
20:49GlanzmannThank you. I'll write that down.
20:49povikwell as i said, some needs to volunteer to write that ucm :-p
20:53Glanzmannpovik: I would have expected that I can disable the jack it in this pull down menu: https://pbot.rmdir.de/FtgnlfqAU7pKLzJG6_pI9Q.png
20:53Glanzmannpovik: But I actually like this way better, because now I can play one audio stream on the speaker and hear another one on the jack. So this is a feature, not a bug.
20:56povikthat mode of playing in parallel through the speakers and jack has a defect
20:57povikthere's noise mixed-in then, at a period
20:57povikdon't know how that happens yet
20:59j`eyGlanzmann: asahi branch in u-boot has the options 058c0bf9bb2601f1618b48822cb5851381e48d53
20:59j`eyah looks like it was already linked
21:00Glanzmannj`ey: Thanks anyway.
21:00j`eyGlanzmann: cool that chainloading works
21:05bluetail[m] Glanzmann: Realizing I was the one not researching before asking the question. rmdir's pastebot has cli integration right in front of me, I just didnt see it.
21:06Glanzmannbluetail[m]: It has actually two: curl and perl. :-)
21:07Glanzmannj`ey: I also adopted debian installer image for the new asahi installer.
21:07___nick___ has quit [Ping timeout: 480 seconds]
21:17Glanzmannmarcan: Please consider the following patch to add Debian bookworm to the installer menu: https://tg.st/u/0001-Entry-for-Debian-bookworm.patch
21:19Glanzmannmarcan: How do you package the /boot/efi/m1n1/boot.bin? Do you put it in the kernel package or in a seperate package or what are you plans with it?
21:19tanty has quit []
21:20opticronawesome, sound and deep sleep are the last things I need for this machine to be a daily driver
21:20Glanzmannopticron: What m1 machine do you have?
21:22tanty has joined #asahi
21:23opticroncurrently have debian on it from the pre-release install process, but I'll probably need to redo that for the chainloading release
21:23mpsGlanzmann: your work on integration different things (from user POV) is awesome, I have to say
21:24mps(I drink too much wine with friends and I'm 'out' today to try anything)
21:25tanty has quit []
21:26Glanzmannopticron: I just changed my debian from prerelease to the newest asahi installer.
21:27Glanzmannmps: Hrhr, I need to hit bed, have to get up tomorrow at 06:00 ...
21:27tanty has joined #asahi
21:29mpsGlanzmann: sleep well
21:31Glanzmannopticron: This is what I did more or less: https://pbot.rmdir.de/AISRqKelnFFJ-NGdjSt6Pg
21:31Glanzmannmps: You, too.
21:31mpsthnaks
21:33Glanzmannmps: Anyway, I tested the spakers and they work. You can find everything you need in the history of this git repository: https://git.zerfleddert.de/cgi-bin/gitweb.cgi/m1-debian but I'm not shipping it for the time being.
21:33GlanzmannUntil it is merged into the asahi branch.
21:34GlanzmannBut the speakers and the jack work.
21:40Glanzmannmarcan: Maybe you want to have a look at this: https://pbot.rmdir.de/l_4gY-U6xwrKxa0yzWvzQA - You can reproduce it with this config: tg.st/u/config-debian-distro-kernel-2022-03-09-4k
21:43user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
21:44amarioguy has joined #asahi
21:47eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
21:48mofux[m]is there documentation / guide somewhere on how to try the current installer? i'm feeling adventurous :)
21:54csileeeeeeeeeeeeeeeeeeeeeeeeoe has quit [Remote host closed the connection]
22:15landscape15[m]mofux: read the wiki on GitHub. There is all the information you need.
22:50gladiac has quit [Quit: k thx bye]
22:57amarioguy has quit [Ping timeout: 480 seconds]
22:59Gues__________________________ has joined #asahi
23:03tpw_rulescould the m1n1 stuff be taught to fwupd?
23:06tpw_rulesand/or use the efi updatecapsule mechanism?
23:07ketteniswithout efi variable support that is hard
23:07tpw_rulesi think you can put it in a specific directory
23:13kettenisah right, that might be good enough
23:17user982492 has joined #asahi
23:27artemist has joined #asahi
23:37pent1ckel has quit [Remote host closed the connection]
23:38pent1ckel has joined #asahi
23:39mofux[m]landscape15: thx, got it working on M1 Air (alarm boots)
23:40mofux[m]seems there is no easy way to connect to wifi yet on the alarm base (wifi-menu not working because dialog is missing)
23:41mofux[m]also sometimes the login hangs after reboot. and sometimes wlan0 is not there after reboot
23:41mofux[m]but i assume it's too early to report these?
23:53j`eywlan0 not coming up sometimes is known, for wifi there should be `iwd`
…