gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
<nicolas17>
gladiac: fix your connection >.>
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
<nicolas17>
gladiac: fix your connection >.>
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
gladiac has joined #asahi
gladiac has quit [Remote host closed the connection]
GustavBlty[m] has joined #asahi
jabashque has joined #asahi
greguu has joined #asahi
riker77_ has joined #asahi
riker77 has quit [Ping timeout: 480 seconds]
riker77_ is now known as riker77
witchbutter has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
PhilippvK has joined #asahi
derzahl has joined #asahi
phiologe has quit [Ping timeout: 480 seconds]
<derzahl>
anyone get chromium to work yet? I had no luck with 100 and 101 after updating for 16k kernel pages. Trying to compile chrome canary 104...I think that should have the 16k page patch built in....
<chadmed>
the patch is in 102
Telvana2 has quit [Read error: Connection reset by peer]
Telvana has joined #asahi
marvin24_ has joined #asahi
marvin24 has quit [Ping timeout: 480 seconds]
nicolas17 has quit [Ping timeout: 480 seconds]
kov has quit [Quit: Coyote finally caught me]
<derzahl>
chadmed: ok thats what I thought. any known tests on asahi?
<chadmed>
not specifically chromium v102 but im pretty sure the patch was tested before it was submitted
chadmed has quit [Quit: Konversation terminated!]
<derzahl>
well one would think so. perhaps not necessarily on apple hardware. dunno. but I have not seen any mention of results. no packages or pkgbuilds specifically for asahi
<jonaburg[m]>
chadmed: this is awesome work in regards to the audio sink config for the 14/16 inch m1 macbook pros. also super cool to understand why it's like that and what's happening
<jonaburg[m]>
something really cool in the asahi community is how the knowledgeable share their knowledge with everyone and explain it for the curious
<chadmed>
heh thanks :) just need to do the same for the other machines when i have the funding/time available to do so. i think it would be interesting to do the mac mini and mac studio, will be v challenging
<jonaburg[m]>
should the feature list not be updated to now include working audio? i assume the plan is to bake this in the installer right?
<chadmed>
i have it at "read the note
<chadmed>
because its not enabled by default for end users
<chadmed>
and we cannot reasonably expect end users to modify their devicetree and rebuild m1n1
<chadmed>
also we dont *want* them to do that because they could explode their machine still
<chadmed>
povik's working on a new patch series that introduces a hard volume limit in the kernel so that cant happen
jack[m]12345 has joined #asahi
<chadmed>
fwiw in my testing ive yet to actually make the machine explode even after blasting bass through the tweeters for a few seconds so its probably reasonably safe but im not in a position to be able to run the test to possible destruction
<chadmed>
for me personally, not exploding after ~2 seconds at max volume is good enough. that's simply not good enough for end users to daily drive though
<jannau>
judging by the boot chime the speaker in the studio is in a different league than the min1
<chadmed>
hence why im keen to get my hands on one :D
<chadmed>
i was going to buy one at launch but the wait times were over a month here in australia and my car needed some TLC so it went on the backburner
<tpw_rules>
chadmed: did you do a blog post on the speakers or something?
<chadmed>
tpw_rules: i dont blog because im nowhere near important enough lmao but i made a couple of tweets about it
<tpw_rules>
ah
<tpw_rules>
i need to set up like a github pages blog
* jannau
wouldn't buy a mac studio for the speaker
<chadmed>
well a mac studio would sure as heck beat the pants off this 2017 dual core macbook pro im using as a desktop with a lenovo thunderbolt dock
<jannau>
I heven't bothered to add the audio nodes to the mac studio dts'
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Telvana has quit []
Telvana has joined #asahi
bps has quit [Ping timeout: 480 seconds]
eroux has joined #asahi
bps has joined #asahi
jluthra has quit [Remote host closed the connection]
jluthra has joined #asahi
voj343[m] has joined #asahi
SorunomeLuckySkies[m] has joined #asahi
conradev has quit [Quit: -]
conradev has joined #asahi
joske has joined #asahi
<joske>
chadmed: is there also DSP work needed for the 13 inch models (MBA en MBP13)?
<Dcow[m]>
joske: yes, those have different speaker configuration
<joske>
Dcow[m]: thx, I guess no-one is working on that
<joske>
yet
<Dcow[m]>
yep
<Cy8aer[m]>
I have seen this: https://github.com/chadmed/asahi-audio and as I understand there is a waveform added to the payload audio in realtime, right? And there are waveforms for the mbp14 right now and mbp16 is just interpolating that the data of the mbp14 is working for it too somehow? How were these waves generated (really measured or calculated?)
<joske>
Cy8aer[m]: IIRC he measured these (and wrote a detailed explanation somewhere how he did so)
<chadmed_>
the notes folder in the repo has experimental notes outlining what i did
<chadmed_>
hopefully theyre detailed enough for someone like m arcan to reproduce on machines i dont have/cant get access to myself
kov has joined #asahi
<Cy8aer[m]>
And I have a mbp14 and therefor the right platform 🙂
<Cy8aer[m]>
But I still wait for the save and switched on audio support with the right output levels before I try it.
timokrgr has quit [Quit: User left the chat]
timokrgr has joined #asahi
<beep-boop[m]>
Have you guys done any tweaks on keyboard settings and shortcuts to make it behave more like macos?
<Cy8aer[m]>
yep should be - just prove it with dconf editor.
<Cy8aer[m]>
(what should I do without emacs bindings...)
<beep-boop[m]>
Cy8aer[m]: I cant see this entry in dconf
<Cy8aer[m]>
mom, just googled it, and that was from 2007. But now I found the emacs option you showed from the tweak tool. That is the paramater I clicked.
<Cy8aer[m]>
It was mouse and keyboard.
<j`ey>
Cy8aer[m]: who you calling Mom? :P
<Cy8aer[m]>
just a moment - may be a german phrase, sorry...
<j`ey>
no need to be sorry, just looked like you were saying mom/mum :P
<Cy8aer[m]>
Inter-Lingual chatting - that is interesting.
Axenntio has joined #asahi
<Sobek[m]>
Also didn't marcan ask people to actually provide feedback about which keyboard layout matched hardware best and what bugs there were ?
<Cy8aer[m]>
Just learned today that "verkehrsverbund" can be found in english like "kindergarten". "Verkehrsverbund" - Something that does not work in Germany - but that is off topic.
<Cy8aer[m]>
But the emacs settings are some gui standard on all archs. I was happy to have `/sys/module/hid_apple/parameters` for having the keyboard more ISO and that works fine for me.
<Cy8aer[m]>
and non general tweaks should be in the `hid_apple` module then?
<j`ey>
yup
<Cy8aer[m]>
I needed `/sys/module/hid_apple/parameters` for emacs - because I had problems to type M-x without twisting my fingers...
minecrell has quit [Read error: Connection reset by peer]
minecrell has joined #asahi
Bai-Chiang has joined #asahi
m42uko has quit [Quit: Leaving.]
eroux has quit [Ping timeout: 480 seconds]
eroux has joined #asahi
Axenntio has quit [Ping timeout: 480 seconds]
AstrOwO[m] has joined #asahi
AstrOwO[m] is now known as AstrOwO[m]1
m42uko has joined #asahi
chadmed_ has quit [Remote host closed the connection]
joske has quit [Quit: Leaving]
darkapex1 has joined #asahi
darkapex has quit [Ping timeout: 480 seconds]
Gaspare has joined #asahi
guillaume_g has quit []
bps has quit [Ping timeout: 480 seconds]
ciggi has quit [Read error: Connection reset by peer]
Gaspare has quit [Quit: Gaspare]
bps has joined #asahi
RevHelix has quit [Read error: Connection reset by peer]
RevHelix has joined #asahi
<derzahl>
as far as optiimizing gcc builds for the cpu, what is best practice? just "-march=armv8.5-a"? I dont think gcc has any more specific definition for the m1 architecture, right?
Telvana has quit []
Gaspare has joined #asahi
Gaspare has quit []
nicolas17 has joined #asahi
Gaspare has joined #asahi
<mps>
I need to delete data from partition on mbp to send it to service. not sure which partitions I have to wipe
<mps>
for linux one is not problem, but don't know which of apfs
<tpw_rules>
the middle one
ptudor has joined #asahi
<mps>
nvme0n1p2?
<tpw_rules>
should be
<tpw_rules>
gdisk calls the ones you need to keep "iBootSystemContainer" and "RecoveryOSContainer"
<mps>
and `dd if=/dev/zero of=/dev/nvmeon1p2` is enough
<tpw_rules>
heck i'd just use blkdiscard
<mps>
ah, thanks for gdisk hint
<tpw_rules>
i wonder if there's a quick way to reset the ssd keys. but i assume that would require a DFU restore afterward
<mps>
I have to do this over network, display doesn't work and I can't boot to macos because linux is default
<tpw_rules>
what kind of machine? did you crack the screen?
<tpw_rules>
you should be able to navigate the boot picker with the arrow keys iirc
<mps>
tpw_rules: spilled water on it
<mps>
after that everything works except display
ptudor_ has quit [Ping timeout: 480 seconds]
<tpw_rules>
ah ok. i had heard of some reports of the displays being fragile so i wasn't sure
<mps>
gdisk shows p2 as 'Container'
<tpw_rules>
yeah that's the main macos data store
<tpw_rules>
the one you need to zorch
<mps>
tpw_rules: thanks
bps has quit [Remote host closed the connection]
bps has joined #asahi
bps has quit [Ping timeout: 480 seconds]
xfim has joined #asahi
<xfim>
Hello
<j`ey>
hi
<xfim>
I asked a few days ago if it was possible to install Asahi on a Mac Studio, although it was not listed as supported.
<xfim>
And the reply was that it was.
<xfim>
However, I just had my Mac Studio a couple hours ago, and after updating MacOs I went straight ahead with the Asahi installer.
<j`ey>
that was nearly a month ago :p
<xfim>
And the first thing that I find is "This device is not supported yet!"
<xfim>
And I can't continue, even after enabling expert mode.
<j`ey>
xfim: yes, you need to use jannau's installer for now
<xfim>
Is there a way?
<xfim>
Ok, j'ey, what is such installer?
<j`ey>
it's not working in the "official" branches, but jannau's forks work
<xfim>
Do you think using such a forked installer can give me other trouble?
<j`ey>
wait for jannau, they can give you the best advice!
<xfim>
OK, perfect.
<xfim>
Any idea when he/she may be found?
<_jannau_>
xfim: mac studio with m1 max or ultra?
witchbutter has joined #asahi
<j`ey>
xfim: just there ^ :P
<_jannau_>
I haven't really prepared a installer with m1 ultra support
<_jannau_>
I will eat in a couple of minute and read over the installation instructions to see what's necessary to get you started
Glanzmann has joined #asahi
<j`ey>
oh sorry, I thought you had a fork of the installer
<Glanzmann>
Hello, I just wanted to pull the subtiles from todays asahi lina stream but the video has disappeared. Could someone summarize what was accomplished today?
<xfim>
m1 max
<xfim>
Hi jannau!
hellstabber has joined #asahi
<xfim>
The basic model, with max, 32 RAM and 500 SSD
<xfim>
I have been preparing for this moment for a month!
<j`ey>
Glanzmann: more tracing, and then work on the PoC of the m1n1 driver
<Glanzmann>
j`ey: I see, thanks for the quick summary.
<j`ey>
Glanzmann: the first step of the m1n1 driver (initdata) made some good progress
hellstabber has quit []
<xfim>
Thank you very much all of you. Please, take your time to eat properly, come on! :-)
eroc1990 has quit [Quit: Ping timeout (120 seconds)]
eroc1990 has joined #asahi
<Glanzmann>
j`ey: Good to hear. I look forward to have accelerated graphics under linux on the m1.
<j`ey>
Glanzmann: yeh! let the CPU do more useful things ;)
Gaspare has quit [Quit: Gaspare]
<Sobek[m]>
Random thought, is it technically possible to write drivers to support Touch ID under linux ? What are the moving parts required ?
<j`ey>
that will require SEP, which requires some APFS support
<Sobek[m]>
How SEP requires APFS support exactly ?
<nicolas17>
I think the kernel (or userland?) needs to read some files off an APFS filesystem and give them to the SEP
<Sobek[m]>
(Is there anything remotely like SEP that has linux drivers btw or will that likely be a significant amount of work simply becuase such component don't have any subsystem in the kernel and interface ?)
<j`ey>
I think SEP is via rtkit (sven?)
<sven>
no
<sven>
SEP has its own mailbox interface
<j`ey>
excellent /s
<nicolas17>
yeah but that's the communication channel
<sven>
even more fun: at least the sks endpoint also includes a nested IPC protocol because ofc
<nicolas17>
I think Sobek[m]'s question is more like, "does the SEP count as an HSM, and can we make a driver for Linux's HSM subsystem, if there is one? or would it need a whole new API?"
<Sobek[m]>
What would rprevent the installer from grabbing those files and dumping them along with the firmware, and then the kernel can grab those off its usual filesystems and hand them to SEP ?
<sven>
what files?
<nicolas17>
sven: why does SEP support need APFS support?
<sven>
nicolas17: "gigalocker"
<Sobek[m]>
nicolas17: @sven, those hypothetical files
<sven>
???
<nicolas17>
what I said about "read files and send them to the SEP" was 100% a guess pulled out my ass, idk how it works
<sven>
anyway, please figure out a reasonable question that I can actually understand and then i'll try to answer
<nicolas17>
sven: Linux needs to handle reading/writing "gigalocker" stuff from APFS on behalf of the SEP?
<sven>
nicolas17: yes
roxfan2 has joined #asahi
<nicolas17>
yeah, so that's what I naively called "files off an APFS filesystem" :)
<sven>
that gigalocker thing is a single file with a constant size on a special APFS partition fwiw
<sven>
it's essentially a key-value store
<sven>
if we wanted to hack this we might be able to just grab the block ids once and then just ignore APFS altogether :D
<sven>
probably good enough to at least bring up SEP and talk to it from m1n1 ;)
<nicolas17>
that sounds like an acceptable PoC tbh :)
<Sobek[m]>
What sort of "clever" uses of SEP could there be in m1n1 ?
<nicolas17>
reverse-engineering the rest of it :)
<sven>
debugging the whole userland-kernel module-SEP stack will be annoying enough even if we already understand how the SEP communication works
<nicolas17>
what's the mailbox protocol to add a new fingerprint? add one on Mac, while m1n1 is logging everything
<Sobek[m]>
I'm more intrigued by your statement that most stuff would occur in the userland, how would the thing be architectured exactly ? (What are the security implication of most stuff being in userland)
roxfan has quit [Ping timeout: 480 seconds]
<sven>
no idea
<sven>
this is all very far away and something to think about once there's a better understand how the SEP protocol looks like
<sven>
*understanding
<Sobek[m]>
Anyway SEP is unlilely to get investigated until stuff like sound, GPU and Thunderbolt have all been sorted out, isn't it ?
<sven>
at least I probably won't look into SEP before at least thunderbolt
<nicolas17>
if it's going to be done by the same people, then yes those are likely the priorities
<nicolas17>
if someone who isn't working on sound/GPU/thunderbolt wants to look into SEP they're welcome to :)
<sven>
we'll also need to understand how e.g. SEP interacts with the NVMe encryption before we can design how the linux implementation will look like
Gaspare has joined #asahi
Glanzmann has quit [Quit: EOF]
<witchbutter>
Are you guys maintaining a project board or anything with things that need to be worked on? How can I find a way to contribute?
<jannau>
xfim: do you have a second computer where you can build arm64 binaries?
Telvana has joined #asahi
<jannau>
how comfortable are you with building software?
<xfim>
OK, let me go and try to understand it step by step.
<kaprests>
witchbutter: I am by no means involved in the development myself, but my understanding is that the wiki and this room are the things to check out if you want to contribute
<jannau>
xfim: github's ubuntu-latest image apparently already has rustup installed
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<xfim>
Correct me if I am wrong: the big picture is to build this installer, and then to copy the resulting binary to the MacOs and run it. Is that it?
roxfan2 is now known as roxfan
<xfim>
emerging rustup right now, the development version already present in the gentoo tree
witchbutter has quit [Ping timeout: 480 seconds]
<jannau>
xfim: yes, you need to copy installer.tar.gz and data/installer_data.json to macos, unpack installer.tar.gz and run ./install.sh as root
<jannau>
you can select and the the minimal asahi linux distro but it will not boot. you should end up with a m1n1 prompt
<xfim>
And once with the m1n1 prompt?
<j`ey>
xfim: do you have a USB C cable? (can be C<->A..)
<jannau>
the installer creates a fat partition (efi system partition). replace /m1n1/boot.bin on that partition with the one you've created
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<jannau>
either via mac os or you can chainload that boot.bin via m1n1 proxy and the usb cable
<jannau>
xfim: that cable looks like it should work
<xfim>
jannau: OK perfect for the cable. But I need some more time to digest everything: a) build the installer; b) copy it to MacOs, along with the json, unpack and run install; c) select minimal and end up with a m1n1 prompt; d) in the meantime in another machine build m1n1 and make the u-boot-1 ending up with a boot.bin file; e) move boot.bin to the Mac Studio either via USB cable or via m1n1 proxy; Is that more
<xfim>
or less?
<xfim>
There are so many steps that can fail here, wow. So much to learn. The alternative would be to wait a few days to incorporate the changes into the main installer and just go with the "curl ... | sh"? If so, how many days you think it would be needed to incorporate it?
<jannau>
xfim: more or less. there is no transfer over usb but you can remotely boot the binary over usb
<xfim>
jannau: oh, so many options. Building the crossdev system in gentoo is something that I would like to learn anyway, as one of my ideas is to use tho Mac Studio to compile the packages for the laptop. so it is worth doing the investment now. (BTW, does this seems reasonable?). But maybe downloading it is just easier now and I leave the crosscompilation for another time.
Bai-Chiang has joined #asahi
amarioguy has quit [Remote host closed the connection]
<xfim>
jannau: the crossdev part is proving to be quite challenging. Can I ask if the option to wait for Mac Studio to be included in the main installer is expected in the foreseeable future? (I can not see myself right now doing this in a couple of days).
<ElvishJerricco[m]>
So I'm considering adding some infra in NixOS for booting with the Asahi kernel and m1n1, but I'm a tad concerned about the firmware stuff. I remember reading at one point that the Asahi kernel has to be written for a specific version of the firmware. Does that also mean the system has to have been booted with a macOS stub based on that version? Because while I'm happy to write some code for pulling the firmware out of IPSWs in Nix to
<ElvishJerricco[m]>
provide to the kernel (we have good support for automating non-redistributeable firmware), I wouldn't be able to automate setting up the macOS stub (given that I think you have to do that from macOS, not to mention lack of APFS support in Linux)
<j`ey>
ElvishJerricco[m]: do you need to resinstall all the time?
<j`ey>
cant you just setup the stub once?
<ElvishJerricco[m]>
j`ey: Well the concern is updating the kernel. If that one day requires a new version of macOS stub, then we can't automate that in NixOS. If all I need to automate is /lib/firmware on the linux side and stage 2 m1n1, then no problem
<ElvishJerricco[m]>
oh wow Element really screwed up your name...
<mps>
maybe apfs linux driver will mature in foreseeable future
<xfim>
jannau: Tomorrow I will continue. Thank you very much for your support.
<ElvishJerricco[m]>
mps: I'm not holding my breath :P
<j`ey>
ElvishJerricco[m]: im hoping that the kernel will be backwards compat with the firmware
<mps>
well, I'm used it for reading, didn't had issues
<ElvishJerricco[m]>
j`ey: Ah, so the goal is that any firmware that was previously supported with Asahi will remain supported in future versions? That certainly helps a lot, but *then* we might have /lib/firmware not matching macOS stub. I don't know if that would be a problem.
<j`ey>
ElvishJerricco[m]: Im not sure if that's the goal or not.. but I think it should be.. depends if the interface changes in the fw are easy enough to work around
<mps>
ElvishJerricco[m]: this is rarely problem on all other drivers, but I'm not sure about apple
<ElvishJerricco[m]>
mps: Yea I don't know if the same stability can be expected from a platform like Apple Silicon
<ElvishJerricco[m]>
Just on the basis that it's literally only designed to run one OS, and that OS manages installing the firmware too
<mps>
for mwifiex wifi on newest kernel I use FW old about 7-8 years
<j`ey>
ElvishJerricco[m]: you wont need to do anything with the fw in nixos, the asahi installer does it.
<mps>
if the hardware didn't changed then old FW shouldn't be problem for new kernels
<ElvishJerricco[m]>
j`ey: NixOS is going to want to manage /lib/firmware
<j`ey>
ElvishJerricco[m]: sure, asahi puts all the firmware on the ESP
<j`ey>
you can move it to that folder
<tpw_rules>
ElvishJerricco[m]: did you see my work?
<ElvishJerricco[m]>
tpw_rules: Yea. I'm looking into how we might integrate something like it into nixos proper
<ElvishJerricco[m]>
But for that I want a level of certainty that firmware versions and macOS stub versions won't be problematic
<tpw_rules>
ElvishJerricco[m]: ahh, great. be sure to keep me in the loop :) the way the asahi stuff expects it to be managed is copying to the ESP like j`ey said
<tpw_rules>
so there at least has to be enough cross-version support to boot into a new version and get the script running to copy new firmware
<tpw_rules>
like the asahi arch install proper does manage /lib/firmware, the installer doesn't do anything to that at all except dd a filesystem which contains that directory, but it's empty for legal reasons
<tpw_rules>
the bigger thing that i don't like is the m1n1 being tied to the kernel
<j`ey>
m1n1 tied to the kernel?
<ElvishJerricco[m]>
m1n1 is tied to the kernel? I thought we could just use it to boot u-boot and then EFI boot the OS
<tpw_rules>
m1n1 has the kernel's device tree
<j`ey>
well now that m1n1 can chainload itself, thats not so important
xfim has quit [Ping timeout: 480 seconds]
<tpw_rules>
i think that problem will lessen over time as things stabilize. i have it managed to some degree with boot.loader.*.extraFiles but if you install a broken one you're hooped and nixos can't roll it back
<j`ey>
tpw_rules: are you chainloading m1n1?
<tpw_rules>
yes
<ElvishJerricco[m]>
So I'm still not clear on my original question. To what degree is the Asahi kernel tied to the version of macOS stub that's installed, and the version of the firmware intended to be copied to /lib/firmware?
<tpw_rules>
even in arch this is a problem: how do you have two different versions of your kernel in the grub menu, each of which need a different device tree?
<tpw_rules>
my 2 cents says it can't be so tied that it's unbootable, because the asahi kernel needs to boot to copy new stuff from the stub to /lib/firmware
<mps>
hm, which FW is needed to boot
<tpw_rules>
none currently
<ElvishJerricco[m]>
wait the Asahi kernel copies stuff from the stub? It can read APFS?
<j`ey>
ElvishJerricco[m]: no
<j`ey>
ElvishJerricco[m]: the installer does
<tpw_rules>
sorry i miswrote there
<j`ey>
the installer copies fw to the ESP
<ElvishJerricco[m]>
got it
<ElvishJerricco[m]>
and the OS loads firmware from ESP I guess
<tpw_rules>
no, there's just a script that runs every boot to copy from the ESP to /lib/firmware
<j`ey>
(in the reference distro)
<ElvishJerricco[m]>
right
<mps>
first time when used wifi I made tarball of firwmare, copied it to linux and untared there