<tpw_rules>
zzywysm: the linux kernel does not support it for aarch64
Catyre has joined #asahi
<tpw_rules>
not 100% sure why, but there's no way to just make a self-decompressing bzImage like you can on x86_64
<zzywysm>
tpw_rules: thanks, i was really baffled
grgy has quit [Remote host closed the connection]
<zzywysm>
tpw_rules: spent a while digging to see if maybe i disabled some key bit of kernel config that turned it off, but i didn't see anything obvious in the Kconfig definitions
grgy has joined #asahi
Catyre has quit [Ping timeout: 481 seconds]
grgy has quit [Remote host closed the connection]
grgy has joined #asahi
grgy has quit [Remote host closed the connection]
grgy has joined #asahi
grgy has quit [Remote host closed the connection]
grgy has joined #asahi
grgy has quit [Remote host closed the connection]
grgy has joined #asahi
grgy has quit [Remote host closed the connection]
zzywysm has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dimbulb has joined #asahi
phiologe has joined #asahi
PhilippvK has quit [Ping timeout: 480 seconds]
grgy has joined #asahi
grgy- has joined #asahi
marvin24 has joined #asahi
grgy| has joined #asahi
grgy has quit [Ping timeout: 480 seconds]
derzahl has quit [Remote host closed the connection]
marvin24_ has quit [Ping timeout: 480 seconds]
grgy- has quit [Ping timeout: 480 seconds]
grgy| has quit [Ping timeout: 480 seconds]
grgy has joined #asahi
kov has quit [Quit: Coyote finally caught me]
zzywysm has joined #asahi
zzywysm has quit []
<tpw_rules>
trying to think through a story about firmware re-parsing on nixos because a systemd unit or boot script or whichever aren't really nixos's jam
<tpw_rules>
so just to confirm, the firmware TAR is specific to the stub partition version, right?
<tpw_rules>
and the stub partition is purely a function of the restore IPSW used to create it?
<tpw_rules>
is there any way to get the URL for that IPSW from information on the ESP aside from manually parsing the installer.log? does apple keep them up indefinitely?
<tpw_rules>
it's okay if there's a lookup table involved, i'd need to populate it with corresponding firmware hashes anyway. just not sure how to interpret all the codes and IDs in the URL
<tpw_rules>
wondering if maybe i should have nixos download and parse the IPSW, but that would require mounting an APFS-format DMG, right?
<tpw_rules>
but on the other hand that would be required to build stub partitions from linux so maybe shaving that particular yak isn't bad?
zzywysm has joined #asahi
zzywysm has quit []
<nicolas17>
tpw_rules: you mean the ipsw URL? the UUID in it is completely random and unpredictable
<Glanzmann>
Cy8aer[m]: Don't tell Linus that emacs doesn't work. He will never use the m1, but IIRC he uses a different emancs and also probably in Fedora they already have pulled the new release.
<tpw_rules>
but the UUID stays constant for the lifetime of the ipsw, right?
<tpw_rules>
i find that "active": true property a bit worrying. do they become inactive at some point
<nicolas17>
"does apple keep them up indefinitely?" there's some really odd set of old iOS ipsws that have been deleted for some strange reason, but in 99% of cases the URL is valid forever
<nicolas17>
and I don't think I have seen any *mac* one do that
<tpw_rules>
okay. well like i said i would have to update my stuff with the new firmware hash for each new supported ipsw anyway so the url not being determinate ahead of time is fine. but it looks like i can key off that build code. i think i saw that in a plist
<nicolas17>
so far, the build number like "21F79" uniquely identifies a macOS IPSW, yes
<tpw_rules>
there's still the apfs problem though. i can't use fuse inside a nixos build environment
<nicolas17>
ugh 21F2092 is actually the M2-specific release so that list may be missing M1 stuff you're looking for, I picked the worst version as the example :P
nicolas17 has quit [Ping timeout: 480 seconds]
<tpw_rules>
no i mean like the wifi firmware and things
<jannau>
if you want to use information from the ESP to get the IPSW URL I don't see why you can't use the vendor firmware and asahi/all_firmware.tar.gz directly. that would be much faster than downloading gigabytes of IPSW
ptudor has quit [Ping timeout: 480 seconds]
guillaume_g has joined #asahi
ciggi has joined #asahi
MajorBiscuit has joined #asahi
<Glanzmann>
jannau: I tried to use the asahi tree this morning with my old config += the_new_bluetooth option. I had no wifi, trackpad worked and there were rcu hangs. I'll retry later today with the asah kernel config and see if it improves.
<Glanzmann>
Are there any known regressions with the current asahi tree?
Major_Biscuit has joined #asahi
MajorBiscuit has quit [Ping timeout: 480 seconds]
<_jannau_>
there a few reports of broken systems after the update, unfortunately mostly without details of what went wrong
<ElvishJerricco[m]>
tpw_rules: I've been trying to figure out what I'm going to do about firmware whenever I get to that with NixOS. I *think* what I'm going to do is write some code to run in initrd that copies the firmware archive that the installer prepares out of the ESP and into a directory in `/run`, then change the NixOS code that handles the `hardware.firmware` option so that it looks there instead.
<ElvishJerricco[m]>
samueldr over in the #nixos-on-arm:nixos.org matrix room had another idea though. He suggested that the firmware could be included in m1n1 stage 1 and passed forward in the device tree. Sort of an unconventional use of the DT but it would mean we don't need this weird magic archive and a way to load it at the right time.
<chadmed>
encode the entire firmware as bytes in the DT?
<j`ey>
(DT can be a maximum 2mb)
<ElvishJerricco[m]>
Yep
<ElvishJerricco[m]>
oh
<chadmed>
thats uhhh thats never going to happen
<ElvishJerricco[m]>
well that probably kills the idea right there :P
<chadmed>
this is just my personal opinion here and im probably wrong but like this doesnt really affect anything but the first boot after some feature enablement right
<chadmed>
so not only will this become increasingly rare as all features are enabled OOB, but outside of that it only affects _some_ users and the solution is to just reboot the system once
<chadmed>
do we really need to overengineer a solution to this?
<chadmed>
just make sure the ESP is mounted and the firmware is present before any modules dependent on firmware are loaded (keep them out of the initramfs)
<ElvishJerricco[m]>
Yea I mean that's essentially what my plan was before the weird DT idea came up. I'm just going to have initramfs extract the stuff from the archive found on the ESP
<ElvishJerricco[m]>
tpw_rules was talking about having the package manager on NixOS download the ipsw and get the firmware from there at build time, but I think just trusting the ESP is probably better, since presumably that *directly* corresponds to whatever stub version is installed
<chadmed>
yeah thats also extremely overengineered and time consuming, will honestly take longer than building the kernel or whatever at that point
<chadmed>
the ESP has all the firmware in it, and there is a PKGBUILD for arch that automates finding "new" fw and putting it in the right spot. you should be able to adapt that for nixos
<_jannau_>
chadmed: I have a half-finished asahi-fwextract ebuild. the post-install handling is missing, I ran it manually to get bt and trackpad firmwares
<ElvishJerricco[m]>
ehh nixos is a bit weird. We don't really have the ability to look at things on the ESP at build time because everything is super sandboxed.
<chadmed>
yeah same with gentoo, but what i did for firmware handling was just install a job into local.d that checks on every boot. since the local service starts only after pretty much everything else, theres no race with partition handling
<chadmed>
_jannau_: nice! that should make things much easier going forward. i was trying to cobble together something manually but gave up and was going to get around to doing that at some point myself :P
<ElvishJerricco[m]>
_jannau_: What/where is asahi-fwextract?
<Glanzmann>
chadmed: So I build out of the asahi a debian kernel package. And than I used dpkg -i kernel.deb to install the kernel which takes care of creating the ramdisk.
<chadmed>
is there any way you can tell it to include modules manually?
<Glanzmann>
I see, I'll try that.
<Glanzmann>
chadmed: Yes, it is possible to add modules to be included manually.
<chadmed>
cool, that dracut config fragment has the list of modules you need to boot the asahi config so add those and you should be good to go
<Glanzmann>
I'll try that later tonight and report back. It might be that some of them were included beforehand statically.
<chadmed>
yeah i had the same issues when sorting out gentoo support
<chadmed>
because theyre soft dependencies on each other they wont get picked up explicitly by initramfs generating tools/scripts/depmod so you need to explicitly tell the initramfs builder to add them in
<Glanzmann>
Got it, than I just got lucky before.
<chadmed>
heh yeah i got lucky on my first run with the gentoo script and then on testing a second (and third and fourth) time this issue reared its head
<Glanzmann>
chadmed: Do you know if I need to regenerate the firmware so that wifi works with the lastest release?
<chadmed>
nah wifi will be fine, you only need to add new firmware for bluetooth
<Glanzmann>
I thought so. But when I build the kernel with my old kernel config, wifi was not working, so I was wondering why. Everything else seems to work.
taraiph has joined #asahi
<taraiph>
holy moly there's a lot of people here
<taraiph>
also holy moly it's been a while since I've been on an IRC channel
<Glanzmann>
Sobek[m]: People (including me) report that they can work between 6-10 hours on a macbook air what we currenly have. So that is like 50% to 70% of what is possible with Macos.
<Sobek[m]>
What does PSCI stands for? (Also, I known that marcan has an Ars Technica account, so I thought he could just answer himself instead of having some word to mouth deformation)
<Glanzmann>
Sobek[m]: Also, IIRC noone has figured out how to put the wifi to sleep and power it back up again.
<Glanzmann>
Or it can be turned off but not back on.
<j`ey>
the driver currently crashes while turning off
<Glanzmann>
Oh, than my recollection was wrong.
<Glanzmann>
j`ey: Did sven succeed with is s2idle bt experiments?
<Glanzmann>
his*
<j`ey>
not sure, ive been on vacation!
<maz>
stop bragging! :D
<Glanzmann>
j`ey: I also took two days off after working 3 weeks including weekends and evenings. :-)
<j`ey>
maz: looks like I managed to escape the hottest of the heatwave..
<j`ey>
Glanzmann: i saw that you went sailing :)
<maz>
j`ey: I came into the office just to avoid cooking...
<Glanzmann>
j`ey: You're based in the uk, aren't you? Than you must be also an sailor. How hot was the heat wave?
<j`ey>
maz: The train must be pretty hot
<j`ey>
Glanzmann: I avoid the sea :P today will be high 30C in some parts of the UK
<sven>
it's actually the first time that i regret not living to the office :<
<sven>
+close
<maz>
j`ey: train had aircon, and I took the 6:47 this morning... I'll try a find a napping pod later...
<j`ey>
sven: keep hydrated
<j`ey>
maz: ah smart
<Glanzmann>
j`ey: We have 34C today but the house keeps cool in the lower floors, so no issue.
<sven>
maybe i should try that as well. i don't think my office has napping pods though :(
<dottedmag>
Fun, Malta used to be the hottest country in Europe during summer. Looks like we've exported it to the northern latitudes somehow :)
<j`ey>
apparently the UK is hotter than spain currently..
eroc1990 has quit [Quit: Ping timeout (120 seconds)]
eroc1990 has joined #asahi
<sven>
37C here :/
<_jannau_>
35C here with 38C predicted for tomorrow, inside ok with 26C without aircon
<tpw_rules>
ElvishJerricco[m]: i'm not well-versed in the boot process but doing it as part of the initrd sounds reasonable. although how would we re-parse the all_firmware.tar.gz into firmware.tar if the installer changes? chadmed is right that that will be rare but it will be a thing. just ask the user to run our packaged version of asahi-fwextract?
<ar>
~28°C here, and is supposed to stay like that until friday, but i'm going to netherlands on thursday
<tpw_rules>
it's been >38C here almost every day for the past couple months :D
<j`ey>
:\
<_jannau_>
tpw_rules: fwextract could be integrated with kernel updates. it's be only necessary after kernel updates and just needs to run once for a given version
<_jannau_>
i.e. if the kernel gains a driver which requires new firmware it could depend on a fwextract update
<bestouff[m]>
32°C here, but I'm at 1000m altitude; I think it's >40°C down in the city
Catyre has joined #asahi
Catyre has quit [Ping timeout: 480 seconds]
bisko has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
artemist0 has joined #asahi
bisko has joined #asahi
miko has quit [Quit: WeeChat 3.6]
idf00[m] has left #asahi [#asahi]
Zaprit has joined #asahi
zzywysm has joined #asahi
zzywysm has quit []
tomtastic has quit [Ping timeout: 480 seconds]
Zaprit has quit [Remote host closed the connection]
Gaspare has joined #asahi
bisko has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
doggkruse has joined #asahi
guillaume_g has quit []
bisko has joined #asahi
qyliss has joined #asahi
c10l2 has joined #asahi
c10l has quit [Ping timeout: 480 seconds]
Gaspare has quit [Quit: Gaspare]
<Manouchehri>
Anyone here attempted to run cuttlefish on asahi?
<Manouchehri>
having trouble figuring out why it's segfaulting
<ChaosPrincess>
what does `file /home/dave/cf/bin/cvd_internal_start` say?
<Manouchehri>
/home/dave/cf/bin/cvd_internal_start: ELF 64-bit LSB pie executable, ARM aarch64, version 1 (SYSV), dynamically linked, BuildID[md5/uuid]=32c79baee3900f28eb763cab7a0249a0, with debug_info, not stripped
<tpw_rules>
yeah, the .text is only 4k aligned so the loader cannot map it into memory
<Manouchehri>
thank you :)
<Manouchehri>
could we put this into the docs somewhere (how to check for 4k alignment)?
Catyre has quit [Ping timeout: 480 seconds]
<Manouchehri>
so ./bin/launch_cvd is fine, just not lib64/ld-android.so?
<tpw_rules>
it's sort of an uncommon problem and it seems google specifically has hacks to defeat the default 64k segment alignment set by the major compilers
<tpw_rules>
chrome had the same problem too
<tpw_rules>
no that one is misaligned to
<tpw_rules>
the load address is 00000000000d9000
<Manouchehri>
but isn't 0x00000000000d9000%64 = 0?
<tpw_rules>
it needs to me % 16384 = 0
<tpw_rules>
be*
<tpw_rules>
iirc
<tpw_rules>
i'm sorry, try readelf -l
<tpw_rules>
i misremembered the options i think
Catyre has joined #asahi
<tpw_rules>
all of the LOAD sections need to have align values of at least 0x4000
<sven>
does the config from the asahi branch work?
<Glanzmann>
sven: Nope, that one doesn't even boot.
<Glanzmann>
Probably because of missing module dependencies.
<sven>
uh, sounds like something in your setup is messed up then
<Glanzmann>
sven: Chadmed has the same issue with gentoo.
<sven>
and i'm not going down the road of debugging your .config again, sorry.
doggkruse has joined #asahi
Catyre has joined #asahi
zzywysm has joined #asahi
eroc1990 is now known as Guest5474
eroc1990 has joined #asahi
Gaspare has joined #asahi
<Glanzmann>
So with the asahi kernel config the Debian initramfs does not include a single module. Apparently because 'zst' is used.
Guest5474 has quit [Ping timeout: 480 seconds]
Catyre has quit [Ping timeout: 480 seconds]
<PyroPeter_>
sven: I think I succeded in rebasing your usb3 patches on yesterday's linux-asahi release, but dmesg says "phy-apple-atc 383000000.phy: tunables and/or fuses not available; only USB2 will be supported" and connecting a usb-c dock to either port of my macbook air does not produce any usb-related messages in dmesg, so I must have messed something up, right?
<sven>
hmm…. Probably
<sven>
is that usb-c dock usb3 or thunderbolt?
the_lanetly_052__ has quit [Ping timeout: 480 seconds]
<sven>
only usb3 will work
<PyroPeter_>
it is a usb3 dock, and it works with linux-asahi without the patches
<sven>
maybe you just need some draft m1n1 PR
<sven>
I don’t remember if I pushed the new tunables format to the atc branch
<sven>
as I said, I dont remember if I already pushed the new tunables format to that branch
<sven>
which chip is this? Normal m1?
<PyroPeter_>
yes, old macbook air m1
<sven>
And I assume both ports dont work?
<PyroPeter_>
yes
nicolas17 has joined #asahi
<PyroPeter_>
I don't understand why I only get a message about 383000000.phy, but not the other phy
<sven>
ah, I remember
<PyroPeter_>
the two phys are corresponding to the two ports, right?
<sven>
in line 1072 the ENOENT needs a - in front of it
<sven>
but that should only affect one of the two ports
<PyroPeter_>
oh yes, R patched that, but I failed to rebase their commits so I forget about that
<sven>
but that should only break one port
<sven>
if it still doesn’t work some commit in mainline broke something :/
<Glanzmann>
sven: The standard asahi config works for me if I set CONFIG_MODULE_COMPRESS_NONE=y, so it seems to be a 'config issue'. However only the wifi is affected, everything else works.
<sven>
🤷🏻♂️
<PyroPeter_>
well, I will apply the -ENOENT fix, and I will figure out how to test out that m1n1 PR, and then I will report back. thanks
<Glanzmann>
sven: Can you recommend a font to watch your emojis in a Terminal?
<PyroPeter_>
Glanzmann: noto-fonts-emoji
<sven>
you don’t need that PR, I just double checked
<Glanzmann>
PyroPeter_: Thanks.
<zzywysm>
Glanzmann: did you try enabling CONFIG_MODULE_DECOMPRESS along with whatever module compression scheme you wanted?
<zzywysm>
when you enable that, it inlines the required decompressor into the kernel
<Glanzmann>
zzywysm: The problem appears to be that the initrd file did not contain any modules at all.
<Glanzmann>
Might it be that the zstd stuff is new, and Debian testing lacks the tooling to build a initrd?
<zzywysm>
yeah, it seems much easier to use xz instead of trying to debug the Debian kernel build and packaging process
<Glanzmann>
zzywysm: So when I build the kernel, make-kpkg complained that zstd was missing. So I isntalled 'zstd'. I could also see the modules in /lib/modules/. However the initrd did not contain any modules.
<zzywysm>
was it complaining that the zstd userspace program was missing, or that your kernel config was missing zstd support?
eloy_ has joined #asahi
<zzywysm>
sorry, forgot to tag Glanzmann in my reply
<Glanzmann>
zzywysm: To be honest, I don't know, because I never got console output.
<zzywysm>
Glanzmann: again, i repeat myself: it seems much better to not break your setup than to cause yourself pain intentionally by switching to a trendy compression algorithm
Chinese_soup has quit [Quit: WeeChat 3.1]
<Glanzmann>
zzywysm: I'm not afraid to break my own environment. I'm afraid to break the asahi Debian installs. So I would like to ship a kernel with a debian config + asahi config options. But with that, wifi does not load with the error message 'brcmfmac 0000:01:00.0: brcmf_pcie_init_ringbuffers: Allocating ring buffers failed'.
<zzywysm>
Glanzmann: is CONFIG_BRCMFMAC set to Y or M?
<Glanzmann>
CONFIG_BRCMFMAC=m
<zzywysm>
is CONFIG_FW_LOADER_COMPRESS_ZSTD enabled?
<Glanzmann>
CONFIG_FW_LOADER_COMPRESS=n
<zzywysm>
i'd enable CONFIG_FW_LOADER_COMPRESS* and see what happens
<Glanzmann>
zzywysm: For 5.18 it worked with CONFIG_FW_LOADER_COMPRESS=n
<zzywysm>
Glanzmann: your hands are covered in oil and soot, and now you complain that you don't want to get your hands dirty ;-)
<Glanzmann>
I can give it a try and report back.
<zzywysm>
Glanzmann: i'd also ensure i had things like CONFIG_CRYPTO_ZSTD and CRYPTO_DECOMPRESS_ZSTD enabled
<Glanzmann>
These were already enabled: CONFIG_RD_ZSTD=y CONFIG_FW_LOADER_COMPRESS_ZSTD=y CONFIG_CRYPTO_ZSTD=m CONFIG_ZSTD_COMPRESS=m CONFIG_ZSTD_DECOMPRESS=y CONFIG_DECOMPRESS_ZSTD=y
<zzywysm>
switch CRYPTO_ZSTD to y
<Glanzmann>
compiling.
___nick___ has joined #asahi
___nick___ has quit []
c10l2 has quit []
c10l2 has joined #asahi
___nick___ has joined #asahi
<jannau>
Glanzmann: the issue is most likely that kmod doesn't support zstd, use a different compressor for modules
<Glanzmann>
jannau: I managed to have working asahi kernel config by setting CONFIG_MODULE_COMPRESS_NONE=y.
<zzywysm>
Glanzmann: i use CONFIG_MODULE_COMPRESS_XZ=y and CONFIG_MODULE_DECOMPRESS=y on NixOS and it works fine with wifi
gladiac is now known as Guest5480
gladiac has joined #asahi
<Glanzmann>
But now I try to get the debian distro config + asahi config to work. The issue appears to be 'brcmf_pcie_init_ringbuffers: Allocating ring buffers failed' which points to 'linux/drivers/net/wireless/broadcom/brcm80211/brcmfmac/pcie.c' There three goto fails, now I add printk to find out which one it is.
<zzywysm>
i know that config works with wifi on 5.18
<Glanzmann>
zzywysm: I have a config that works with 5.18, I got the asahi config working for Debian for 5.19 by setting CONFIG_MODULE_COMPRESS_NONE=y. Now, I'm trying to get my 5.18 config working for 5.19.
<zzywysm>
Glanzmann: and how did you do that? did make oldconfig not work?
<zzywysm>
Or make olddefconfig?
doggkruse has quit [Ping timeout: 480 seconds]
<Glanzmann>
zzywysm: I took the old config and set the new option for bluetooth.
<Glanzmann>
This ends up with a system mostly functioning expect wifi.
doggkruse has joined #asahi
<zzywysm>
did you do that by hand, or via the menuconfig system?
<Glanzmann>
by hand and than I run olddefconfig.
<zzywysm>
yeah, you need to stop editing kernel configs by hand RIGHT NOW
<Glanzmann>
Maybe I broke something while doing that, so let me try ith without bletooth.
<zzywysm>
it's expected that you will break your kernel in interesting ways editing it by hand. when you select an option in menuconfig/nconfig, it will also automatically enable other prerequisite features
<zzywysm>
ask tpw_rules how i broke his NixOS machine doing that
<zzywysm>
:)
<Glanzmann>
zzywysm: Hrhr. I expect 'make olddefconfig' to fix it afterwards. But I'll try right now my unmodified config for 5.18 for 5.19.
<zzywysm>
if your kernel still doesn't work right after trying that, give make olddefconfig a spin with the config in the pull request i shared above
<Glanzmann>
I'll do the same.
<Glanzmann>
Okay, if I tell you the case, you're going to shoot me. I have 'CONFIG_ARM64_4K_PAGES=y' but no 4k patch ...
<Glanzmann>
I try 16k pages next ...
doggkruse_ has joined #asahi
doggkruse has quit [Ping timeout: 480 seconds]
doggkruse_ has quit [Ping timeout: 480 seconds]
doggkruse has joined #asahi
<Glanzmann>
sven jannau marcan zzywysm: With the right page size, wifi works ...
Axenntio has joined #asahi
StellarOrbit has joined #asahi
<StellarOrbit>
Hello everyone. i installed via the 3rd option which gives me only m1n1 and u-boot so I can install OpenBSD on my m1 mac mini and I don't have keyboard input on u-boot. Is there any form of configuration I need to get it to work? It's a wired keyboard that can be connected either via usb-a or usb-c and I have tried both with no success
<StellarOrbit>
I need keyboard init so that I can report an OpenBSD snapshot stop error upon boot
<zzywysm>
StellarOrbit: some keyboards are just hopelessly goofy and are never going to work within uboot
<StellarOrbit>
zzywysm I have a few different keyboards. This one is a Keychron C2 keyboard, so would you recommend I try another keyboard to see if that will work?
<zzywysm>
that would be the first thing to try
<StellarOrbit>
ok sounds good. I will be back to try this out.
<zzywysm>
while i have a KeyChron K3, it is rather boutique
<zzywysm>
you'll have better luck with less boutique, more cheap crap from Micro Center
<StellarOrbit>
Gotcha
StellarOrbit3 has joined #asahi
StellarOrbit has quit [Quit: Connection closed]
Axenntio has quit [Ping timeout: 480 seconds]
<tpw_rules>
some of them have a "bios mode" which might help
StellarOrbit3 has quit []
___nick___ has quit [Ping timeout: 480 seconds]
<jannau>
they are gone, but only the usb-c ports will work in u-boot. kettenis is aware of a regression of the current installer's uefi boot binaries with openbsd
<Glanzmann>
StellarOrbit: I have keychron k3 on my mini, which does __not__ work in u-boot.
nela has joined #asahi
zzywysm has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
StellarOrbit has joined #asahi
StellarOrbit has quit []
zzywysm has joined #asahi
ciggi_ has joined #asahi
ciggi has quit [Read error: Connection reset by peer]
StellarOrbit has joined #asahi
<StellarOrbit>
Ok I am back now. I connected a different keyboard and was able to select a different boot and it booted right into OpenBSD, though I admittedly don't know what I did to get back in
<StellarOrbit>
I feel like such a noob even though I have been using GNU/Linux on amd64 for over a decade
<StellarOrbit>
Is it possible that u-boot is trying to boot into a different boot environment that doens't exist?
yuyichao has joined #asahi
<StellarOrbit>
If this is the case, what is the best way for u-boot to choose the right boot source?
ciggi_ has quit [Read error: Connection reset by peer]
ciggi has joined #asahi
<Glanzmann>
StellarOrbit: There are two options, either you only have on efi partition.
<Glanzmann>
Or you use a parameter for m1n1 to tell it the right partition.
<Glanzmann>
StellarOrbit: Don't use my rust m1n1 because it is very old. Build your own or use the one from the asahi installer.
<Glanzmann>
one* efi partition.
yuyichao has quit [Ping timeout: 480 seconds]
taraiph has joined #asahi
<taraiph>
good afternoon
<StellarOrbit>
Glanzmann I used the asahi install script and chose option 3 to get m1n1 and u-boot for uefi and installed OpenBSD via the install image I wrote to my usb via dd
<Glanzmann>
I see.
<Glanzmann>
So the OpenBSD u-boot doesn't have the choosen.asahi,efi-system-partition patch in it.
<Glanzmann>
So probably just make sure that you have only one efi parition.
<StellarOrbit>
I am using u-boot from asahi I assume because u-boot loads, it shows the asahi linux logo and boots into openbsd from there
<Glanzmann>
StellarOrbit: I only installed OpenBSD once, it was more or less straight forward, but I installed it into the internal nvme.
yuyichao has joined #asahi
<Glanzmann>
StellarOrbit: IIRC u-boot boots from the first efi partition it finds. First looking on the nvme and than on usb.
<StellarOrbit>
Glanzmann that's what I did. I wrote the installer to the usb and booted into the installer from asahi u-boot and installed it normally and rebooted into it just fine from u-boot. It was when I upgraded to the snapshot that I had the issue
<StellarOrbit>
It was installed to the internal nvme
<Glanzmann>
Ah, I see.
<StellarOrbit>
I am writing to you from OpenBSD installed on here. I typed boot and chose something random (kinda just spammed keys) and it got me back in
<Glanzmann>
Looks good to me, but check with kettenis.
<Glanzmann>
It also appears that you have only on esp parition (the msdos one).
<kettenis>
yes, that looks ok
<StellarOrbit>
Glanzmann true. I think I am booted into the old kernel at the moment because I just ran sysupgrade just to see what happens and it tried to download the sets again
<kettenis>
the "unknown" partitions are the various apfs partitions
yuyichao has quit [Ping timeout: 480 seconds]
<StellarOrbit>
kettenis right. What do you think the issue could be? I am gonna reboot to see what happens if I let it boot normally again
StellarOrbit44 has joined #asahi
StellarOrbit has left #asahi [#asahi]
<kettenis>
sorry, but your description of what you did is too vague
<StellarOrbit44>
kettenis whatever the case, it works now. I just rebooted just fine
<taraiph>
has anyone gotten games to work on asahi?
<taraiph>
because I'm new to linux as a whole and I think I'm a bit deep
<StellarOrbit44>
Thanks for your help kettenis and Glanzmann !
StellarOrbit44 has quit []
<Glanzmann>
jannau: Where do I get the firmware for the touchpad/keyboard? I tried to rerun the asahi installer, but I did not see it. Debian is complainging while building the ramdisk: W: Possible missing firmware /lib/firmware/apple/tpmtfw-*.bin for built-in driver dockchannel_hid
<Glanzmann>
I assume that is the firmware for the keyboard/touchpad.
<probie_>
taraiph: what do you mean by "games"? Something like Chess, Go or various card games will probably work. The lack of GPU acceleration is likely to be an issue for anything more involved
<j`ey>
Glanzmann: are you on m2?
<Glanzmann>
Nope.
<Glanzmann>
Are, so the firmware is only for the m2, now I get it.
<taraiph>
I see, so steam is out of the question
<Glanzmann>
And it will only be extracted on the m2.
<taraiph>
I was wondering why there was no sign of multilib anywhere
<jannau>
Glanzmann: mtp firmware requires a stub with m2 support, that is currently only the the special mac os 12.4 for m2
<Glanzmann>
For the Debain users I updated the kernel. You can get it using 'curl -sL https://tg.st/u/ksh | bash' I only tested it on the macbook air m1. It no longer has the 4k patch.
<Glanzmann>
jannau: Ah, okay.
taraiph has quit [Quit: Page closed]
<Glanzmann>
jannau: Can I run the 'mtp firmware' also on the m1, if I get access to a 'm2 stub'?
<sven>
why would you even want to do that?
<Glanzmann>
sven: My question boils down if the mtp firmware is only for the m2 or can also be used with the m1 touchpad/keyboard?
<j`ey>
only for the m2
<Glanzmann>
Thanks.
yuyichao has joined #asahi
<osteoblast22[m]>
Is there an alternative(fork?) to mariadb that works on asahi linux ?
yuyichao has quit [Ping timeout: 480 seconds]
<clover[m]>
do you mean, are there linux aarch64 binaries for mariadb?
<osteoblast22[m]>
No, mariab crashes on asahi linux because it uses "jemalloc" which doesn't work because of the 16k pages
<osteoblast22[m]>
so maybe there is another mysql server that supports 16k pages?
<clover[m]>
have you tried postgres? :)
<osteoblast22[m]>
I'm not developing a new app, trying to run an old app which requires a mysql server not postgres
<clover[m]>
i see
rurza has joined #asahi
<psykose>
if you can tweak the jemalloc build options that would fix it
<psykose>
sounds easier than using an entire other database
rurza has quit []
<jannau>
it should be enough to build jemalloc/mariadb on a apple silicon system with 16k page size
Gaspare has quit [Quit: Gaspare]
nicolas17 has joined #asahi
<osteoblast22[m]>
I used PKGBUILD from this pr and it worked, thanks
<nicolas17>
looking at the chat log, "<taraiph> I see, so steam is out of the question, I was wondering why there was no sign of multilib anywhere", I sense a much deeper misunderstanding here >.>
<nicolas17>
isn't multilib what you need to run x86 games on x86_64 Linux distros? if the game is x86 then multilib won't be enough to make it work on m1 anyway
<j`ey>
and theres no 32 bit arm multilib since m1 doesnt support 32bit!
<nicolas17>
that too
<nicolas17>
does steam even support arm linux, of either bitness?
<osteoblast22[m]>
Has anyone got hostapd to work with macbook's wifi ?