ChanServ changed the topic of #asahi-alt to: Asahi Linux: porting Linux to Apple Silicon macs | User-contributed/unofficial distribution ports | Logs: https://alx.sh/l/asahi-alt
<zerdox>
chadmed: I am not sure about kernel config files, I guess I did not set up anything of that. Commands you see in pastebin are almost complete.
<chadmed>
hm okay
<chadmed>
interesting
<chadmed>
because i do remember running into similar issues when first developing all that stuff but ive done multiple installs since and havent hit it
<chadmed>
and the only fix i encountered was kernel config related
<chadmed>
very odd indeed...
<zerdox>
chadmed: I have no idea how I can run mismatched kernel version in asahi. I upgraded the whole system and rebooted before bootstrapping
<zerdox>
Am I mounting partitions right? Do I need to do something with /boot? Is it ok to rm -rf whole asahi partition before installing Gentoo?
<zerdox>
This is the things I am not sure about cause they are not documented anywhere. When I run install script it says like "make sure your /boot is mounted anywhere you want and efi partition is mounted to /boot/efi. also add it do fstab". This is what I am unsure about
<zerdox>
Feels like I am just missing something obvious in my installation
<chadmed>
no that should all be fine, theres nothing specific you need to do in that regard
<chadmed>
only different thing i do is instead of rm -rf / on the asahi partition, i clobber it and make a new one
<chadmed>
its hard to really understand WHAT is causing the error because you have no kernel debug output
<chadmed>
if you have access to /boot somehow (virtualbox can be passed through a pointer to raw block devices on macos hint hint) you can add earlycon and debug to your kernel command line
<chadmed>
and that should at least give us an idea of whats causing the problem
<ChaosPrincess>
does your gentoo install do ESP as /boot or separate
<chadmed>
separate by default because people are used to it, hence mounting the ESP at /boot/efi
<chadmed>
i run ESP as boot on my own machine though and will switch it over to do that once/if asahi does the same
<ChaosPrincess>
Yea, if separate, then you need to do tricks
<ChaosPrincess>
i run esp on boot and its just a diskutil mount away
<chadmed>
yeah i had to do some surgery to /boot on my x86 mac in macos today, it was an ordeal :P
<chadmed>
for context zerdox's issue is the kernel halting with no debug info on first boot
<chadmed>
id be curious to see if the WDT kills the machine after 2 minutes
<chadmed>
if it doesnt then the kernel isnt actually halted and the issue is simpledrm
<zerdox>
i could wait for 15 minutes and it was hanging as is. if i understood right
<chadmed>
hmm. if the kernel was halted the WDT _should_ reboot the machine after a couple of minutes
<chadmed>
so i think the machine is alive, its just not happy with the firmware framebuffer driver
<zerdox>
chadmed: my /boot dir is empty except content in /boot/efi. is it really ok?
<chadmed>
no, your kernel and initramfs should be in there
<chadmed>
what did you mount to /boot when you installed gentoo?
<zerdox>
nothing. i created /boot/efi and mounted efi partition here
<zerdox>
that what i was saying earlier. I don't think this is the right way
<zerdox>
you can check pastebin again to ensure
<chadmed>
yeah
<chadmed>
i mean
<chadmed>
your commands there are fine
<chadmed>
if your /boot is on your root partition then there should be content there. how do you know its empty if you cant access linux?
<zerdox>
i mean its empty before i run install.sh. and before i run install I don't mount anything to /boot except /boot/efi
<chadmed>
oh right yeah thats fine
<zerdox>
ok
<zerdox>
i could try to debug kernel if i got a bit more information how to do it properly
<chadmed>
well youd need access to your /boot somehow so you can add kernel command line parameters
<zerdox>
grub won't help with it as i remember from that github issue
<chadmed>
if you have access to grub then yeah you can add command line parameters from there
<chadmed>
take "quiet" out and add "earlycon debug" in its place
<zerdox>
yes. before choosing gentoo linux i am getting grub loader
<zerdox>
ok. i'll do it rn
<chadmed>
awesome
<zerdox>
chadmed: linux params i got by default:
<zerdox>
linux /boot/vmlinuz-6.2.0-rc3-asahi-ARCH root=UUID=<my-uuid> ro
<zerdox>
do i need to add earlycon degub here? i can't see any quiet here
<chadmed>
add "earlycon debug" to that
<chadmed>
yeah it should still print something
<zerdox>
the difference is that now I got "Booting a command list" on top. everything except this is the same, no useful output
<zerdox>
i guess booting a command list is just a message from grub. nothing changed in fact
<chadmed>
yeah so the kernel isnt able to write to the fw framebuffer
<chadmed>
is your machine connected with ethernet or wifi
<chadmed>
and did you add sshd to a runlevel before rebooting out of the livecd
<zerdox>
my machine is connected to wifi, i share it from my android phone. i did not add sshd to a runlevel before rebooting
<chadmed>
ahhhh youre kind of up shit creek then
<chadmed>
last hope would be sshing into the machine and looking at dmesg
<zerdox>
i can reinstall again (๑•﹏•)
<zerdox>
but i am not sure how this helps. once i exit livecd i dont have any access to os
<chadmed>
one thing that has changed is i have stabilised a newer version of the kernel since the other day
<zerdox>
i did previous install 2 or 3 days ago i guess. so next time i add ssh and you will take a look?
<chadmed>
no im not sshing into your machine, youd do it yourself from your local network
<chadmed>
i might move the script to do ESP-as-boot by default
<chadmed>
just so if this shit keeps happening its 1000x easier to play with m1n1/kernels
<chadmed>
its late here so im not ognna do anything tonight but ill make some changes to the script tomorrow and push them
<zerdox>
i don't really understand the point. in what state of my machine i would ssh to and why. if I can ssh to a machine that it somehow works and i can see everything without ssh. If I get current state (even if i got added sshd) I dont have any access to the OS or even livecd anymore.
<chadmed>
ok so remember before i said if the machine had actually crashed then it would have rebooted itself after two mins
<chadmed>
the fact that its just frozen on that screen means that the kernel _might_ not have actually crashed, and your machine is running fine behind the scenes
<chadmed>
sshing in would confirm this
<chadmed>
and you would be able to get the dmesg log and paste it somewhere for me to look at
<zerdox>
hm.. i could ssh even without logging in first? if yes, ok then
<chadmed>
yeah ssh spawns a new session on the machine, provided sshd is set to load on boot
<zerdox>
i can try to do it tomorrow after you push the updates. but i need some help with it, i don't really know how to do it. i guess i also need to enable all services and set config for wpa_supplicant so it would get network connection before i can ssh
<zerdox>
and i mean i need all this done before i exit livecd
<chadmed>
if you follow the handbook it shows you how to do all this
<chadmed>
because its considered a standard install to have all this stuff set
<chadmed>
also use iwd for wifi its more reliable and supported
<chadmed>
its 2023 society has progressed beyond the need for wpa_supplicant
<zerdox>
ok, thank you. i will try to use it on livecd next time and on arch also
<zerdox>
i usually drop handbook when it's time to configure kernel, as asahi tutorial on wiki says. so i just reboot after install script
<chadmed>
echo "This script will now exit. Continue setting up your machine as per the"
<chadmed>
echo "Gentoo Handbook, skipping the steps related to setting up the kernel or"
<chadmed>
echo "GRUB as these have been done for you."
<chadmed>
this might have caused you some problems tbh...
<zerdox>
there are some necessary steps?
<zerdox>
i posted my whole installation process, really. thats almost the full list of commands related to my gentoo installation
<chadmed>
you need to configure a user, syslog, networking, fs tools, cron, portage options, some system services
<chadmed>
like i said from the outset, its hard to diagnose and fault find when youre not following the instructions...
snyanz has quit [Quit: Konversation terminated!]
<zerdox>
why do i need to do it when i can do it after booting to gentoo? the only thing i do now is downloading networking tools so i can get online later. second time i tried to install gentoo i also started configuring the system, as you said. i added user, set up networking, fs tools, cron and all that (because second time i was much more confident in what i am doing and i was sure i will get my os). but i also got unbootable and was very disappointed because
<zerdox>
spent about hour or more to configure the system which i don't have access to and which doesn't affect will I boot or not
<chadmed>
because thats how installing gentoo works? youre starting from literally nothing, and by default you get nothing out of the box except a toolchain, package manager and coreutils
<zerdox>
the most important staff later in handbook is configuring kernel which i skip, configuring bootloader which i also skip (i would like to use refind later tho). and configuring the system which i can do later when I login. the only necessary that seems to me things are installing networking tools and setting up fstab which is just genfstab
<chadmed>
thats the spirit of gentoo
<chadmed>
refind doesnt work properly on arm64 macs anyway
<chadmed>
ive tried it a couple of times
<chadmed>
i use systemd-boot
<chadmed>
anyway, you cant just skip steps youve been asked to do because you personally dont think its meant to work like that
<chadmed>
if you want a more user friendly out of the box experience just stick to arch
<zerdox>
but... not setting up crone and not creating the user won't stop from booting the system like this. this is not the issue
<chadmed>
probably not yeah but the point is that youre not getting support unless youve followed the basic instructions laid out before you lmfao
<zerdox>
i already used arch for about a year and then i got to void linux which was much better. now i decided to never go back to systemd. this is the reason why i decided to use gentoo and not just asahi
<chadmed>
and fair enough, but im not going to act as guarantor for your system when i have no idea what state its in
<chadmed>
this is such a basic fundamental tenet of tech support
<chadmed>
if you want to go off script because you think its fine then go ahead be my gues
<chadmed>
there wouldnt even be any gentoo infrastructure in the first place if i didnt have the same thought myself
<chadmed>
but that usually comes with the understanding that youre on your own to fix any issues that crop up
<zerdox>
ok. i will post one more installation including all of the commands where i will set up user, install wifi tools, set crone, time sync which won't affect my boot process...
<chadmed>
yknow what
<chadmed>
im going to make the changes to the script right now
<chadmed>
done
yuka_ has joined #asahi-alt
yuka has quit [Remote host closed the connection]
rudi has joined #asahi-alt
rudi has quit []
_rudi has joined #asahi-alt
<MichaelLong>
*yay* it is a long time ago I had to setup gentoo from scratch, my current systems are so old by now...
<chadmed>
my server root fs is at least 6 years old
<chadmed>
i dont mind the gentoo install experience tbh, i find it therapeutic
<MichaelLong>
had to figure out basics like not to suspend on lid close (I'm still running without graphical environment)
_rudi has quit [Ping timeout: 480 seconds]
<chadmed>
just disable suspend in the kernel problem solved :P
<MichaelLong>
nah it is on the /etc/systemd/logind.conf HandleLidSwitch=ignore :)
<chadmed>
ahhh systemd issue
rudi has joined #asahi-alt
rudi is now known as Guest10285
Guest10285 is now known as _rudi
_rudi has quit [Quit: Fell off the bus]
zerdox has quit [Ping timeout: 480 seconds]
compassion has joined #asahi-alt
compassion has quit []
compassion has joined #asahi-alt
compassion has quit []
compassion has joined #asahi-alt
_rudi has joined #asahi-alt
deteg1337 has joined #asahi-alt
deteg1337 has quit [Remote host closed the connection]
janrinze has joined #asahi-alt
<janrinze>
another round of bugfixes from Asahi Lina has been submitted. @glanzmann will we see them soon for Debian?
_rudi has quit [Read error: Connection reset by peer]
<mps>
janrinze: does it builds on debian? I got '../src/gallium/drivers/asahi/agx_state.h:36:38: error: '__VA_ARGS' undeclared (first use in this function)' error on alpine linux
zerdox has joined #asahi-alt
<jannau>
mps: yes, it will compile. the error is inside #else after '#ifdef __GLIBC__'
<jannau>
s/__VA_ARGS/__VA_ARGS__/
<mps>
jannau: in src/asahi/compiler/agx_compile.c
<mps>
hm, src/gallium/drivers/asahi/agx_state.h
<jannau>
src/gallium/drivers/asahi/agx_state.h line 36
<mps>
uh, did I really upgraded kernel after build :) lets check again
<jannau>
the UAPI change is not in asahi-6.2-12
<janrinze>
jannau: which kernel update do I need? i have linux debian 6.2.0-asahi
<janrinze>
aha.
<janrinze>
Ok. building mesa first and then see how i can do a kernel build.
<mps>
things becoming stable so I stopping to be careful on building and installing
<janrinze>
hopefully grub will offer to boot the older kernel as a fail over if my newer kernel fails.
<janrinze>
mps: just my thought. being careful never hurt ;-)
<mps>
janrinze: right, but work on pkgs becoming routine and then attention is lowered
<janrinze>
mps: mesa build was fast!
<mps>
yes
<janrinze>
jannau: when using latest mesa build on older kernel, what would be the worst case scenario?
<mps>
so, kernel tagged v6.3-rc4 will probably have UAPI 1007
<janrinze>
jannau: because if I would have to revert to 6.2 kernel i'd like to know the risks
<jannau>
no, v6.3-rc4 has no asahi driver at all
<jannau>
the uapi change is only in lina's usual branch
<janrinze>
jannau: so i won't be able to test this yet?
<jannau>
best case fallback to software rendering, in the worst case crashing X11 servers / wayland compositors
QuickTimeX has joined #asahi-alt
<janrinze>
jannau: got it. As long as it boots to the console i'm fine.
<QuickTimeX>
Hello channel! Is it possible to install Debian 11 Bullseye instead of Bookworm? The current guide uses bookworm but I need bullseye
<mps>
yes, in my test it used swrast and xorg didn't crashed
<janrinze>
QuickTimeX: It's quite an endeavour to roll your own. The Glanzmann installer does indeed bookworm (testing)
<QuickTimeX>
Thanks! I am interested in running openmediavault which requires bullseye. but not if it is too difficult. maybe I will play with nextcloud in docker first
<janrinze>
QuickTimeX: Before glanzmann worked on it I would roll my own. However that would require several disk partitions and a working install. Next mount an unused partition and debootstrap Bullseye. Not sure if the DEBs from Glanzmann will install on Bullseye though..
<janrinze>
QuickTimeX: Also setting up after debootstrap is not for the faint of heart..
<janrinze>
QuickTimeX: and booting into that image I would add a grub entry in the working Linux install.
<janrinze>
QuickTimeX: My experience was that this will yield a setup that won't be maintainable and won't be compatible with future upgrades or installs
QuickTimeX has quit [Remote host closed the connection]
deteg1337 has quit [Remote host closed the connection]
janrinze has quit [Remote host closed the connection]
<janrinze>
jannau: I've managed to get my homebrew kernel to run with latest mesa. (including your patch)
pthariensflame has joined #asahi-alt
pthariensflame has quit []
nopeslide13 has quit []
nopeslide13 has joined #asahi-alt
<zerdox>
chadmed: Thank you for the help. I tried again and it worked just fine. I found out that I was also missing required licenses. Can I try to provide PR so we can handle this case in install script?
<zerdox>
ChaosPrincess: thank you too.
jamespmorgan has quit [Remote host closed the connection]
nsklaus has quit [Quit: disappears in a cloud of smoke]
<chadmed>
zerdox: interesting that you were missing licenses, the script adds the required licenses to package.license
<chadmed>
by all means file a PR though
<chadmed>
glad it's finally working for you
<zerdox>
hm.. that was the only difference from my previous installation. maybe changes you made to the script made everything work
<chadmed>
could be, although if you look at the commit it was mostly just rewording stuff and making the fwextract script run automatically
<chadmed>
so who knows
<chadmed>
youd think a piece of sand that can only ever be in one of two states would behave a little more predictably
<zerdox>
yeah i looked up commit. you changed /boot/efi to /boot and some other stuff
nsklaus has joined #asahi-alt
<chadmed>
yeah oh well as long as it works!
<chadmed>
keep in mind of course that your EFI System Partition is now your /boot
<chadmed>
this makes it waaaaay easier to muck about with kernels and m1n1
<chadmed>
because you can just mount it directly in macos
<zerdox>
now i need to know how to automate updating my system with requirements from asahi tutorial. also i would dig a bit deeper about how do i customize kernel and all that stuff that i am unfamiliar with. i remember script asked me to install different kernel and play with it afterwards
<chadmed>
ah right no you dont install a "different" kernel
<chadmed>
the kernel sources and build artefacts are at /usr/src/linux
<zerdox>
yeah i remember it from install script
<chadmed>
the point of that message is to say that the kernel configuration is probably overly bloated for a gentoo install because its the default arch/asahi one
<chadmed>
so if youre comfortable customising kernel configs you can take stuff out like the useless HID drivers for devices you dont own etc
<chadmed>
for example i run a minimal kernel built with clang using LTO and only the bare minimum modules i need to use the system
<chadmed>
which is basically just the asahi drivers plus the cmedia codec driver for my dac
<zerdox>
i am not into all that kerneling stuff you know.. i used wrong wording for it. i will take a look at configurating current kernel
<zerdox>
also that boot penguins are a bit cringy imo. i guess they are related to kernel also, right?
<chadmed>
yeah thats just a kernel config option
<chadmed>
1 tux = 1 cpu
<zerdox>
owkay. good luck to you and good luck for me learning all this. ima also create dotfiles repo this time so i can hop easier cause i will rice this shit now
<chadmed>
lmfao awesome have fun
<chadmed>
keep /etc/portage synced up with that too
<chadmed>
you will end up with loads of stuff in there
<chadmed>
and its a pain in the arse to try and reconstruct
<zerdox>
ok. i will just right down things you say cause for now i don't understand them