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
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
cylm_ has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
jamespmo_ has joined #asahi-alt
jamespmorgan has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
dgb has joined #asahi-alt
pthariensflame has joined #asahi-alt
pthariensflame has quit []
dgb has quit [Quit: dgb]
kallisti has joined #asahi-alt
dgb has joined #asahi-alt
<Glanzmann>
ahdeb[m]: Thank you.
<dgb>
Glanzmann: hey i just want to say thanks for all your hard work on debian! do you have any plans on a debian archive? i got tired of building new kernels and briefly thought about using a ci, but it's not really worth it if no one else uses it. but if you have any plans or if there's an audience i'd be happy to help.
<cy8aer>
And Glanzmann is very fast in updating it.
<dgb>
oh thanks. i don't recall seeing this when i first started and i haven't checked his git in a while.
<Glanzmann>
dgb: It is pretty new. I started it on the day the gpu code was available. Because we now need to many packages which are not upstream and won't be for quiet a while.
<Glanzmann>
too*
<Glanzmann>
There is no CI, but I have a cronjob which notifies me of any changes at 18:00 each day. I'll test them and if I don't see any regressions I push them out. But I also keep all old versions in the repository. So if we have any regressions, we can easily go back. Also m1n1 makes backup copies.
<Glanzmann>
The kernel is the debian standard config + asahi options.
<dgb>
oh yeah i have to rebuild libmutter because i'm too stubborn to use a mainstream de
<dgb>
is there a debian installer in the works? i have some time after the new year and i don't mind giving it a crack
<Glanzmann>
dgb: I used to have a modified version that works for asahi.
<Glanzmann>
But at some point I decided that the debian installer is so broken and constantly being broken and noone really cared, so I decided to drop it.
<Glanzmann>
So now we have two options to install Debian a) you use the image b) you use the live system and do everything manual
<Glanzmann>
dgb: But if you want to make a functional debian installer, you can checkout 'bootstrap.sh' from 2 weeks ago and modify the commented out d-i section.
<dgb>
i used your flash drive image because i wanted to use btrfs and it was quite some typing. i figure it'll make it easier to have an actual installer.
<dgb>
that reminds me, i was having trouble with wifi until i recalled i've been using dfs channels for so long i forgot about it. wireless-regdb was missing in your image at the time.
<dgb>
i didn't expect missing wireless-regdb would result in a wrong password prompt though.
<Glanzmann>
dgb: Never heard about wireless-regdb, but I added it.
<Glanzmann>
dgb: In the livesystem you can also install xorg and cut paste the commands for setup Linux. Most commands are in /root/quickstart.txt
<Glanzmann>
Personally I stopped using the d-i after it let me down a few times with simple raid 1 + ext4 setups and now only use livesystems (now my own, before grml) to setup Debian.
<cy8aer>
The main problem of the di was that it did text resolution changes (even in text mode). But I guess/hope that these problems would be solved with a dcp kernel (and - of course: grub, self install the kernel, firmware)
<Glanzmann>
I wrote a few scripts so I'm also faster than with the d-i.
<Glanzmann>
cy8aer: For me the man problem with d-i that they broke different things every other day and I just got sick of it.
<cy8aer>
Glanzmann: yeah one of the d-i developers should fall in love with asahi . Then the problems would be vanish...
<Glanzmann>
Yes. But to be honest the changed to the d-i was minimal. It was just: another kernel, extracting the firmware and loading the wifi module. And install grub with the removable flag.
<dgb>
Glanzmann: thanks. for some reason it never occur to me to install a desktop... i'll remember next time for sure.
<Glanzmann>
you can also install ssh and ssh in from another box.
Dcow has joined #asahi-alt
<dgb>
i don't have a second box but i guess i could've edited quickstart.txt into .sh, not saying it's a good idea ofc...
Dcow has quit [Ping timeout: 480 seconds]
<Glanzmann>
dgb: This is what I always end up doing, if I don't have x or a second computer.
<Glanzmann>
When I have a second computer, I boot into live, install ssh, set a password for a user or install my ssh key and do the rest remotely.
<dgb>
Glanzmann: that reminds me, is there a reason why you are installing system packages with apt individually instead of tasksel?
<dgb>
there's also a `laptop` which is `task-laptop` in apt
<Glanzmann>
dgb: But I don't see how that would help me.
swaggie has joined #asahi-alt
<dgb>
Glanzmann: afaik `standard` installs `~prequired ~pimportant ~pstandard` for the complete "standard system utilities," so future upstream changes wouldn't break your list
<Glanzmann>
dgb: I see. I try to keep the list as small as possible to make the installation quicker and the live system boot faster.
<Glanzmann>
dgb: I also have a version of the live system for amd64 which I internally use. There I have issues with the initrd to become to big to be booted by any bootloader I found on amd64 (grub, syslinux, ipxe). So I made the initrd as small as possible and have a rc.local which pulls in the rest.
<dgb>
makes sense. i guess i'm just not too worried when it comes to m1.
Dcow has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
pjakobsson_ has joined #asahi-alt
pjakobsson has quit [Ping timeout: 480 seconds]
<Glanzmann>
dgb: No on m1 the initrd size is so big that not yet found a limit. And I already tried several gb.
<arnd_>
is anyone seeing Xorg crash immediately in OsLookupColor+0x188 on Debian Bookworm?
<arnd_>
I just restarted X after running for two weeks and it did not come back but crashed with
<Glanzmann>
You can also build it your self using the libunwind.sh in my repository.
<mps>
Glanzmann: can you build xorg without libunwind
<Glanzmann>
Issue is that libunwind was recently updaed. The version had a 4k issue. See asahi mailinglist.
<arnd_>
ok, thanks
<Glanzmann>
mps: I know, but I try to stick as close to debian as possible.
<Glanzmann>
mps: However if the anti tearing modesetting patch makes it upstream. I'm the first guy compiling a new xorg for debian. :-)
<mps>
Glanzmann: I already built xorg with this patch, and I don't see differences
<mps>
Glanzmann: anyway, wonder why you as minimalist person use debian ;-)
* mps
was long time debian user
<Glanzmann>
mps: I use Debian, because it has the most packages for any distribution. It is stable, so it does not break all the time. And I can upgrade to another major release. I have systems which I have not reinstalled for 20+ years. I also have machines which used to be physical, than VMs, which used to be 32 bit and now 64. All without reinstallation ...
<mps>
Glanzmann: ok, ok, understand. just couldn't resist, sorry
<Glanzmann>
Oh and Debian packages are well tested. And the upgrade is well tested.
<mps>
also I have more than 20 years debian experience
<Glanzmann>
I started with Debian 2001. At that time I was a sysadmin and student at FAU. And we reinstalled the CIP Pool (200+) machines with Debian. We used Caldera before.
<arnd_>
I did something wrong, now I get the same problem with both 4kb and 16kb kernels. time to get out a laptop
<mps>
I forgot at which year I switched to debian from slackware but iirc 1996 or 1997
<Glanzmann>
mps: My first Linux distribution was Slackware 93 or 94. Don't remeber. Than SLES. Than RHEL. Than I also tried OpenBSD, NetBSD, FreeBSD, Solaris. And I build twice a linux distribution by myself. Once with linuxfrom scratch was 2000 or 2001 and once really from scratch. :-)
<mps>
arnd_: did you upgrade second stage m1n1+dtbs+u-boot
<Glanzmann>
arnd_: I also upgraded the live system for you. So you can also use that to fix things, if you like to.
<mps>
Glanzmann: looks like we switched to linux at nearly same time, I started in '92 or '93, forgot exact time
<Glanzmann>
mps: I only remeber that I still had Linux 0.9x on my first install.
<mps>
yes, iirc I had same
<Glanzmann>
So I assume it was 93.
<mps>
yes, time flies, doesn't it :-)
<Glanzmann>
Yes, it does.
<arnd_>
mps: no, I did not touch touch the boot loader at all. I did install the packages from https://thomas.glanzmann.de/asahi/README.txt, but am still running my own kernel with dcp but no appledrm
<Glanzmann>
arnd_: Did you only install libunwind8 or did you put all the packages?
<j`ey>
(not to nit pick, but DCP is DRM_APPLE, DRM_ASAHI is the GPU)
<Glanzmann>
arnd_: Is xorg still crashing or does it only not start?
<mps>
arnd_: idk for debian but a lot of people have this issue forgetting to upgrade second stage m1n1+dtbs+u-boot. I think Glanzmann could give proper answer
<Glanzmann>
j`ey: Maybe that is the problem.
<Glanzmann>
mps: I build a package for that it is called 'm1n1'.
<Glanzmann>
mps: It takes a backup copy of the old one and copies the new over.
<mps>
yesterday I tried to start xorg with simpledrm and it didn't worked with latest dtbs
<Glanzmann>
mps: Maybe that is arnd's problem.
<mps>
Glanzmann: my guess it is related, but I can't be sure ofc
<Glanzmann>
mps: Once I want to the GPU kernel I did not look back. Currently I have no issues other than xfreerdp. But it was broken with simpledrm as well but differently.
<mps>
j`ey: I'm thinkink to push APPLE_DRM to alpine kernel
<mps>
as module maybe
<Glanzmann>
I also had a lot of trouble with the kernel config on Debian. At first only mps config worked. System did not even boot. But than once to many users complained about missing kernel. I gave adopting the debian config another shot and it worked. And I automated the process. So hopefully in the future it will be easier.
<arnd_>
Glanzmann: it's the same crash as before. I installed libunwind8 and libunwind-dev manually and also did an 'apt upgrade' for the other packages on your server
<Glanzmann>
arnd_: Let me try to reproduce.
<Glanzmann>
arnd_: Have you tried my kernel already?
<arnd_>
Glanzmann: no, let me try that next. What version of m1n1 do I need for that? I'm currently on linux-next from just before the merge window, with a m1n1 that matches the asahi-rc8 kernel
<Glanzmann>
arnd: So I updated all packages (both from Debian and myself) and I can still boot into Xorg.
<Glanzmann>
arnd_: My latest kernel is 6.1.0 (no rc). But it should be pretty close to rc8.
<Glanzmann>
arnd_: You can install the package 'm1n1' from my repository.
<Glanzmann>
It does a backup copy of boot.bin and installs the new boot.bin
<Glanzmann>
But it does __not__ check if /boot/efi is mounted. I'll add that check soon.
<Glanzmann>
So make sure it is mount or copy it manually over once the package is installed: cp /usr/lib/m1n1/boot.bin /boot/efi/m1n1/
<arnd_>
I think I did all that now, but see no changes
<Glanzmann>
Damn it.
<Glanzmann>
So the system boots, but Xorg does not start?
<arnd_>
maybe I'm stil booting an old m1n1 from a different partition. I had installed two separate instances in order to have a fallback
cylm_ has joined #asahi-alt
<arnd_>
right, I can log in remotely, everything else seems fine. Before updating libunwind, I could still use Xorg with the 4K kernel, now it's broken with all kernels
<Glanzmann>
arnd_: strange.
<Glanzmann>
arnd_: Okay and if you start Xorg manually as root from the first console, it crashes?
<arnd_>
same thing
<Glanzmann>
If you're unsuere about the m1n1.
<Glanzmann>
Just mount every efi partition and put the updated boot.bin there.
<arnd_>
is there any way to find out which one was actually booting from?
<j`ey>
isn't debian uboot just picking the first?
<Glanzmann>
arnd_: What is your hardware platform?
<Glanzmann>
arnd_: If it is debian, it is always the first.
<arnd_>
this is a t6002
<arnd_>
I know which grub and kernel I'm booting, those are the ones I want, the question is which second stage m1n1 is picked up by the first stage
<Glanzmann>
arnd_: The one that is corresponding to the stub.
<Glanzmann>
m1n1 stage1 has the efi partition number hardcoded in.
<arnd_>
ok
<Glanzmann>
But that does not explain why X is not working for you.
<Glanzmann>
arnd_: Have you tried the live system?
<Glanzmann>
you could install a minimal xorg and see if it comes up there.
<Glanzmann>
I only have T8103 and there it works on the mini and air.
<Glanzmann>
arnd_: Yep. And your m1n1 vesion is not mine as a result.
<mps>
hm, /proc/device-tree/chosen/asahi,system-fw-version => 13.0 is this ventura macos
<Glanzmann>
mps: Yes, I keep my macos upgraded.
<mps>
I still have 12.5
<Glanzmann>
mps: I see. But I also have all my devices updated.
* mps
can't remember when booted to macos last time
<Glanzmann>
Hopefully it is just the old m1n1 version.
<arnd_>
no, I updated the other parttiion to your stage2 uboot v1.2.3, no changes
<Glanzmann>
arnd_: You could try the livesystem to rule out your userland.
<Glanzmann>
Let me test if xorg is working with the live system.
<arnd_>
I'll do that later then, will work with remote login from my laptop for now, need to get a few things done that I can hopefully do with an inferior keyboard and screen ;-)
<Glanzmann>
arnd_: I let you know if the xorg works for me in the irc.
<arnd_>
ok
<arnd_>
thanks a lot for the help so far. It's probably something I did in the end anyway
<arnd_>
"[ 22.533] (EE) 0: /usr/lib/xorg/Xorg (OsLookupColor+0x188) [0xaaab0e6b96c8]\n [ 22.533] (EE) unw_get_proc_info failed: no unwind info found [-10]" sounds like the unwinding still has problems. Maybe I need to install some -dbg package to get a proper backtrace?
<Glanzmann>
arnd_: I had this issue when I had the wrong graphic card hardcoded in the 30-modsetting.conf
<Glanzmann>
arnd_: I tried the live system. It works. I followed the instructions here: https://git.zerfleddert.de/cgi-bin/gitweb.cgi/m1-debian/ logged in as root without password. configured wifi, apt update; apt install lightdm xterm blackbox; /etc/init.d/lightdm start and logged in as root without password.
<Glanzmann>
arnd_: Or you kernel lacks the necessary colormapping.
<Glanzmann>
arnd_: But I would start with the live cd and see if that works. Because than you know it is a general problem or one with your userland.
<Glanzmann>
And than take it from there. I would also get rid of the second installation and use backup copies of m1n1 to revert back quickly using macos in case of an failure and the debian live usb for anything else.
swaggie has joined #asahi-alt
<arnd_>
Glanzmann: after I went into the recovery boot menu and picked the other stage1 loader, it booted into the right stage2 partition and it all seems to work now. No idea what changed since both stage2 are the same, and I was booting the same kernel
<arnd_>
getting rid of the second installation is probably a good idea, this seems to be causing most of my problems, even though I still don't understand what exactly is going on
<arnd_>
I'll need to make sure it all still works with my own kernel though, but now I really need to get to my actual work
<Glanzmann>
arnd_: I see. Perfect.
swaggie has quit [Ping timeout: 480 seconds]
<Glanzmann>
arnd_: My assumption is that your other stub is the wrong macos version. As a result the gpu driver refuses to probe. As a result, xorg can't start because it doesn't get the color mapping right. But who knows.
Glanzmann has quit [Quit: EOF]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
<cy8aer>
@Glanzmann: saw your retrospection some hours ago... yes 93, I had an 486dx2 80 amd. And it did not run longer than two days (slackware). And nobody cared about this because these machines were not stable with DOS neither... But it was the bus wih 40MHz...
<cy8aer>
(so the next machine got SCSI...)
<cy8aer>
because it was the ide card crashing...
swaggie has joined #asahi-alt
swaggie has quit [Ping timeout: 480 seconds]
jamespmorgan has joined #asahi-alt
jamespmo_ has quit [Ping timeout: 480 seconds]
swaggie has joined #asahi-alt
Doliometer has quit [Remote host closed the connection]