You are not logged in.
There are i486 ISOs which are completely deviating from the Archlinux way of ISOs, especially for machines with low memory.
The ISO is a little bit rough around the edge, so there is a project https://git.archlinux32.org/archi486 with a script building
that ISO (this has nothing to do with mkarchiso from Archlinux or the normal ISO of Archlinux32).
The https://bbs.archlinux32.org/viewforum.php?id=11 part of the forum has some examples of how to use pacstrap for instance
in compact flashes and then move them to machines with little memory (which is far the easiest way IMHO).
https://git.archlinux32.org/archi486/tree/motd documents some things you can do when running low on memory (the most
prominent is to add swap as soon as possible).
I also did PXE boots with iPXE, but they tend to be a little bit complicated to set up.. (and there are network cards not even
supporting PXE in such old machines).
I defenitely encourage people to try such old machines, because it can be fun to see the newest kernel and systemd on
a say pentium II, and the machine is actually usable.. :-)
Offline
Interesting. Thanks Andreas. I will check it out. I do like using these old machines and in particular as "consoles" and support services for even older 1970s vintage mini and microcomputers such as PDP-11s, Z80s running CP/M, etc. I like to be able to write modern Python to communicate with the console of the older machines to download code, send images of floppies, etc and they are still very useful for this-- especially if they have an ethernet or Wifi that functions and can connect them to the modern world on the other side.
Chris
Offline
Hello again,
Excuse me that it was I who opened the discussion and I have absented myself completely. I'm sorry about that but there have been important reasons.
What chrise says above is my experience. Previously I had arch (ISO 01-Feb-2022) installed, on my IBM ThinkPad T41, Pentium M, 2G ram, and everything was going perfectly. At the end of February I updated the system (sudo pacman -Syu) and the reboot froze at "Loading initial ramdisk ..."; all attempts to repair it have been unsuccessful.
I have reinstalled using all available isos, and it always freezes at the same point.
The latest ISO from March also freezes at the same point.
Regards,
Alberto
Offline
I have also used a more modern machine to install to one of the Dell hard drives and then move that hard drive to a Dell laptop after installation. It still hangs at "Loading initial ramdisk". However, with loglevel=7 and nomodeset, I was able to see that it is getting as far as udev triggers (I forget now the exact message). So, it appears that we are actually loading and starting the kernel and it is when systemd and/or udev starts loading stuff that it falls over. I guess that isn't a surprise but does suggest again that there is something in the initrd that is unfavorable. Note that I still hang even with the 'fallback" initrd. Seems like we need a really simple, not very capable initrd that just gets the root mounted and not much other special hardware support.
For now, I need one of these machines to talk to an old EPROM/PAL programmer so I am installing Windows98. Imagine that. :-(
Chris
Offline
I tested now on all my real machines, not one is showing this effect.. some have LTS kernels (for instance because drivers like 'gdth' got thrown out of the kernel,
but this is for an almost 20 year old server, I doubt laptop hardware would be affected).
Offline
Hi again Abaumann,
Thank you very much for taking the time to help us solve our problem.
I am glad that you have verified that everything is ok with the published iso.
There is something else I would like to try and I will comment here the result.
I want to try with the original hard drive instead of the ssd card.
Alberto
Offline
Presumably you're still booting the original iso on the SSD. That iso should have all you need along with the install instructions to take that iso and make an installation on the hard disc.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
I have now one machine (Acer Travelmate 524TEV) which hangs on booting the cdrom when checking for hard disks.
The hard disk itself is fine (I can access it via Slackware).
I see tons of errors around systemd-auto whatever things coraking on EFI, GTP and stuff.. now my floppy drive (yes,
there is such a mystical thing) has it's LED turned on and the machine hangs.
The kernel itself shows ata drives (in this case via pata_ali or so).
Though I added now 'debug' to the kernel options, I don't really understand what's going on.
Offline
Yep, I can boot with kernel 5.15.7 not with any 5.16.x..
This is a trend I see more and more: the kernel looses support for old things fast now (gdth, VGA on Alix, PATA boots now obviously).
Offline
I'm only using to keep my eeepc functional, and for me 5.16.x boots and functions just fine.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
[abaumann@eurobuild6 linux-5.16.8]$ diff .config ../../config
5c5
< CONFIG_CC_VERSION_TEXT="gcc (GCC) 11.2.0"
---
> CONFIG_CC_VERSION_TEXT="gcc (GCC) 11.1.0"
7c7
< CONFIG_GCC_VERSION=110200
---
> CONFIG_GCC_VERSION=110100
10c10
< CONFIG_AS_VERSION=23800
---
> CONFIG_AS_VERSION=23601
12c12
< CONFIG_LD_VERSION=23800
---
> CONFIG_LD_VERSION=23601
This could explain something..
Offline
Hello again guys,
As I said in my last post, I tried to install arch in the original hard drive, because the ssd card that I used before is with a SATA to IDE adapter, but with the original hard drive is the same.
I see that you are still investigating what is the reason for this problem.
I am sorry that I cannot contribute to the work by helping a little, but my knowledge is very limited.
If there's anything I can do to help by testing or something, let me know and I'll be happy to do it.
Regards, Alberto
Offline
It's not 100 percent clear to me looking at this thread afresh what your exact problem is. You say that the Feb-2022 iso booted fine on your SSD hooked over an adaptor into a PATA/IDE port, but on updating it you got some block early on in the boot process. And abaumann says that 5.16.x kernels are broken on some hardware, so it's maybe worth reimaging that SSD with the Feb iso, booting from it, switching to the LTS kernel then trying a pacman update again. I think if it's broken with the mainline kernel even on the SSD we need to fix it first on the SSD before considering moving to the IDE hard drive you have installed in that machine.
Last edited by levi (2022-03-16 17:51:43)
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
Very well, I can do that if that way I help you.
Please, correct me if I have not understood correctly.
1. boot from usb
2. arch-chroot /mnt
3. Pacman-Syu
Offline
I have done.
But it does not work; It stays frozen in the same place.
Offline
You seem to have at the very least omitted to installed the LTS kernel there. You need to do that before you reboot ideally before you do your first update.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
Hello Levi, you are right, I did not install the LTS Kernel, I am sorry, I have already said that I am inexperienced.
I'll do it tomorrow and tell you how it goes.
Thank you. :-)
Offline
I'm just glad we seem to be communicating better now. That can be difficult in a text-based medium, but we seem to be there now.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
I just have to add here that I do have at least three machine which are on an LTS kernel, I almost forgot that they don't work
anymore with the normal kernel:
- gtdth on a Intel server is missing (RAID controller)
- VGA BIOS completely unreadable and borked on Alix 1.E (as I need the Xorg VESA output, Geode LX doesn't work ATM)
- HP Gen1 server: this one actually also shows this "stuck" behaviour, but I have to test again (the problem is, I can only
boot it via two USB sticks in the case I'm borking up the installation, one using plopp and a second one containing the Arch32 ISO).
I might want to dig into PXE booting there to speed up the testing cycles, but I fear the hardware is pre-PXE.
In general, the problem is usually: the ones having the old hardware are not necesserily good kernel developers (pointing at me),
and the ones who are competent kernel coders don't have the old hardware or better things to do. :-)
Offline
Hi again,
I have tried to install the LTS Kernel, but it gives me this error message.
pacman -S linux-lts"error: linux-lts: signature from "Erich Eckner ... " is unknown trust
Offline
New update: I was able to install the lts kernel, after importing the public key.
# curl -Ss https://archlinux32.org/keys.php?k=5FDCA472AB93292BC678FD59255A76DB9A12601A | gpg --homedir /etc/pacman.d/gnupg --import
# pacman -S linux-lts
# grub-mkconfig -o /boot/grub/grub.cfg
# reboot
# pacman -R linux
# grub-mkconfig -o /boot/grub/grub.cfgI was finally able to get into my arch installation.
But I must say that it is showing strange behaviour.
It keeps popping up on the screen...
"systemd-journald[179]: Sent WATCHDOG=1 notification.
Also if I try to update the system ...
error: failed retrieving file ... mirror archlinux32.org Could not resolve host
Offline
The first problem was because I had the "debug" option set to GRUB_CMDLINE_LINUX_DEFAULT. But I have already solved it.
GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 quiet"I'm working on solving the problem with the mirror.
If you see this before I figure it out, a little help is welcome.
Offline