#archlinux32 | Logs for 2024-07-09

Back
[00:15:00] <KitsuWhooa> yeah something is broken in the iso
[00:15:01] <KitsuWhooa> sigh
[00:28:43] -!- drathir_tor has joined #archlinux32
[06:58:34] <Cthuutloops> so working out some issues with a fresh install, the UIDs are linking to user names. so i go to look at /etc/passwd and it doesnt exist. yet i am able to log into root and user account....
[06:59:09] <Cthuutloops> sorry that should read "the UID's are NOT linking"
[07:00:35] <Cthuutloops> also i am getting "I have no name!" for my user name at the bash prompt
[07:02:28] <Cthuutloops> nvm i tried lookin at /etc/passwd as root and its there
[07:05:00] <Cthuutloops> its weird because i think my useraccount isnt linked so doesnt get permissions to anything but its in /etc/passwd with the UID that seems reasonable
[07:08:03] <Cthuutloops> its weird cuz when i ls -l it doesnt give usernames only uids like root is 0
[07:09:03] <Cthuutloops> the root user can see the usernames but my account only gets uid's
[11:10:19] -!- Cthuutloops has quit [Remote host closed the connection]
[13:32:18] <KitsuWhooa> I suspect we need a newer kernel
[13:39:43] <KitsuWhooa> ugh I need to undo the kernel pahole stuff again
[15:04:31] -!- Peetz0r has joined #archlinux32
[15:37:24] <Peetz0r> ohai. I'm trying to install a modern linux distro on a 20 year old laptop (Athlon XP cpu, 512 MB ram). So after learning that my CPU doesn't have SSE2 and "debian i686" requires that, I found the list at https://www.archlinux32.org and I picked the i686 version of archlinux32
[15:37:26] <phrik> Title: Arch Linux 32 - Architecture Overview (at www.archlinux32.org)
[15:38:04] <Peetz0r> here's the full output of pacstrap: https://bpa.st
[15:38:04] <Peetz0r> So that was the context, now the main question. why does it fail with signature issues?
[15:38:05] <phrik> Title: View paste NTWQ (at bpa.st)
[15:38:52] <Peetz0r> also more context: I'm running this in a VM on a modern system with the intention of moving the disk over after installing
[15:39:38] <Peetz0r> but it's a legacy mode vm with `-cpu pentium3` so it should be close enough
[16:10:50] <bill-auger> Peetz0r: it is probably this issue https://bugs.archlinux32.org - that is a current topic of discussion currently
[16:10:51] <phrik> Title: FS#360 : Archlinux32-Keyring was trusted but became untrusted (at bugs.archlinux32.org)
[16:13:51] <Peetz0r> Thanks, I'll go read that :)
[16:17:28] <bill-auger> based on that log, my guess is that you did not initialize the keyring yet - you should definitely try that
[16:20:33] <bill-auger> # pacman-key -init && pacman-key --populate archlinux32
[16:22:54] <bill-auger> then `pacman -Sy archlinux32-keyring` before running pacstrap
[16:24:39] <Peetz0r> oooh I need to do more steps because it's archlinux32 :)
[16:25:07] <Peetz0r> makes total sense actually
[16:59:55] <KitsuWhooa> you shouldn't have to
[17:00:04] <KitsuWhooa> things are broken unfortunately
[17:00:10] <KitsuWhooa> so that's a workaround for now
[17:05:50] <KitsuWhooa> also make sure you end up with i686 and not pentium4
[18:06:04] -!- drathir87 has joined #archlinux32
[18:06:34] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[18:06:39] drathir87 is now known as drathir_tor
[18:23:51] <Peetz0r> yeah I told qemu to pretend it's pentium3
[18:24:03] <Peetz0r> before I did that it indeed gave me pentium4 packages, but now it gave me i686 packages
[18:24:07] <Peetz0r> so that works as it should :)
[18:38:02] <KitsuWhooa> ah
[18:38:08] <KitsuWhooa> I know qemu executes SSE2 with pentium3 just fine
[18:38:13] <KitsuWhooa> that's why I pointed it out
[18:38:19] <KitsuWhooa> but glad to hear it did the right thing
[19:09:44] -!- drathir_tor has quit [Remote host closed the connection]
[19:17:48] -!- drathir_tor has joined #archlinux32
[19:41:02] <Peetz0r> so it booted on the real hardware but it couldn't find the rootfs (or any /dev/sd* at all)
[19:46:03] <Peetz0r> https://u.pixie.town
[19:56:46] <KitsuWhooa> what's the hardware on it?
[19:56:52] <KitsuWhooa> does dmesg say anything about it?
[20:02:34] <Peetz0r> the hardware is an HP ZE4268 laptop from 2003. The CPU is an AMD Athlon XP 2200+, 512 MB ram, onboard ATi graphics, the harddisk is an Hitachi Travelstar 40 GB 4200 rpm thingie
[20:02:56] <Peetz0r> dmesg didn't say anything that grabbed my attention but I didn't look too closely
[20:03:22] <KitsuWhooa> do you see any ide controllers with lspci?
[20:03:36] <KitsuWhooa> ...I suspect you can't run lspci
[20:03:37] <KitsuWhooa> hm
[20:07:41] <KitsuWhooa> it seems like the kernel either doesn't have a driver for the disk controller, or there's some kind of bug
[20:07:50] <KitsuWhooa> first step would be to figure out what the controller is and what driver it'd use
[20:08:27] <KitsuWhooa> that model number doesn't really bring information up sadly
[20:12:21] <KitsuWhooa> I wonder if you can put lspci in the initramfs :p
[20:13:00] <KitsuWhooa> or painstakingly go through /sys/bus/pci
[20:14:03] <Peetz0r> surprisingly enough I do have lssci but no lspci, lsusb, lsblk, etc
[20:14:23] <KitsuWhooa> do you have find?
[20:14:50] <Peetz0r> you could search for hp ze4200 maybe
[20:14:54] <Peetz0r> no find, but I do have ls|grep
[20:15:55] <KitsuWhooa> do you have readlink?
[20:16:10] <KitsuWhooa> actually, hm
[20:16:18] <KitsuWhooa> try ls -la /sys/bus/pci/devices/*/driver
[20:16:24] <Peetz0r> yes I have readlink
[20:16:31] <KitsuWhooa> readlink /sys/bus/pci/devices/*/driver
[20:16:31] <KitsuWhooa> :p
[20:19:15] <Peetz0r> I have only one pci device with a driver, and it's ohci-pci
[20:19:20] <Peetz0r> (yes it only has usb 1.1 ports)
[20:19:43] <KitsuWhooa> ...Right, because the shell won't expand if there's no driver file
[20:19:45] <KitsuWhooa> uh
[20:21:32] <KitsuWhooa> this might be a pain, but how about
[20:21:36] <KitsuWhooa> grep "" /sys/bus/pci/devices/*/{vendor,device}
[20:22:13] <KitsuWhooa> and then go through every vid/pid pair until you find what your storage controller is :p
[20:25:59] <Peetz0r> nope, shell doesn't like curly braces. But I have dealth with barebones shells before
[20:26:10] <Peetz0r> https://u.pixie.town
[20:26:10] <KitsuWhooa> ah, it's probably not bash
[20:26:17] <Peetz0r> brb figuring out what each of those is
[20:26:20] <KitsuWhooa> yeah, sorry, I keep forgetting :p
[20:26:34] <Peetz0r> yeah it's busybox :)
[20:26:49] <KitsuWhooa> busybox doesn't have lspci?
[20:26:57] <KitsuWhooa> what if you run `busybox lspci`
[20:28:11] <Peetz0r> nope
[20:28:15] <KitsuWhooa> huh
[20:28:28] <Peetz0r> https://u.pixie.town
[20:28:29] <Peetz0r> there's all of them
[20:29:13] <KitsuWhooa> yeah, no worries
[20:29:16] <KitsuWhooa> I was wrong
[20:31:08] <KitsuWhooa> looks like this might be a ULi chip
[20:31:54] <KitsuWhooa> yup
[20:31:56] <KitsuWhooa> 10b9:5229
[20:33:19] -!- bdju has quit [Read error: Connection reset by peer]
[20:34:17] <Peetz0r> I was just about to say that :)
[20:35:15] <KitsuWhooa> CONFIG_PATA_ALI=m hm
[20:36:05] <KitsuWhooa> does /proc/config.gz exist?
[20:36:13] <Peetz0r> do I need to do anything to get that module in my initramfs?
[20:36:27] <KitsuWhooa> I'd expect it to be there by default
[20:36:37] <KitsuWhooa> but maybe upstream removed it
[20:37:24] <Peetz0r> yes, the /proc/config.gz has CONFIG_PATA_ALI=m
[20:37:42] <Peetz0r> but does initramfs have the module?
[20:37:55] -!- bdju has joined #archlinux32
[20:38:00] <KitsuWhooa> I do not know that
[20:38:02] <KitsuWhooa> I also not know how to check
[20:38:14] <KitsuWhooa> I guess you can ls /usr/lib/modules
[20:38:20] <KitsuWhooa> should be pata_ali.ko
[20:38:24] <KitsuWhooa> .ko.whatever
[20:38:26] <KitsuWhooa> zst I think
[20:39:13] <KitsuWhooa> /usr/lib/modules/*/kernel/drivers/ata/pata_ali.ko/zst
[20:39:15] <KitsuWhooa> /usr/lib/modules/*/kernel/drivers/ata/pata_ali.ko.zst
[20:40:21] <Peetz0r> nope, but it does have /usr/lib/modules/*/kernel/drivers/block/virtio_blk.ko.zst
[20:40:31] <KitsuWhooa> okay that explains that then :p
[20:40:35] <Peetz0r> so yeah, running the installer in a vm and then moving the disk backfired
[20:40:46] <Peetz0r> but only slightly
[20:40:48] <KitsuWhooa> I don't think it would've made a difference
[20:41:26] <Peetz0r> I guess I can move the disk back, boot the vm again, and manually tell mkinitcpio to add the right module
[20:41:41] <KitsuWhooa> I think that's your only choice right now unfortunately
[20:42:59] <KitsuWhooa> > Adds block device modules. If the autodetect hook runs before this hook, it will only add modules for block devices used on the system. Exceptions are the ahci, sd_mod, usb_storage, uas, mmc_block, nvme, virtio_scsi and virtio_blk modules which will always be added unconditionally.
[20:43:27] <bill-auger> Peetz0r: have you got the OS to boot on metal now? - the reason for the first screenshot you sent is obvious
[20:43:29] <Peetz0r> yeah, everything you need in 2024
[20:43:32] <Peetz0r> except that I'm in 2003 :p
[20:43:38] <KitsuWhooa> that's what I'm thinking
[20:43:46] <KitsuWhooa> maybe we should add a downstream patch to include all pata controllers
[20:43:56] <KitsuWhooa> how does my laptop boot though, for example
[20:44:38] <KitsuWhooa> how does anything boot
[20:44:48] <Peetz0r> I guess if you installed it not in a vm but on the device itself (e.g. the "normal" usecase), it autodetects and adds the modules you need
[20:44:49] <KitsuWhooa> I guess it doesn't overwrite the config manually
[20:44:58] <KitsuWhooa> ah
[20:44:59] <KitsuWhooa> hm
[20:45:03] <KitsuWhooa> yeah you're right
[20:45:05] <Peetz0r> but my machine doesn't boot from usb, and having to burn a cd is super annoying in 2024
[20:45:22] <KitsuWhooa> would it even boot though off cd?
[20:45:40] <KitsuWhooa> it's on the same pata controller
[20:46:10] <KitsuWhooa> as for usb, I have a CD with plop bootloader around
[20:46:15] <KitsuWhooa> it has a usb stack that sometimes works well enough
[20:46:28] <Peetz0r> eh, I don't know exactly bot booting from cd is weird
[20:46:28] <KitsuWhooa> and a floppy too :p
[20:46:41] <Peetz0r> and yeah maybe I should make a cd with bootloader stuff on it for usecases like these
[20:46:59] <Peetz0r> but my only other retro machine doesn't even boot from CD because it doesn't have a cd drive at all
[20:47:20] <Peetz0r> (and it's floppy drive is broken, and it really doesn't run modern linux at all anyway since it has like 12 MB ram)
[20:49:01] <KitsuWhooa> https://gitlab.archlinux.org
[20:49:02] <phrik> Title: install/block · master · Arch Linux / Mkinitcpio / mkinitcpio · GitLab (at gitlab.archlinux.org)
[20:49:48] <KitsuWhooa> so yeah, it's under checked
[20:50:07] <KitsuWhooa> I do wonder if we should unconditionally add pata
[20:50:27] <KitsuWhooa> and maybe remove nvme and ahci by default
[20:50:52] <KitsuWhooa> I don't think there are ahci native machines that are 32 bit only
[20:51:19] <KitsuWhooa> maybe even virtio
[20:58:22] <Peetz0r> I'd definitely keep those since I think the usecase of swapping disks between old and new machines (or old machines and vm's on new machines) is not that exceptional
[20:58:49] <KitsuWhooa> booting arch32 on a modern machine doesn't sound like a good idea :p
[20:59:04] <KitsuWhooa> as for vms, they all use the emulated disk controllers by default
[20:59:13] <KitsuWhooa> unless you explicitly pick virtio
[21:00:22] <Peetz0r> https://u.pixie.town
[21:00:23] <Peetz0r> I think this is using virtio
[21:00:43] <KitsuWhooa> huh
[21:01:03] <Peetz0r> and I let it pick the defaults (but it autodetected the OS from the install media as a modern linux so it picked the most modern preset, I think)
[21:01:11] <KitsuWhooa> yeah, I think that's what happened
[21:02:06] <KitsuWhooa> ...virt manager can't create a VM while virtualbox is running
[21:02:06] <KitsuWhooa> sigh
[21:04:14] <KitsuWhooa> but anyway, yeah, fair enough about virtio
[21:11:11] <Peetz0r> aaaaand it boots fine now :)
[21:11:54] <KitsuWhooa> cool
[21:32:55] <Peetz0r> so now it boots but for some reason dbus-broker doesn't want to start
[21:33:01] <Peetz0r> (it does the same thing in the vm)
[21:33:40] <KitsuWhooa> maybe give old dbus-daemon
[21:33:41] <KitsuWhooa> a try
[21:34:05] <Peetz0r> https://u.pixie.town
[21:35:00] <KitsuWhooa> you can try to debug it if you want :p
[21:36:17] <KitsuWhooa> maybe related? https://bugzilla.redhat.com
[21:36:18] <phrik> Title: 1704920 – dbus-broker fails to start: launcher_load_service_dir @ ../src/launch/launcher.c +753: Permission denied (at bugzilla.redhat.com)
[21:36:36] <KitsuWhooa> actually no, it's not related
[21:37:59] <KitsuWhooa> ah
[21:38:11] <KitsuWhooa> it seems like it can't read the systemd machine id?
[21:39:51] -!- bill-auger has quit [Remote host closed the connection]
[21:40:39] <KitsuWhooa> Peetz0r: can you access /etc/machine-id ?
[21:43:27] -!- bill-auger has joined #archlinux32
[21:43:39] <KitsuWhooa> perhaps systemd is preventing access to it in the unit file
[21:49:49] -!- KillerWasp has quit [Server closed connection]
[21:50:00] -!- KillerWasp has joined #archlinux32
[21:50:23] <Peetz0r> sorry, was doing other things for the past 15 minutes
[21:50:27] <Peetz0r> but I'm back
[21:50:35] <KitsuWhooa> no worries
[21:50:36] <Peetz0r> /etc/machine-id does exist
[21:50:47] <Peetz0r> and it has an okay looking id inside
[21:51:26] <KitsuWhooa> well, for some reason dbus-broker is trying to open it (via libsystemd) and it's getting permission denied
[21:51:49] <KitsuWhooa> is there anything by auditd in dmesg?
[21:51:52] <KitsuWhooa> that might be related
[21:54:22] <Peetz0r> https://u.pixie.town
[21:54:48] <KitsuWhooa> I have no idea
[21:55:02] <KitsuWhooa> (as a sidenote, you can boot with console=ttyS0,115200n8 and then use the serial console to be able to copy text
[21:55:05] <KitsuWhooa> )
[21:55:27] <KitsuWhooa> view -> consoles -> serial 1
[21:55:33] <Peetz0r> that's way too smart :p
[21:56:11] <KitsuWhooa> :p
[21:56:18] <Peetz0r> (what I really wanted to do is just enable sshd and call it a day, but that depends on network being up, which I wanted to do with networkmanager, which relies on dbus, and guess where we are now)
[21:56:24] <KitsuWhooa> yup
[21:56:57] <KitsuWhooa> as I said, if you just want it to work, use dbus-daemon
[21:57:00] <KitsuWhooa> instead of dbus-broker
[21:57:36] <KitsuWhooa> otherwise you're in for debugging :p
[21:57:43] <KitsuWhooa> but I suspect it has something to do with systemd sandboxing
[21:57:48] <Peetz0r> well I was spemnding time anyway so figuring out why and fixing it at the cause would be fun
[21:57:57] <KitsuWhooa> okay, then that is very much appreciated
[21:58:01] <Peetz0r> but first, rebooting with serial
[21:58:08] <Peetz0r> since yeah, copypasting is nice
[22:01:29] <Peetz0r> This is the dbus-broker.service: https://bpa.st
[22:01:30] <phrik> Title: View paste D37A (at bpa.st)
[22:01:45] <Peetz0r> maybe `ProtectSystem=full` is suboptimal?
[22:02:03] <KitsuWhooa> try without it I guess
[22:02:38] <KitsuWhooa> it's the same as upstream though
[22:03:12] <KitsuWhooa> upstream -> Arch
[22:05:10] <KitsuWhooa> stupid question; machine-id r--r--r--, right?
[22:05:23] <KitsuWhooa> *machine-id is r--r--r--, right?
[22:19:06] <Peetz0r> yep
[22:21:36] <Peetz0r> I disabled all the Protect* things in the service, no difference
[22:31:22] <KitsuWhooa> what if you try to run it manually as root?
[22:31:47] <KitsuWhooa> like sudo /usr/bin/dbus-broker-launch --scope system --audit
[22:37:19] <Peetz0r> [root@hp-ze4268-arch ~]# /usr/bin/dbus-broker-launch --scope system --audit
[22:37:19] <Peetz0r> No listener socket inherited
[22:37:31] <Peetz0r> fails in an entirely different mode from what the logs show
[22:38:28] -!- drathir_tor has quit [Remote host closed the connection]
[22:39:08] <Peetz0r> this is what `journalctl` gives me: https://bpa.st
[22:39:09] <phrik> Title: View paste HTIQ (at bpa.st)
[22:39:40] <KitsuWhooa> yeah
[22:40:43] <KitsuWhooa> ../dbus-broker-36/src/launch/launcher.c +326 points to https://github.com
[22:41:30] <KitsuWhooa> 01:37:19 <Peetz0r> No listener socket inherited <-- ah, so you probably can't run it standalone
[22:44:06] <Peetz0r> yeah I suspected something like that
[22:49:09] -!- drathir_tor has joined #archlinux32
[22:59:33] <Peetz0r> ok I'm trough all the things that I can think of
[22:59:44] <Peetz0r> so I'll probably swap to the original dbus anyway
[23:00:55] <KitsuWhooa> interestingly, dbus-broker works in the iso
[23:02:09] <Peetz0r> yeah, it does indeed work fine over there
[23:02:10] <Peetz0r> but, why?
[23:04:45] <KitsuWhooa> I have absolutely no idea
[23:30:26] -!- drathir_tor has quit [Ping timeout: 260 seconds]
[23:32:46] <KitsuWhooa> I wrote a test program that gets the machine id, ran it through the dbus-broker systemd service, and it succeeds
[23:32:51] <KitsuWhooa> so something is fucky with dbus-broker itself
[23:33:46] -!- drathir_tor has joined #archlinux32
[23:34:00] <KitsuWhooa> maybe a rebuild will fix it
[23:34:16] <KitsuWhooa> oh, hm, there's a new version in staging...
[23:36:09] <KitsuWhooa> Peetz0r: manually installing dbus-broker from staging fixes it
[23:36:10] <KitsuWhooa> 🤦‍♀
[23:36:18] <KitsuWhooa> I'll push it to stable...
[23:36:33] <KitsuWhooa> actually no
[23:36:34] <KitsuWhooa> it still failed
[23:36:38] <KitsuWhooa> nevermind, I'm dumb
[23:37:45] <Peetz0r> oh I just swapped to dbus-daemon and rebooted to check if it works
[23:38:10] <KitsuWhooa> it's fine, it doesn't work
[23:38:14] <KitsuWhooa> false alarm :p
[23:48:17] <KitsuWhooa> okay something is really fucky
[23:48:23] <KitsuWhooa> I made a user, I logged in as the user
[23:48:38] <KitsuWhooa> and on login I keep getting bash: /etc/profile: Permission denied
[23:49:06] <KitsuWhooa> /etc/profile is rw/r/r
[23:49:12] <Peetz0r> huh that's definitely Quite Borken (tm)
[23:50:38] <Peetz0r> meanwhile, I was trying to install lxqt and it couldn't find the right version of libfm-qt
[23:50:44] <Peetz0r> https://bpa.st
[23:50:45] <phrik> Title: View paste FPGQ (at bpa.st)
[23:51:02] <KitsuWhooa> yeah, things have been fucked for a while
[23:51:09] <KitsuWhooa> but something is really broken with normal suers
[23:51:11] <KitsuWhooa> *normal users
[23:51:16] <Peetz0r> for context, I'm somewhat new to pacman's dependency handling and I'm unsure if it's wise to just press Y there
[23:51:34] <Peetz0r> so far I haven't created a nonroot user yet :x
[23:51:43] <Peetz0r> I guess I may do that first
[23:52:01] <KitsuWhooa> it should be fine I think?
[23:52:39] <KitsuWhooa> but yeah, uh, non root users do not have access to /etc at all
[23:52:47] <KitsuWhooa> what could even cause this
[23:52:58] <KitsuWhooa> pam?
[23:54:00] <KitsuWhooa> Oh
[23:54:12] <KitsuWhooa> /etc has wrong permissions
[23:54:16] <KitsuWhooa> ...that'd do it
[23:54:56] <KitsuWhooa> chmod 755 /etc
[23:55:15] <KitsuWhooa> that fixed dbus-broker too
[23:55:27] <KitsuWhooa> of course it gets permission denied if it tries to access /etc/machine-id...
[23:55:28] <Peetz0r> oh huh that's weird
[23:56:11] <KitsuWhooa> !later
[23:56:12] <phrik> Nao is later.
[23:56:15] <KitsuWhooa> !list later
[23:56:15] <phrik> KitsuWhooa: notes, remove, tell, and undo
[23:56:28] <KitsuWhooa> !later tell
[23:56:29] <phrik> KitsuWhooa: later tell <nick1[,nick2[,...]]> <text>
[23:57:46] <KitsuWhooa> !later tell Cthuutloops You're going to need to `chmod 755 /etc` to fix the numeric ids
[23:57:46] <phrik> KitsuWhooa: Bazinga!
[23:57:52] <KitsuWhooa> ...bazinga? lmao
[23:58:25] <KitsuWhooa> so what package makes the filesystem
[23:58:55] <KitsuWhooa> `filesystem`
[23:59:00] <KitsuWhooa> deep thought should be jealous