IRC channel logs

<etno>Hi hurd people ! In my attempt to get the Debian build (06/23) run on real hardware (Dell Inspiron 1750), I could get AHCI to read my disk (vremap did not like unaligned PCI BARs). The thing is now going through the drivers initialisation, and hangs with "start acpi: acpi".

<etno>I tried to keep only the ext2fs+exec servers; I get "start ext2fs:" and it hangs. Have you already experienced this "any server hangs at startup" symptom?

<damo22>etno: i think acpi server has problems on real hw, i never figured out all the irq issues

<damo22>but the acpi server is required to discover the irq number of the disk controller

<damo22>at least when using IOAPIC

<damo22>your best bet is to capture the log

<damo22>and share it here with a pastebin

<etno>damo22: APIC is disabled in the debian build AFAICT (at least, it is in mine 😅). I am not yet familiar with the disk controller concept. As I understand it, it implements the block layer in userland. With the linux ahci driver, I thought that ext2fs could directly make use of sd0p2. So I modified the grub config to start ext2fs and exec only.

<damo22>ah kindof, the block layer is implemented in gnumach (linux ahci driver) in the debian build but can be overridden to use userland rumpdisk driver instead

<etno>(I will try to get the logs, I may have to copy it by hand)

<damo22>i dont know if your hardware supports irq10 for ahci

<etno>I am really eager to use rumpdisk (that is for a good part making my interest in Hurd got now). But to get a base system running on hardware, I am trying to go as bare as possible.

<etno>The startup log says that it reads irq 10 on the BAR5. Then it can read the partition table.

<damo22>ok

<damo22>you need pci-arbiter

<damo22>as the first boot server

<etno>However, I had to disable port1 (likely the ATAPI cd-rom ) because it hung on IE. Which points to something definitely wrong with interrupts

<etno>> you need pci-arbiter as first < so, it would be pci-arbiter, ext2fs, exec

<damo22>yes i think so

<etno>Unfortunately, after keeping {pci-arbiter, ext2fs, exec} (and redoing the chaining + task-resume thingy), I get to the same kind of result: modules are printed, followed by "task loaded" for every of them. Ultimately, "start pci-arbiter:" and hang

<damo22>hmm

<damo22>i can share my grub entry

<damo22>with rumpdisk

<etno>I would be glad to try it ! :-)

<damo22> https://paste.debian.net/plain/1303510

<damo22>you need to substitute the partition number for part:2

<etno>Thanks damo22, I will give it a try as soon as I can (got some work work interrupt :D)

<youpi>you also need the acpi translator in the boot chain

<youpi>grub already does that for you

<gnu_srs1>Why do I get this response when mailing to #bug@bugs.debian.org or control@bugs.debian.org?

<gnu_srs1>Your message to Python-modules-team awaits moderator approval

<gnu_srs1>The reason it is being held: Message has implicit destination

<youpi>gnu_srs1: because the maintainer field of the package is the python-modules-team mailing list, but it's moderated so that'll have to pass through moderation

<etno>damo22, youpi: it turns out that I was missing the host/device ports on the first server (shame). Except ACPI, which hangs forever, any other server, when it is the first one throws a panic with "invalid opcode" 😳

<gnucode>hello hurd people.

<gnucode>I guess I am going to try to install the Hurd on a T410 today.

<gnucode>with a SATA SSD.

<gnucode>Will the regular Hurd installer work? Or do I need to install Debian GNU/Linux on the T410 first. Then create a partition and somehow install Debian Hurd from GNU/Linux.

<gnucode>I also wonder if the trackpad will work...

<gnucode>and I just realized that I left me Debian installer CD back at my house...grr.

<gnucode>I guess I'm going to send in another wiki update to talk about the AArch64 port.

<gnucode>did someone port netsurf to Debian?

<gnucode>apt search netsurf-common shows netsurf 3.11-1

<gnucode>I think that is a no.

<gnucode>netsurf-common shows up, but netsurf-gtk does not.

<gnucode>apparently netsurf released a new version that support flex.

<gnucode>I think that is the same thing as flexbok

<gnucode>flexbox*

<dsmith>fluxbox ?

<gnucode> https://css-tricks.com/snippets/css/a-guide-to-flexbox/

<gnucode>flexbox

<gnucode>also, if I am going to try to port netsurf to debian...which I should.

<gnucode>I can always use apt source netsurf-common.

<gnucode>but that doesn't give me the .git directory.

<gnucode>I guess I should clone netsurf and copy the .git ?

<nikolar>gnucode: can't you just make a copy of the package and change the hash of the commit or something similar

<nikolar>copying .git is a bad idae

<nikolar>idea

<gnucode>"make a copy of the package" means what?

<nikolar>i am more used to arch packaging so sorry if i am using wrong names and terminology

<nikolar>but basically copying the recpie file for the package (PKGBUILD in arch), and making changes to that

<gnucode>hmmm, I'm really not familiar to the way debian builds stuff.

<gnucode>let me go read the hurd wiki. They might have specific steps.

<nikolar>you should probably read about packaging on debian wiki

<nikolar>and working with local copies or something similar

<nikolar>gnucode: ^

<gnucode> https://www.debian.org/ports/hurd/hurd-devel-debian#porting_issues

<nikolar>that's if you already know how to work with debian package

<nikolar>packages

<gnucode>my main issue, is I have ran "apt source netsurf-common". Thay gave me the source code.

<nikolar>but it's the wrong version/commit?

<gnucode>I currently have netsurf-common-3.10

<gnucode>the latest version is 3.11

<gnucode>so yes, I need to run apt source netsurf-common-3.11

<gnucode>but once I have the latest source code...and I make changes so that it builds...

<gnucode>how do I generate a patch? a diff? without git ?

<nikolar>there is a diff command

<nikolar>i assume you use that

<nikolar>patching is pretty common on debian so i assume it's not that hard to find out

<gnucode>hmmm.

<nikolar>i suspect this should be useful https://wiki.debian.org/Packaging/Intro

<gnucode>well I guess I can try that.

<gnucode> https://www.debian.org/devel/ <-- that is probably helpful too

<nikolar>in the worse case, you can make your fork with hurd changes until you figure something else out

<nikolar>yeah that seems like a good resouece

<gnucode>ok I am trying out the Debian GNU/Hurd installer on my T410.

<gnucode>ok, the installer has network access

<gnucode>and my keyboard works.

<gnu_srs1>gnucode: FYI: Regarding contributions to GNU/Hurd I did enable file record locking and bootstrapped/ported ada/gnat and go to Hurd :)

<gnucode>gnu_srs1 you are referring to my changes to the wiki's developer page?

<gnu_srs1>Yes, at the developer page

<gnu_srs1>Ask youpi for confirmation.

<gnucode>gnu_srs1 I believe you...what's your name?

<gnu_srs1>Ask youpi

<gnu_srs1>I'm hosting the mahler buildd.

<gnucode>what is the mahler build? Is that a CI?

<gnucode>also thanks for the file record locking stuff!

<gnucode>and youpi I've got quite a few wiki patches that are in several different emails. If you'd like, I can send a new patch series. So that all of the changes are in one place.

<gnu_srs1>there are two buildd's for GNU/Hurd: ironforge and mahler: See e.g. https://buildd.debian.org/status/architecture.php?a=hurd-i386&suite=sid

<gnucode>is mahler running the Hurd on real hard ware?

<gnucode>and it looks like the Hurd installer detected my SSD. That's pretty cool.

<gnucode>WARNING: the kernel failed to re-read the partition table on /dev/wd0 (No such file or directory). As a result, it may not reflect all of your changes until after reboot.

<gnucode>Hopefully that is not a huge problem.

<gnucode>looks like it is trying to create format in / & /home as ext2. so that's a good sign.

<gnucode>and it's installing!

<gnucode>so apparently I need to reboot.

<gnucode>"Before the installation can proceed, a root file system must be mounted on /target. The partitioner and formatter should have done this for you.

<gnucode>My current thought, is that that warning about the kernel could not re-read the partition table needs I just need to reboot and try again.

<gnucode>ahci: 00:1f.2: Port 0 Communication not establed. Power on your device.

<gnucode>whatever that means.

<gnucode>I think I saw something that just said that / was not writable.

<gnucode>so of course writing stuff to that disk is going to fail...

<gnucode>writing stuff to disk -> installed base packages

<gnucode>installing*

<gnucode>I also might just be using the wrong installer too...

IRC channel logs

2024-01-09.log