Kernel Traffic
Latest | Archives | People | Topics
Latest | Archives | People | Topics
Latest | Archives | People | Topics
Home | News | RSS Feeds | Mailing Lists | Authors Info | Mirrors | Sleeping Cousins

Hurd Traffic #66 For 8 Nov 2000

Editor: Zack Brown

By Paul Emsley  and  Zack Brown

Mach 4 | Hurd Servers | Debian Hurd Home | Debian Hurd FAQ | Debian Hurd Mailing List | Hurd Reference Manual | Hurd Installation Guide | Cross-Compiling GNUMach | Hurd Hardware Compatibility Guide

Table Of Contents


Want to help write KC Debian Hurd? See the KC Authorship page the KC Debian Hurd homepage, and the Thread Summary FAQ. Send any questions to the KCDevel mailing list.

Mailing List Stats For This Week

We looked at 174 posts in 652K.

There were 61 different contributors. 31 posted more than once. 17 posted last week too.

The top posters of the week were:

1. Realtec 8029 NIC Initialization problems

24 Oct 2000 - 30 Oct 2000 (7 posts) Archive Link: "Newbie - Network problems"

Summary By Paul Emsley

People: Daniel MattnerRobert Bihlmeyer

Daniel Mattner had installed the Hurd but had problems with his network card. He explained " The kernel recognises my network card during bootup. Unfortunately it doesn't initialise it. I seem to have a not common Realtec card (8029). Where can I get a working version of a network card driver and how do I install them ?? " Robert Bihlmeyer helped him diagnose the problem, asking:

Does Mach output a line about "eth0" while booting? Then your chances are good that the driver is there. In that case, the only thing you're missing is the IP stack. How to set that up is detailed in the Easy Guide in step 6.

If Mach is silent about your card, only then do you need to secure a driver for it.

Daniel Mattner then said that he tried unsuccessfully to initialize the card following the Easy Guide using:

/hurd/pfinet -i=eth0 -a .. -g .. -m

Robert Bihlmeyer spotted a typo in this noting " Hmm, the '...=bar' syntax is only for long style options, isn't it? So, try '--interface=eth0' or '-i eth0'." Daniel Mattner reported that that solved the problem, and his network card was initializing properly.

2. The Latest Set Of Boot Floppies

25 Oct 2000 - 31 Oct 2000 (10 posts) Archive Link: "Hurd boot-floppies "C""

Summary By Zack Brown

People: Philip CharlesGordon Matzigkeit

Philip Charles announced:

I have prepared a new set of Hurd boot-floppies and they are sitting in a tar.bz2 file of 42 MB. There is no documentation included and the 1.2 floppies are omitted. The file is small enough for me to upload somewhere. Is anyone interested?

The main change is in the installation menu where inappropriate options have been removed, eg, the network/ftp installation option is gone as the installation ram-disk is still Linux.

IMHO, it is worth considering placing these on the Debian mirror despite the "purity" issue, however there will need to be some discussion about this. As I am stuck in the New Maintainer's pipeline I cannot do anything about it at the moment.

I have worked out how to make the source available in a space saving way and I will be sorting this out over the next couple of days.

I hope to look at the CD after this. A considerable amount of work needs to be done to make certain that everything works and I will probably get onto this next week. This will be quite a long job and will need coordination between Hurd hackers and CD builders.

As a Debian maintainer, Gordon Matzigkeit volunteered to sponsor putting the work on the mirrors, but said he'd wait for further discussion. He added, "My own opinion is that this is a step closer to native Hurd boot floppies. Once we can get the changes to the programs accepted for installing the Hurd, then we can work on getting the programs changed to run on the Hurd. In any event, these floppies make it cleaner for installing the Hurd on a spare machine."

3. L4 Instead Of Mach

27 Oct 2000 - 2 Nov 2000 (44 posts) Archive Link: "L4 instead of gnumach?"

Summary By Zack Brown

People: Ron FarrerOKUJI YoshinoriFarid Hajji

There was a lot of discussion about porting the Hurd over to the L4 microkernel, and ditching gnumach entirely. Apparently the porting issues would be considerable, and there was a lot of resistance to a migration. There was also a lot of support, and the subject seems very controversial. Ron Farrer posted the initial proposal this week, arguing:

  1. No one really wants to work on gnumach.
  2. OSkit-mach is too slow. The COM interface in oskit is (IMO) too bloated (this is in regards to many people saying "use oskit-mach").
  3. No one has really "actively" worked on gnu/mach in over 5 years (roughly).
  4. gnumach only has x86 support in it, other platforms have a LOT of work get get things working.
  5. gnumach is not 100% GPL'd
  6. L4 IS 100% GPL'd
  7. L4 is actively being developed.
  8. L4 is cross platform today
  9. The Alpha and MIPS ports are almost entirely written in assembly
  10. L4 is fast!
  11. L4 team is working on SMP righ NOW.
  12. the HURD developers don't have to work on the kernel anymore as L4 has it's own team of developers. :)
  13. various other, and probably minor, issues.

Along the way, the possibility of dropping 'glibc' support from the Hurd also came up, as did the idea of creating a 'VK' (virtual kernel) API, that would provide a standard set of semantics to the Hurd. As Farid Hajji argued, Mach, Linux, L4, or any other kernel, could provide those semantics, once they'd been defined. At one point several people suggested that a new mailing list was really required, and OKUJI Yoshinori announced:

This is an announcement of a new mailing list, "l4-hurd". The list l4-hurd will be used to discuss L4 port of Hurd. You can send e-mail to the address If you want to subscribe to the list, see the webpage

Please use the list only if your message is specific to L4-Hurd. General questions about Hurd (or L4) should be sent to help-hurd (or l4-hackers).

4. The Hurd On 'plex86'

29 Oct 2000 (10 posts) Archive Link: "Second shot at PAM"

Summary By Zack Brown

Topics: Bootloaders, Emulators: plex86

People: Stefanus Du ToitMarcus Brinkmann

In the course of discussion, Stefanus Du Toit asked Marcus Brinkmann, "Marcus, I saw on the plex86 list that you tried to get gnumach/hurd booting at some point. How far have you gotten with this?" Marcus replied, "I am watching plex86, and I think Yoshinori does, too, although probably only as far as GRUB is concerned and not GNU Mach. I would like to try out the latest version, which can run linux well, but plex86 does not work on linux 2.4, which I have installed, so I have to wait. If someone has linux 2.2, they could try out plex86 ( CVS version and tell us how far they come..." Stefanus said he'd give it a try.

5. 'gnupg' Under X; 'rxvt' Success

29 Oct 2000 (8 posts) Archive Link: "gnupg under X"

Summary By Paul Emsley

People: Martin StenzelRoland McGrath

Martin Stenzel had problems running gnupg in an xterm:

When I try to work with gnupg (1.0.4) under X I get the error message:

gpg: cannot open /dev/tty: Device not configured

and gnupg aborts.

In terminal mode I have no problems. I executed the /dev/MAKEDEV script that did generate a /dev/tty. I am not sure whether this script comes as standard, executing ls -l /dev/tty gets me:

/dev/tty: can't stat '/dev/tty' (Device not configured).

Martin added that he hadn't been able to test gnupg in 'rxvt', which he hadn't been able to compile successfully for the Hurd. Roland McGrath addressed the terminal control issues in xterm:

The symptoms you see suggest something broken in how the pty got set up.

You said that tty works, so you can see that the real tty is be /dev/ttyp0 or whatever it is. What does "ls -l /dev/ttyp0" give you? If it gives you the same error, then /dev/tty is fine and it's just redirecting to the actual terminal device, which has a problem. If using /dev/ttyp0 directly works ok but /dev/tty doesn't, then it is a problem with the redirection that /dev/tty is supposed to do.

Looking at the source, I suspect you will find that /dev/ttyp0 works ok. I can't figure out how ENXIO (Device not configured) would from term.

The libc code that gets invoked for /dev/tty will return ENXIO if the cttyid port is null. So, I suspect that xterm failed to properly set the controlling terminal of its child process (your shell). This is done with the TIOCSTTY ioctl. So take a look at the code in xterm trying to do that and see what results it is getting.

Marcus thought that this would probably explain ^C and ^Z not working in xterm either. He also posted a patch to get rxvt to compile. Martin replied that the patch did work, and that gnupg worked properlywith rxvt. So the problem had been with xterm all along.

6. Adaptec 2940UW

30 Oct 2000 (5 posts) Archive Link: "Adaptec 2940UW problem"

Summary By Paul Emsley

People: Martin MaciaszekMarcus Brinkmann

Martin Maciaszek tried to boot the Hurd, but got a hang after his Adaptec AHA-2940UW was detected. Marcus Brinkmann replied that the Linux driver used in gnumach was old and broken; and suggested that Martin cross-compile a new gnumach with an updated driver for the 2940, using the latest Linux 2.0.x.

Martin replied, " that's too much work for now. " But Marcus suggested that it was actually not so difficult, saying, "As you probably already have a compiler for i386 linux, let me tell you that with make-cross, you can have a compiler in a couple of kilobytes (not megabytes). And, gnumach can be compiled with the linux compiler as well."

There was no reply.

7. Latest XFree86 4.0 Experimental Packages

30 Oct 2000 (1 post) Archive Link: "XFree86 4.0.1 phase2v21 packages"

Summary By Zack Brown

Topics: Apt

People: Marcus Brinkmann

Marcus Brinkmann announced:

I uploaded the binary-hurd-i386 packages of the *highly* experimental xfree86-1-4.0.1-0phase2v21 packages (to get more info on the phase progress, see the X strike force page at The packages are not build from the 0phase2v21 source, but from this source plus additional patches which I sent in the last weeks to the debian-x list (see the Debian list archive for details). I should probably make a grand diff of the whole beast, but I am currently short on disk space. Anyway, that's only so that you can see it break on your home machines.

I don't know how to update to these packages, it involves a lot of --force-conflicts etc. The resource is NOT apt-able, it would be a waste of time to set this up. You need some binary-all packages from, too.

There is a good chance that your card is not supported, or the support is broken. The original X 4.0.1 package worked *better* than the phase2v21 package which is built from the current CVS X version. The X 3.3.6 worked *better* as well. I am only uploading the phase2v21 packages to be cool.

Yesterday, phase2v29 appeared, so maybe that will include most of the Hurd fixes, and with phase2v4024 it will work out of the box.

Find the following files at


There was no reply.







Sharon And Joy

Kernel Traffic is grateful to be developed on a computer donated by Professor Greg Benson and Professor Allan Cruse in the Department of Computer Science at the University of San Francisco. This is the same department that invented FlashMob Computing. Kernel Traffic is hosted by the generous folks at All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License version 2.0.