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 #24 For 17 Nov 1999

By Zack Brown

Do you pine for the nice days of minix-1.1, when men were men and wrote their own device drivers?
Are you without a nice project and just dying to cut your teeth on an OS you can try to modify for your needs?
Are you finding it frustrating when everything works on minix? No more all-nighters to get a nifty program working?
Then this post might be just for you :-)
-- Linus Torvalds, 1991

Table Of Contents


Next week is Thanksgiving here in the US, so I'll be taking a break from KC Debian Hurd (and Kernel Traffic). See you in 2 weeks!

Mailing List Stats For This Week

We looked at 77 posts in 217K.

There were 25 different contributors. 16 posted more than once. 8 posted last week too.

The top posters of the week were:

1. Problems Building The Hurd By Hand

6 Nov 1999 - 9 Nov 1999 (11 posts) Archive Link: "Trying to build the Hurd"

People: Marcus BrinkmannChris Lingard

Chris Lingard had successfully installed the Hurd, but then wanted to recompile it from scratch. He'd thought he had the necessary packages installed on his system for that, but apparently the build required TeX, which had not been installed. Although the tetex package was available in the archives, Chris found that it depended on the dpkg-perl package, which he couldn't find anywhere. Marcus Brinkmann replied, "The problem is that some parts of tetex are binary-all and updated regularly by the maintainer, while the binary-any part which needs to be recompiled (tetex-bin) is still very old, and is not updated by me. So they get out of sync, making it problematic to get it going."

Stefan Weil pointed out that the Hurd build may have required the tetex package just to get 'makeinfo'. He suggested downloading the (fairly small) 'makeinfo' source from some repository, rather than installing the huge tetex package. Marcus replied, "Of course you are right. I think the current texinfo package in the Hurd (version 4.0) has makeinfo in it and conflicts with older tetex versions. Installing texinfo package might be sufficient to compile the Hurd." But Chris replied that he wanted to build the PostScript manual, which did require the full TeX package.

Elsewhere, Kapil H. Paranjape gave the location of the dpkg-perl package at debian/dists/potato/binary-all/devel, in the archives. Chris, in his original post, had mentioned the possibility of compiling the Hurd under Linux first, and then moving it over to its intended partition. Kapil posted a lengthy explanation of how to do this, and there was a bit of discussion about his method.

2. Hang When Booting To Multi-user Mode

6 Nov 1999 - 8 Nov 1999 (5 posts) Archive Link: "Bootup Problem"

People: Marcus Brinkmann

John Goerzen reported a successful Hurd installation, and had no problems in single-user mode. However, booting to multi-user mode would get to the point of displaying, "Automatic boot in progress..." followed by the date. Then it would hang. Marcus Brinkmann replied that several people had reported that problem, but he had never experienced it himself. He suggested booting first to single-user mode and then logging out. He believed this would take the system to multi-user mode. Stefan Weil confirmed the problem, but pointed out several details: first, gnu-19991025.tar.gz never failed when booting to multi-user mode; second, gnumach-1.2 (1999-09-07) would give reproducible lock-ups; and third, the problem mysteriously went away for no apparent reason.

Stefan's conclusion was that his laptop, which changed CPU speeds as part of its automatic power management features, must be causing the problem. To him this meant that the problem had to do with timings.

If this is indeed the culprit, the subject came up in Kernel Traffic Issue #43, Article 4, and received some lengthy attention.

3. Access To The System Granted To Non-Logged-In Users

6 Nov 1999 - 8 Nov 1999 (9 posts) Archive Link: "A few other problems"

People: Marcus BrinkmannDaniel Burrows

John Goerzen noticed that users who were not even logged into the machine could go anywhere in the filesystem and read any files they chose (including things like /etc/passwd). Daniel Burrows screamed, and he and Marcus Brinkmann pointed out that the Hurd had a fourth set of permissions for users who were not logged in. They suggested changing the permissions to be more restrictive. Marcus added, "It's not a bug, it's a feature."

4. Partition Naming; Linux Compatibility

8 Nov 1999 - 10 Nov 1999 (11 posts) Archive Link: "Newbie questions again..."

Topics: Bootloaders, POSIX

People: Gordon MatzigkeitMichael BacarellaZsombor GergelyMarcus Brinkmann

Zsombor Gergely had a number of questions. He asked about partition naming conventions, and Marcus Brinkmann pointed out that Mach and GRUB had different conventions. In particular, adding a new drive might not change the names of the drives as Mach saw them, but would change their names as GRUB saw them. This was because GRUB used BIOS routines for its partition detection. It followed that syncing the names used between Mach and GRUB would be extremely difficult. Gordon Matzigkeit replied:

Okuji has done some fantastic work on an `ioprobe' command that has overcome this hurdle. There's still more work to do, but it is now feasible to correctly support both Mach and Linux-style partition naming.

For those who are curious, ioprobe works by setting the trap flag, then single-stepping through an INT13 call, scanning the machine code for IO port usage, and recording the ports that were used. This should be enough for us to determine the correct Mach or Linux-style name for the device.

Zsombor also asked about Linux compatibility in general, and whether it would be possible to exploit the full power of the Hurd while staying compatible with Linux. Michael Bacarella replied:

I personally don't care all that much about Linux compatibility. It'd be nice, but a better goal would be POSIX compatibility and forget the idiosyncracies of other operating systems. There's plenty of room to make our own :)

To soapbox a bit, this is the GNU OS. People should want to use it because it's truly free on many levels. Not because it's like some other OS.

5. panic: Cannot load user-bootstrap image:error code 6000

8 Nov 1999 - 10 Nov 1999 (4 posts) Archive Link: "kernel panic"

Topics: Bootloaders

People: Marcus Brinkmann

Marshal Wong had some problems with his Hurd installation, the biggest one being a "panic: Cannot load user-bootstrap image:error code 6000" error and hang at boot-up. Marcus Brinkmann suggested using a newer bootdisk from, since he suspected Marshal's GRUB was old, and unable to decompress the 'serverboot.gz' file. As an alternative, he suggested gunzipping the file by hand, and giving "serverboot" as a module to GRUB. Marshall tried this approach and reported success.

6. Gordon Matzigkeit Interviewed In French GNU Magazine

10 Nov 1999 (1 post) Archive Link: "Re: [Fwd: interview in french GNU magazine]"

Topics: Bootloaders

People: Denis BodorThomas BushnellOKUJI YoshinoriRoland McGrathMark KettenisMarcus BrinkmannGordon Matzigkeit

Gordon Matzigkeit posted an email containing an interview with him (conducted by Denis Bodor) about the Hurd. Since it would only appear in public in a French magazine, he posted the original english to the list. I reprint it here:

Denis Bodor

How many people work on the Hurd ?

Gordon Matzigkeit

Currently, there are five actively involved hackers: Marcus Brinkmann does nearly all the Debian work of porting packages and system integration, OKUJI Yoshinori works on GRUB and GNU Mach, Mark Kettenis is working on the GNU C Library, Roland McGrath and Thomas Bushnell work as a team to fix difficult bugs and add new features to complex parts of the Hurd and C Library.

Outside of that core, there are probably about 30 people who have posted on the Hurd lists, and are trying to get involved with it. We're lacking in documentation right now, and so it is difficult for people just to jump in.

My official role is to run around and find jobs that are being neglected, then either help volunteers take over them, or try to do them myself. I'm mostly a coordinator.

Denis Bodor

The version of the hurd gonna be 0.3, but is it still quite stable. Is there a lot of work to go to 1.0 ?

Gordon Matzigkeit

We need a lot more packages to be ported, more documentation, and a simpler installation process before 1.0. Each new package helps us uncover new bugs, and to improve the system.

Denis Bodor

Some people said that the Hurd can run Object Oriented programs (like Gnome, Gnustep,...) faster. Is it true, Why ?

Gordon Matzigkeit

The microkernel that the Hurd is based on (currently GNU Mach) is designed for message passing. It is possible to make changes to the message implementation in GNOME or GNUSTEP so that they work more efficiently under Mach.

Denis Bodor

do you think that the best way to make the Hurd popular is to make it run directly GNU/linux binaries ? what the hurd need to do that ?

Gordon Matzigkeit

GNU/Linux binary compatibility will be useful, but there are other, more important changes and features that we should probably add first.

The simplest form of binary compatibility would be to emulate glibc's Linux interface on the Hurd. This should not be difficult, since glibc can already run on both the Hurd and Linux. Then, most dynamically linked applications would be able to run on the Hurd. There are other problems, such as dependence on the Linux /proc filesystem, but hopefully not too many binaries require those functions.

Denis Bodor

Is there one developpment team by country ? Here in france, there is no specifique project related to the Hurd, so it's perhaps a good ID to make a call in this paper.

Gordon Matzigkeit

We're an international team, and welcome anybody who wants to get more involved in working on the Hurd. We always need more people who are talented C programmers, and understand kernel implementation, RPC, and thread programming.

People who want to write documentation are especially welcome, since they help other people to understand the architecture, and become kernel hackers.

Denis Bodor

Have you something more to say to incitate our readers to try and play with the hurd ?

Gordon Matzigkeit

The more people who use the Hurd, the faster it will become a stable system. It is a good system to learn how to hack kernels, because so much of the implementation is exposed, and available for you to modify.

7. gnumach Incompatible With Latest GCCs

12 Nov 1999 (2 posts) Archive Link: "gnumach with gcc-2.95.1 or newer"

People: Mark Kettenis

Stefan Weil tried to build gnumach with gcc-2.95.1, and found that the compile would fail at every __asm__ statement. He asked if there were any plans to make the Hurd compatible with the latest compilers. Mark Kettenis agreed that this was definitely a bug, though he didn't expect to be able to work on it himself.

8. Selection Of Programs In The Base Installation; Colors On Text Consoles

14 Nov 1999 (2 posts) Archive Link: "Feedback on Hurd install"

People: Daniel De KokMarcus Brinkmann

Daniel De Kok reported success with his Hurd installation, but complained about the lack of compilers in the base install. He also couldn't seem to get any colors on the text terminal. Marcus Brinkmann replied that the base installation was not supposed to have anything more than what would allow users to continue upgrading on their own. He added that colors were just not supported at this time.







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.