Kernel Traffic
Home | News | RSS Feeds | Mailing Lists | Authors Info | Mirrors | Stalled Traffic

Hurd Traffic #76 For 17�Jan�2001

Editor: Zack Brown

By Paul Emsley �and� Zack Brown

Mach 4 | Hurd Servers | Debian Hurd Home | Debian Hurd FAQ | debian-hurd List Archives | bug-hurd List Archives | 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 45 posts in 160K.

There were 24 different contributors. 9 posted more than once. 6 posted last week too.

The top posters of the week were:

1. Booting lockup

1�Jan�2001�-�8�Jan�2001 (7 posts) Archive Link: "GRUB lockup during boot"

Summary By Paul Emsley

People: Roger Williams,�Robert Bihlmeyer,�Ognyan Kulev,�Neil Walfield

Roger Williams reported his problems while booting the Hurd. He saw a freeze after: (scsi 0) Downloading sequencer code 419 instructions downloaded. He wondered if his Adaptec 2930B Ultra SCSI card was really supported. Neil Walfield replied that he has an AIC7895 and it works fine and that it may be configuring the next device (the network card) that is the cause of the problem.

Roger then replied: "I believe you are correct about the conflict not coming from the SCSI controller. I tried going in and removing each of my PCI & ISA cards one at a time, including the SCSI controller, and the lockup still occurs. I even had a spare video card that I pulled out of the dustbin and installed. " . But still his problem remained.

Robert Bihlmeyer thought that the problem maybe due to a FIC SD11 mobo. "On second thought, the conflict I encountered may be with the VIA chipset that is used by many Athlon boards. So disabling eatadma may be a good idea even if you use another board. If this affects most Athlon boxes, it would be good to turn the eatadma driver off per default. "

Prabhu Ramachandran offered a pointer to an older message on cross compiling gnumach and Ognyan Kulev chipped in with "I have AHA-2940 and gnumach doesn't work with too. Someone said it doesn't work with AHA-2920. The drivers of gnumach are from Linux 2.0.36 and have to be updated. I tried and I didn't succeed by just replacing files."

2. Hurd CD ISO image

8�Jan�2001�-�10�Jan�2001 (7 posts) Archive Link: "ISO image"

Summary By Paul Emsley

People: Philip Charles,�Vadim Chekan

Vadim Chekan wanted to know if he had to install Linux to use the ISO Hurd image. Philip Charles replied "The Hurd CD boots Linux and creates a ramdisk in memory and this is used to partition the Hard Drive, format the Hurd partition and install the the tarball. The Linux phase only invoves RAM. At no time is Linux installed on the HD." Philip then went on to say that the CD documentation is lagging and that the E series on which he is currently working will be better.

Where is the D ISO image? See here:

3. Package Updates

8�Jan�2001 (2 posts) Archive Link: "Package updates now entering the pool"

Summary By Zack Brown

Topics: Apt

People: Marcus Brinkmann,�Philip Charles

Marcus Brinkmann announced:

here is a list of packages I uploaded today, they should be installed in the archive rather soon now. They will now go into the package pool rather than the distributions directories. This has some consequences to some people, but I don't know much about that (notably, there are issues with the cd creation scripts, but Phil seems to have it under control).


libterm-slang-perl will follow (should be there, but upload failed). Note that the new debconf requires apt, which isn't there, because 0.3.19 is broken and 0.4.x not there (yet). --force-depends is your friend, but I don't know if it will work without apt (haven't tested it).

Philip Charles replied:

An interesting list, but they are not in the archive yet.

The CD creation scripts are being worked on and just need to be debugged. We are now into the second CD.

End of thread.

4. Status Of PPP

10�Jan�2001�-�12�Jan�2001 (5 posts) Archive Link: "ppp"

Summary By Zack Brown

People: Daniel E. Baumann,�Roland McGrath,�Marcus Brinkmann,�Petros Sidiropoulos,�Ben Collins

Petros Sidiropoulos asked about the status of PPP in the Hurd, and Daniel E Baumann replied:

The good news is I have ported user space PPP and it has been done for some time, but haven't been able to test it due to certain glibc features that are not there yet (some socket ioctl() calls), although Marcus tells me he will be done soon. I tried doign some things manually and copying over some libs that he gave me, but I got strange errors about "Protocol Family Not Supported" for AF_INET. Marcus has basically told me that I may be missing some things that he has on his system and it is best to wait for new debian packages. So whenever new debian packages come into the mix I can get it working. I am not about it jump into the fun that is glibc. I am leaving that to Marcus ;). I could have written it all using fsys_set_options, but that is not too portable, nor the recommended method.

So I just ask that all be patient, it will be there soon enough.

But Roland McGrath interjected, "These bugs are no reason not to post your code." There was no reply, but under the Subject: PPP port available via SourceForge CVS, Daniel announced:

Upon popular request, to make the code avaiable, I have imported the user space PPP port ino the HURD SourceForge CVS tree. You can check it out by doing the following:

Anonymous CVS Access

cvs login

cvs -z3 co ppp-hurd

Developer CVS Access via SSH

export CVS_RSH=ssh

cvs -z3 co ppp-hurd

Most if the code is in ppp-hurd/ppp, the other directories have not been touched so files in the examples, libhack, pppctl directories have not been changed. Also there is a ChangeLog in ppp-hurd/ppp as well as a a hurd directory in ppp-hurd/ppp which contains some of the same files as the parent directory, but with specific hurd implementations of those functions. This code will not be tested or debugged until there is a glibc and hurd update containing support for various socket() ioctl() calls, which Marcus tells me "now it's really done". So soon we will have a working PPP port ;), then I will make a debian package and hopefully HURD life will be a little better for some :).

Marcus Brinkmann thanked him heartily, and added:

I am currently uploading gnumach 1.2-9 (not necessary for this) and hurd-20010111, which contains the Hurd part of those changes.

I send the glibc patch to Ben Collins, and he said he will add it to the next version of the glibc Debian package (at the weekend, I think).[1] So by the beginning of next week, I will have the Debian glibc packages recompiled. I could provide you with experimental pakages right now, but I think the few days more wait are okay.

5. More on Hurd CDs

10�Jan�2001 (3 posts) Archive Link: "Progress on the Hurd CDs, E series."

Summary By Paul Emsley

Topics: Apt

People: Philip Charles,�Marcus Brinkmann

There was some discussion between Marcus Brinkmann and Philip Charles about the E series Hurd CDs.

Philip started off with:

At the moment dependency checking has been disabled so all packages are included, except those on the exclusion list. This could be changed if required. However, debian-cd will not include badly built packages and I am reluctant to try to disable this.

The test installation was the best so far. dselect was used and the dependency problem was not as bad as feared. There were far fewer crashes than in previous installations. I suspect the Hurd, dselect and my AMD K5 machine (32 MB RAM, 50 MB swap) don't get on very well. The machine bought a couple of farms.

A new tarball can be dropped in with little effort. The second CD boots off the Hurd installation compact boot image.

They went on to discuss checking, apt and exclusions.


Philip asked:
Should the dependency checking be disabled? (I would suggest yes).

Marcus replied: "Do you mean the checking in debian-cd? I hope not that you mean any checks in dpkg or dselect :) We certainly want to fix all dependency problems. Most often this is a problem of not up-to-date packages. Several people here are becoming Debian developers now (they are in progress), so they can upload recompilations and package updates soon. I am happy to provide some initial help how to do it correctly (there are a couple of minor problems in dpkg build tools etc. that require careful attention occasionally). "

Philip replied:

debian-cd normally checks for dependencies so that the first CD has no un-met dependencies, and the same for the set. A consequence of this is that if package cannot have its dependencies met it does not get onto the set. This has been disabled so everything that is not excluded is on the CDs, but packages on the first CD might depend on what is on the second.

Being one of the people in developer pipeline, the sooner the better. I am not happy about the way I am forking debian-cd and boot-floppies.


Philip asked "What is the status of apt in relationship to the CD? apt is on the exclusion list, yet people use it by forcing its installation. "

Marcus replied "apt is not available. I think 0.1.9 is in the ftp archive, but this is horrible broken. apt 0.3.19 doesn't compile out of the box, but I tested the current CVS version and this works very well. We can only hope that an apt 0.4.x is released soon. I can make a temporary package available on (now that it is clear that the next official version really will work), so I might just be going to do that, and alpha is on the CD, IIRC. "

Philip confirmed that it was and said that the sooner we have a working version of apt, the better.


Philip asked: "What should be in the exclusion list? exclude.txt will be on my site ."

Marcus replied:

As we are with two CDs already, I don't think anything should be excluded for technical reasons. We can save some space by excluding obviously silly packages which are not installable (let's say, everything that is not installable and of priority optional or extra). Something like that.

The MOST important issue distribution wise is the following: dpkg needs to be able to treat Arhcitecture fields of linux-all, linux-any, hurd-all, hurd-any, so that linux specific packages can be marked and treated as such without listing all linux architectures. This field is also parsed by the ftp maintenance scripts and at some other places, so everywhere updates are needed. This feature is now actually feasible because of package pools. This is really most important.

When this feature is available, we can start to file bug reports against such packages (makedev: linux-all, modutils: linux-any etc). This will clean up a lot of the mess.

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.