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

Hurd Traffic #41 For 29 Mar 2000

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

Mailing List Stats For This Week

We looked at 90 posts in 287K.

There were 44 different contributors. 18 posted more than once. 13 posted last week too.

The top posters of the week were:


1. Value Of The Hurd Login Shell
15 Mar 2000 - 21 Mar 2000 (51 posts) Archive Link: "Re: "Small" Bug"
People: Marcus BrinkmannNorbert NemecZsombor GergelyMichael Thaler

One of the longest threads I've seen on the debian-hurd mailing list. Continuing an earlier discussion started by Alan P. Laudicina, Michael Thaler pointed out that the Hurd's login behaviour currently rejects a user if a nonexistent login is entered, even before the password is entered. The objection had been raised that this behaviour would inform potential hackers if a given login name existed or not, which could be conceived as a vulnerability. Michael added his voice to this objection, and asked very simply, what advantage would be gained from the current behaviour, over the traditional system of waiting for the password to be entered before rejecting a failed login. Marcus Brinkmann and others replied that the behaviour really made no difference at all for security.

Zsombor Gergely veered off a bit, asking what the purpose was of the Hurd's login shell. Marcus replied (perhaps somewhat exasperatedly), "You are desperately in need for a shell, to read your email (via IMAP), browse the net, check a domain name, or whatever. A machine is nearby running the Hurd. You go to the machine and use it." Norbert Nemec replied that for comparison, one could consider having a guest account on a Debian system as being essentially the same thing as the Hurd login shell. Marcus replied, "Just for the record: No standard unix guest account can clone the flexibility of the not logged in user on the Hurd." Edmund GRIMLEY EVANS asked why this was, and Marcus replied, "Because the fourth permission set for the not logged in user is independant from the others. (I think you can do a lot with frobbing groups and group ownerships on unix, but on the Hurd you can do all this and have an additional permission set :) "

The discussion went on for awhile.


2. Selecting Drivers For The oskit-mach Kernel
20 Mar 2000 (2 posts) Archive Link: "Re: cross compilation tool versions"
People: Igor KhavkineRoland McGrath

Igor Khavkine asked,
I wonder if it's possible to include only selected drivers from oskit into the oskit-mach kernel. If you get the debian oskit package it seems to have all the drivers compiled in. Would I have to recompile the oskit sources with selected options for that?
Roland McGrath replied,
I have been meaning to add some configure options for this but have not gotten around to it. If you are recompiling the oskit yourself anyway, then you can select the drivers to compile by editting a few files in the oskit sources (oskit/linux/ethernet.h and oskit/linux/scsi.h) to remove the lines for drivers you don't want. But, there is no need to recompile the oskit just to avoid using all its drivers. In oskit-mach, edit the source file oskit/ds_osenv.c and replace the call `oskit_linux_init_devs ();' with individual calls to the list of driver initializers you want, i.e. `oskit_linux_init_scsi_ncr53c7xx ();' and `oskit_linux_init_ethernet_eepro100 ();' and the like. The aforementioned oskit header files list all the driver names.
End Of Thread.


3. Another Simple Hurd Translator In Perl
21 Mar 2000 (1 post) Archive Link: ""
People: John TobeyMarcus Brinkmann

John Tobey pointed back to Marcus Brinkmann's simple translator to run 'fortune', and announced that he had reimplemented it in 'perl'. He added that it required 'pith' and could be invoked with 'settrans -fgac test ~/hurd/ fortune', and posted the code:


use Hurd::Trivfs;

if (!@ARGV) {

die "Usage: $0 command arg...\n";

$support_write = 0; $/ = undef; @Run::ISA = ('Hurd::Trivfs'); Run->start; exit 0;

sub Run::io_read {

my ($self, $prot, $bufref) = @_; if ($prot->po->hook) {
# After the first read, return eof. $$bufref = '';
} else {
$prot->po->hook (1); $$bufref = `@ARGV`;

There was no reply.x1


4. Device File Creation
21 Mar 2000 - 24 Mar 2000 (5 posts) Archive Link: "Making Devices on First Install"
People: Marcus Brinkmann

Bryan Walton installed the Hurd using the tarball method, and had no problems. Now, with a functioning system, he wanted to install device files according to the hardware he had available: CD, floppy, PCMCIA etc.; he asked what devices he needed to make, and what labels they should have. Marcus Brinkmann replied:

All devices are listed in the MAKEDEV script in some way, however, this might not help you alot. Note that not all hardware you want to create devices for might be supported.

SCSI and IDE devices are sdX and hdX, where X is a number from 0 to something. (hd0, hd1, hd2, hd3). This is the whole disk. Partitions on this disk are either in BSD disklabel format (then you know how they are named), or they are dos labels, then these dos partitions are called slices (linux uses dos partition table format as well, mind you). Slices are represanted by sY, where Y is a number. For example: hd0s1 is the first primary partition on the first disk etc. (The number is the same as in linux). The floppy device is called fd0 for the first floppy (does anybody still have more than one drive?). Serial ports are com0, com1 etc, but they probably won't work with the Hurd term translator. PCMCIA is not supported. Network devices are created on the fly. So, for example:

$ cd /dev
$ MAKEDEV com0
$ MAKEDEV hd0s1



5. Difficuties Downloading Packages From Server
25 Mar 2000 (3 posts) Archive Link: "Re: Open SSH available now!"
People: Marcus Brinkmann

Soete Joel was unable to find the 'openssl' and 'openssh' packages at, where Marcus Brinkmann had put them. Marcus replied that the server configuration seemed to have changed, to not give a directory listing anymore. He had no control over the server configuration, and explained that apparently it was now necessary to specify filenames in the URL. He listed the relevant filenames (libssl09-dev_0.9.4-4_hurd-i386.deb, libssl09_0.9.4-4_hurd-i386.deb, openssl_0.9.4-4_hurd-i386.deb, ssh-askpass-ptk_1.2.1pre24-1_all.deb, ssh_1.2.1pre24-1_hurd-i386.deb, and ssleay_0.9.4-4_hurd-i386.deb), but then replied to himself an hour later,
It turns out that it is screwed up. This will probably not fixed until beginning of next week. Please try again later, and complain to the webmaster if it doesn't get fixed within reasonable time.
End Of Thread.


6. 'libgcj' Ported To The Hurd
25 Mar 2000 - 26 Mar 2000 (7 posts) Archive Link: "libgcj"
Topics: Compiling Packages
People: Chris LingardMarcus BrinkmannJim Franklin

Chris Lingard announced that he'd ported the Java library libgcj-2.95.1 to the Hurd, and gave a link to the sources. It had taken a lot of patching to get it to work, and required 'gcj' and 'c++' from a very recent version of 'gcc'. He offered to post his patches, and Jim Franklin expressed interest in seeing them. Chris posted them, and Marcus Brinkmann had some criticism about the quality of part of the code. Chris replied:

I offer this thing as a prototype, could it be useful? If people are going to use it then it is worth spending some time on. If it turns out to be an experiment, to see the Hurd compiling c++ and Java; but it is of no use, then a hack is the best way.

It just seemed an amusing experiment to build this library. I now have c++, Fortran and Java, plus the more normal gcc languages. I think I will port Ada next.

Marcus replied, "Hell, if you ported some software, make sure it propagates, so someone can use it when he needs to. Even if currently nobody wants to do c++ and java, some other day many might want."







We Hope You Enjoy Hurd Traffic

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.