Hurd Traffic #18 For 6 Oct 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

Mailing List Stats For This Week

We looked at 75 posts in 210K.

There were 21 different contributors. 12 posted more than once. 8 posted last week too.

The top posters of the week were:

 

1. Installation Difficulties; Terminal Translator
21 Sep 1999 - 2 Oct 1999 (19 posts) Archive Link: "Hurd stopped working :("
Topics: GGI
People: Marcus BrinkmannKalle Olavi NiemitaloBrent FulghamPontus Lidman

Pontus Lidman recently upgraded his Hurd system to use the latest packages on one of the mirror sites, but found that his system started locking up right after the partition check during boot-up. Brent Fulgham replied that he'd experienced a similar problem with a slowly failing disk drive. He asked how old Pontus' disk drive was, and Pontus replied that it was pretty old.

Kalle Olavi Niemitalo also replied to Pontus' original post. He had had a similar problem, which he'd tracked down by downgrading to a known good installation, and then upgrading gradually, on a package-by-package basis. In that way he managed to trace his own problem to something in /hurd/init, at which point he stopped his investigation.

As an aside, Kalle added that he intended to write a translator for displaying text with colors, but Marcus Brinkmann replied, "Definitely don't spend your time on colored text. Anyway, this change has to be done in GNU Mach, and now that we are already talking about it, I have a patch ready for attributes and colors, and Brent did NMU ncurses with a appropriate termcap entry. (and when we switch to linux console, we will have coloured text, too, of course). However, working on a fancy terminal translator is encouraged, to make a kernel driven console driver unnecesasarz at some later time."

Kalle replied that the "fancy terminal translator" was what he had meant: "a program which pokes characters to video memory without using the Mach console driver."

But Marcus pointed out:

No, this is unrelated. The term translator handles input stream of characters. It would be useful to get the scancode directly and do mapping in the translator.

What you mean should probably be handled by a device similar to framebuffer device in Linux. The GGI people have some ideas here, I am sure, and i think they are interested in contributing a translator. Maybe this translator could do text mode as well.

At this point the discussion broke down and, as there were several points of confusion, petered out.

 

2. apt-get Failure
26 Sep 1999 (2 posts) Archive Link: "apt-get trouble"
Topics: Apt, FS: FTPFS
People: Marcus Brinkmann

Kai Großjohann installed apt-get but got errors. Marcus Brinkmann replied, "There is no usable version for apt on the Hurd yet. The version in the archive is broken. I will port apt really soon now, so just hold your breath and use ftpfs or debian-ftp." EOT.

 

3. Some Explanation Of The Hurd's Partition-Size Limit
27 Sep 1999 - 28 Sep 1999 (5 posts) Archive Link: "partition size"
People: Marcus BrinkmannMichael BacarellaThomas Bushnell

Someone asked if the Hurd's 1G limit on partition sizes had to do with the boot loader, and Marcus Brinkmann replied that no, it was actually the amount of virtual memory the mach kernel could address at one time. He added, "There are already patches to fix this, but they are kept back for stability. (People checking out an old CVS version can access them if they want to, but nobody will help you with this, I think)."

Michael Bacarella expanded on this, replying to the original poster with, "It's a limitation of mmap(). The Hurd filesystem translators, in all cleverness, mmap() the entire partition for it's added conveniance. I'm not specifically sure what causes tasks to only be able to have 1 gig of mappings, but I'm guessing things like stack/text segments occupy the other gig. On the other hand, when reading literature on the Mach microkernel, it discusses how UNIXisms like text/data segments were phased out in favor of a broad unified memory model without text/data partitioning so I don't see how a mmap() limitation applies here. A task should have a whole 2 gigs of address space, according to my limited understanding."

Thomas Bushnell concluded peripherally, "That's more or less correct. The limit of 1GB is what we tell people; some people have reported getting partitions a bit larger to work fine."

 

4. PowerPC And SMP Support
29 Sep 1999 (3 posts) Archive Link: "PowerPC or SMP support?"
People: Marcus BrinkmannMark Kettenis

Troy Benjegerdes asked if anyone was working on SMP support in the Hurd; Marcus Brinkmann replied, "No. Note that the SMP support belongs in Mach, not in Hurd. The SMP support in GNU Mach is at least broken. There is a patch on alpha.gnu.org/gnu/hurd/contrib/uch (ftp://alpha.gnu.org/gnu/hurd/contrib/uch) but I am quite sure that some hacking is involved to make it work."

Mark Kettenis also replied to Troy, saying, "The Hurd itself is almost entirely multithreaded. However, the SMP support in GNU Mach is not working at the moment. I'm not aware that any of the developers owns a SMP machine, so I don't think there will be working SMP support in the near future. If the SMP support in OSF Mach is in a better shape, the Hurd would probably be able to take advantage of it."

Troy had also asked if anyone had tried building the Hurd on top of the MkLinux osfMach kernel or the Mach kernel in Darwin on a PowerMac, and Mark replied:

I have done some work to make the Hurd running on top of OSF Mach on the ix86. I didn't have the time to finish it yet. There are some differences between OSF Mach and GNU Mach, but I think it might be possible to port the Hurd to OSF Mach on the PowerPC with some effert.

The Mach kernel in Darwin is based on Mach 2.5, which is pretty old and probably lacks a lot of functionality that the Hurd depends on.

 

 

 

 

 

 

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 kernel.org. All pages on this site are copyright their original authors, and distributed under the terms of the GNU General Public License, version 2.0.