Eric,

If you're using a recent GNU Mach kernel, your "kernel" line in GRUB should
read:

kernel=/boot/gnumach.gz root=hd1s1 -s

This will boot into single user mode.  The GNU Mach kernel, as well as the
serverboot module, are now compressed.  Consequently, your module line will
read:

module=/boot/serverboot.gz

As far as extracting the tar file, it uses the directory "gnu" as its root.
So typically I'll create a dummy gnu directory and mount my destination
drive to that directory, e.g.:

cd $HOME
mkdir gnu
mount /dev/hdb1 gnu
tar -zxpf hurd.tar.gz
umount /dev/hdb1

I hope this helps.

Kevin Musick


-----Original Message-----
From: Eric Augustine [mailto:[EMAIL PROTECTED]]
Sent: Thursday, June 15, 2000 9:22 AM
To: [EMAIL PROTECTED]
Subject: help-hurd




Hello,

        I apologize for the following lengthy message, however I'm
        hoping there's something wrong in my procedure so I've written
        it out in painful detail:

        I've gone through both documents on the installation of
        The Hurd:

        Make a partition with fdisk, typed 83 of size 850M (/dev/hdb1)
        Then put the Hurd-owned FS on the partition:
                mke2fs -o hurd /dev/hdb1  (v1.18 of e2fsprogs)

        I mount the partition, cd / and extract the archive:
                tar --same-owner -zxvpf /archive/hurd/gnu-20000301.tar.gz
gnu

                (using the exact command line from Matthew Vernon's guide
                 gives me an error since the archive is created without
                 the leading /)

        This gives me a filesystem like this (under /gnu):

                bin   doc        home     lib         mnt
servers  usr
                boot  dpkg-hurd  hurd     libexec     native-install  share
var
                dev   etc        include  lost+found  root            src
                dict  games      info     man         sbin            tmp

        (If I cd into /gnu and then extract the archive, as according
         to the installation directions at debian.org I'd have /gnu/gnu
         and then the above filesystem... this didn't seem right.  Though
         after enough failures I tried this as well)

        I then umount /gnu and boot off of the grub floppy.  Once grub
        comes up I hit 'c' and get the GRUB> prompt and enter:

        root=(hd1,0)            

        which should point to /dev/hdb1.  This gives me the recognized
        ext2fs and the type '83'.  The next step,

        kernel=/boot/gnumach root=hd1s1 -s

        gives me "File not found" (as did using gnumach.gz).  As a test 
        I attempted:

        kernel=/native-install

        Which gave me an illegal file type - so then I copied a 
        decompressed  gnumach.gz to /gnu (under Linux) and
        rebooted once again and when I get to the appropriate
        point I try

        kernel=/gnumach root=hd1s1 -s

        This gives me the expected information about ELF object
        code.  Moving on to the "module=/boot/serverboot" command
        I run into the same problem - so I apply the same trick,
        but to no avail.  Any variation gives me 'File not found.'

        What am I doing wrong here?  I originally followed the 
        instructions to the letter, even creating filesystems
        on /dev/hdb1 so that the docs would match up directly.
        I began variations when the instructions did not work
        for me.

        I'm installing from RedHat Linux 6.2.  The target disk
        is a 10G drive, however, only the first (850M) slice is 
        allocated for The Hurd.


Thanks, in advance


        --Eric Augustine

Reply via email to