Text 6298, 129 rader
Skriven 2006-06-12 04:55:18 av Maurice Kinal (1:140/13.1)
   Kommentar till text 6293 av Paul Rogers (1:105/360.0)
Ärende: follow up
=================
Hey Paul!

Jun 11 20:44 06, Paul Rogers wrote to Maurice Kinal:

 PR> Not an initrd or kernel?  I don't know what you're doing there.
 PR> Ahh, lilo.conf says it's an initrd.  OK.

Right.  Piece of cake right?

 MK>> M-Systems 40-pin iDOC which dmesg shows as;

 MK>>     hda: M-Systems iDOC V1.01 030804, ATA DISK drive
 MK>>     ide0 at 0x1f0-0x1f7,0x3f6 on irq 14

 MK>> and was mounted on the build system's (/dev/sda1 mounted as

 PR> Mounting an IDE/ATA drive as SCSI?!  You're explaining this
 PR> backwards I think.

No that one is a ide flashdisk.  /dev/sda is a sata drive and the rest are all
flashdisks of various types on usb hubs (sdc is a compactflash and sdf is a usb
flashdisk).  Clear as mud now?

 PR> Or the host system booted from sda1, and then you mounted
 PR> your thumb-drive (/dev/hda1)?  That I can understand.
 PR> (Linux understand sata drives as SCSI???  Wierd.)

Right.  Now you're getting it.

 MK>>     disk=/dev/hda bios=0x80

 PR> Damn straight!  "Button, button, who's got the button?"

No button on that one.  She is solidly mounted directly on the 40-pins
belonging to ide0.  This could easily work on anything you have.

 PR> I don't understand why this isn't enough to tell lilo where
 PR> to write the boot.b.  Disk= does that and boot= is what it
 PR> will be at boot time?

It may change as in the case of the usb or if I move hda over to ide1 and then
it'll be hdc.  Doesn't matter because it'll still boot since the mbr moves with
the disk and lilo is in there.  Given that the root file system will always
mount to /dev/ram0 it won't matter where you physically put the disk as long as
the interface is bootable from the bios.

 PR> Where'd you get the map?

lilo made it when I did 'lilo -C /mnt/flashdisk/etc/lilo.conf' to
/mnt/flshdisk/boot.  I did a rough comparison to the one on the host system and
it is different.

 PR> What you SEEEEEEM to be telling me is lilo wants these to point
 PR> to the files on the host system at the time it's writing the
 PR> boot records, not something the code it writes is going to read
 PR> at boot time, ala grub's /boot/grub/menu.lst, or what the
 PR> target system is going to use itself.

No.  lilo wants to know the targetted system so that the target's mbr gets
written to and NOT the host's.  The host is already booting or it would have
never gotten this far.

 PR> It was on the thumb drive, but referring to files on the
 PR> host, so is going to have to be changed pretty much as
 PR> soon as you get it booted to refer to "local" files?  Or
 PR> there was one of those already there.

Now you have me confused.  The host was a sata drive (dev/sda1) that I booted
to and mounted the ide flashdisk (/dev/hda1) on /mnt/flashdisk on the sata
drive.  hda1 has it's own kernel that it can boot.  Once hda1 was made bootable
it no longer needs the host drive to boot up ever again.  It is it's own little
world but can mount /dev/sda1 and call it a target and itself the host, except
then all the references to /dev/hda will have to be /dev/sda in the target's
special lilo.conf or the wrong mbr will get written to.  Is it getting any
clearer?

 PR> Running the host's lilo in the host's address space.
 PR> Only when I did that I didn't even get a "L".

Yes but in that case you were targetting the host drive from the host.  The
method I am sugestting is totally different but achieves the same result ...
except it works.  :-)

 PR> But let me check, I think perhaps the lilo.conf I made was
 PR> for the target system's address space, and that's what the
 PR> chroot was achieving.  Still, it doesn't quite explain why
 PR> that only got me a "LI", unless I ran lilo by hand with the
 PR> same lilo.conf.

In a chroot enviroment you have to ensure that /proc is mounted and the correct
/dev that matches the physical layout of your pc else it'll fail.

 PR> OK, that's admiting to the fact that it must have been
 PR> referring to the target system's address space.

Correct.  The host can already boot just fine.

 MK>>     Chop wood, carry water.

 PR> Well, Pascal appreciates Zen too, why not you?  Go to the
 PR> same dojo?

:-)

 MK>> Cool eh?

 PR> "Because it's there."  ???  <shrug>

Why else?

 PR> website demonstrates, I'll BUILD the sucker if I have to!

Why not because you can?

 PR> See, I had this idea of a bunch of points in a 7-space that
 PR> I wanted to examine by projecting into a subset 3-space cube
 PR> and rotating it in real time as an astronomer might.  Even
 PR> had/have an S-100 graphics card for this honkingly HEAVY 19"
 PR> Mitsubishi fixed frequency monitor.  Never got anything done
 PR> in that direction though.

Pack it into the trailer and we'll see what we can do for you when you get
here.

Life is good,
Maurice

--- Msged/LNX 6.2.0
 * Origin: The Pointy Stick Society XVIII - Don't know what I want (1:140/13.1)