Very Computer

I just installed Debian 2.1 on my system (Asus P2B, BT-958 SCSI, two
UW-SCSI hard drives & LS-120 on hda), and I can't seem to upgrade the kernel.

I compiled a new kernel, tried creating a boot disk.  A plain ("dd")
boot disk refuses to even start booting, a LILO boot disk gets as far as
"LI", and now I've rendered Debian unbootable by trying LILO on the hard
drive.

LILO is configured to install on /dev/sdb4 (my Debian partition), as
I use OS/2 Boot Manager as the primary loader.  The partition is just over
4 Gb in size.

When I ran lilo -v on the new kernel, I got a warning that BIOS device 0x82
might not be accessible.

Then when I tried to boot, as soon as I chose Linux from Boot Manager, I
got

        L   01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 ...

ad infinitum.

I checked the LILO documentation, and it says this "shouldn't happen"
(and a few equally helpful things).

I tried booting into my other Linux partition, SuSE 5.2 on /dev/sdb3.
I mounted sdb4 on /new, tried restoring the old kernel image, and
running "lilo -v -r /new"

and got "First boot sector does not have a valid LILO signature".

The LILO docs just say I might have the partition name wrong, but
I can't have since I've mounted it.

The final piece of the puzzle is this: LILO installed perfectly well
via the Debian installation routines (with the default install kernel).
I'm using the exact same lilo.conf file as was installed by default in
/etc (which I assume is what it used before).  All I did was change the
symlink /vmlinuz to point to the new kernel.

I've tried adding/removing parameters like "compact" and "linear", to
no avail.

I've also tried using the exact same lilo.conf contents as my SuSE
installation on /deb/sdb3 has (with the partition name changed).
But it refuses to accept it.

It can't be the hard drive, because SuSE works well, and Debian installed
perfectly well, too.

It can't be a one-time fluke, either, because this same thing happened
to me earlier in the week, using the debian-style kernel install (I
just reformatted & reinstalled because I couldn't bother fighting with it).

Can somebody PLEASE help me...?

Thanks...
-----------------------------------------------------------------
 Alex Taylor                  BA - CIS - University of Guelph

-----------------------------------------------------------------

If I'm reading you right, you have both EIDE and SCSI disks on the same
machine, and you're trying to boot Debian on /dev/sdb4.

You can't do that. [1]

LILO is telling you it's having a really hard time finding your boot
partition.  Unless you've got a newfangled MB and BIOS, an EIDE drive
preempts all SCSI devices.  You may want to muck around in BIOS and/or
SCSI setup to see if your board is an exception to this rule, but
chances are you're SOL.

The SuSE install is an interesting twist.  Are you sure there's not
something you've forgotten to tell us here?  How about a copy of
/etc/lilo.conf?

Best bet is to boot from floppy, then move /boot to a small partition
(or directory) on your first EIDE drive.  Your Linux root partition may
be on any partition on your system, but LILO has to be able to find your
kernel.  BTW, since all your distros share the same essential kernel,
you should be able to share /boot among them.  You can specify different
kernels and/or root partitions as needed.

Quote:> I checked the LILO documentation, and it says this "shouldn't happen"
> (and a few equally helpful things).

Read it again, carefully and cluefully.  You'll understand it eventually
(or die trying).

[1] Or I'm dead wrong.
--

    What part of "Gestalt" don't you understand?
    Welchen Teil von "Gestalt" verstehen Sie nicht?

web:       http://www.netcom.com/~kmself
SAS/Linux: http://www.netcom.com/~kmself/SAS/SAS4Linux.html    

 10:03pm  up 6 days, 20:07,  9 users,  load average: 0.64, 0.52, 0.43

Sounds as though you didn't get support for your BT-958 in your kernel.  Do
you have boot disks from your original Debian install?

Note: E-mailed *and* posted.

:I just installed Debian 2.1 on my system (Asus P2B, BT-958 SCSI, two
:UW-SCSI hard drives & LS-120 on hda), and I can't seem to upgrade the kernel.
[SNIP]
:When I ran lilo -v on the new kernel, I got a warning that BIOS device 0x82
:might not be accessible.
:
:Then when I tried to boot, as soon as I chose Linux from Boot Manager, I
:got
:
:       L   01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 01 ...
:
:ad infinitum.

That's your confirmation that BIOS device 0x82 is indeed not accessible.

I suspect what's happening is that you've configured your BIOS to boot
from the SCSI disks, not the LS-120, so the two SCSI disks get
configured as devices 0x80 and 0x81.  The LILO installer doesn't know
you've done that, and assumes that hda (the LS-120) is the first BIOS
disk.  Perhaps you installed the LS-120 after you did the other Linux
installations, so the confusion never occurred before.

You just need "BIOS=0x80" and "BIOS=0x81" directives in your lilo.conf
file to tell the LILO installer how to address your SCSI disks at boot
time.  The LILO docs will show you the details.

--

PGP public key 1024/9A9C7955
Key fingerprint = 2F E5 82 F8 5D 06 A2 59  20 65 44 68 87 EC A7 D7

I've fixed it!!  Thanks for all the advice, I should have known it would turn
out to be something stupid (if utterly unintuitive).  Read on for details...

First off, I mentioned in my original post that this had happened several days
earlier, and I'd eventually just given up and reinstalled.  What I didn't say
was that I had tried to boot from the install CD (with "rescue=/dev/sdb4"), and
that refused to mount the root filesystem (got a weird error that looked like
a FAT filesystem header).

This time, however, I created a Rescue floppy (same thing, really), and managed
to boot to /dev/sdb4 from it successfully.

I changed the symlink /vmlinuz back to the original install kernel, and re-ran
LILO.  However, I got the exact same error.

FYI, this is the output LILO was giving:

---------------------------------------------
ifurita:/# lilo -v
LILO version 21, Copyright 1992-1998 Werner Almesberger
Reading boot sector from /dev/sdb4
Warning: /dev/sdb4 is not on the first disk
 hda: hda1 hda2 hda4
Warning: BIOS drive 0x82 may not be accessible
Merging with /boot/boot.b
 hda: hda1 hda2 hda4
Warning: BIOS drive 0x82 may not be accessible
 hda: hda1 hda2 hda4
Warning: BIOS drive 0x82 may not be accessible
Boot image: /vmlinuz
 hda: hda1 hda2 hda4
Warning: BIOS drive 0x82 may not be accessible
Added linux *
/boot/boot.0814 exists - no backup copy made.
Writing boot sector.
---------------------------------------------

And my /etc/lilo.conf file is:
---------------------------------------------
boot=/dev/sdb4
install=/boot/boot.b
map=/boot/map
# linear
read-only
vga=normal
delay=20
image=/vmlinuz
        root=/dev/sdb4
        label=linux
---------------------------------------------

Anyway, what I eventually did was this:

(1) I didn't like the messages about /dev/hda in the lilo output, since
    that's my floppy drive, which shouldn't even enter into the picture!  So
    I made certain I had no disks in the drive (I thought I'd tried that
    already, but I could be wrong).

(2) Scouring dejanews for LILO problems, I noticed a common theme:
    BIOS boot virus protection.  The Debian documentation said to turn it
    off, but I think I'd only gotten around to doing that *after* my
    initial install (which was probably pointless, but I figured what the  
    hey).  However, a few people seemed to think turning it on might do
    something.  So I went into Setup & re-enabled it.

I booted back into my system, using a Rescue floppy, took the disk out of the
drive, ran lilo -v, and noticed with some optimism that the BIOS errors no
longer appeared.  (I still got some messages about /dev/hda not having any
valid partitions, for some reason.)  But I rebooted and LILO booted the kernel
successfully (finally)!

What it had restored was the default install kernel, since I'd changed the
symlink back (as you recall).

So then, I changed the symlink /vmlinuz back to my new kernel, and that
installed successfully, too.  

So now I'm happily booting away again.

Since I did both (1) and (2) at the same time, I'm not certain which did the
trick, but I suspect (2).  

However, I still don't know why I'm getting messages about /dev/hda when I run
LILO.

Also, I'm still unable to create a bootdisk.  The LILO bootdisk just stops at
"LI", and the raw kernel image refuses to boot at all.

But that's less of a priority, now.

Anyway, thanks for everybody's help & advice.  Usenet really is a wonderful
safety net...

-----------------------------------------------------------------
 Alex Taylor                  BA - CIS - University of Guelph

-----------------------------------------------------------------

On 14 Apr 1999 02:05:47 GMT,

Quote:

> (1) I didn't like the messages about /dev/hda in the lilo output, since
>     that's my floppy drive, which shouldn't even enter into the picture!  So
>     I made certain I had no disks in the drive (I thought I'd tried that
>     already, but I could be wrong).

Actually, /dev/hda is the primary drive on the first IDE controller.
Your floppy is /dev/fd0.

--
Brian Moore                       | "The Zen nature of a spammer resembles
      Sysadmin, C/Perl Hacker     |  a*roach, except that the*roach
      Usenet Vandal               |  is higher up on the evolutionary chain."
      Netscum, Bane of Elves.                 Peter Olson, Delphi Postmaster

Not so, my floppy is an LS-120 drive on the first IDE controller, and is
my only IDE device.  Hence, /dev/hda.

I have no devices on /dev/fd* - in fact, I've disabled my MB's floppy
controller in the BIOS, since I have no need of it.

--------------------------
Alex Taylor

On 15 Apr 1999 22:46:07 GMT,

It is also the primary drive on the first controller, no?

This story makes MUCH more sense now that you've explained you're not
talking about a floppy, but a removable IDE drive.

Quote:> I have no devices on /dev/fd* - in fact, I've disabled my MB's floppy
> controller in the BIOS, since I have no need of it.

Have you disabled booting from the IDE drive in your BIOS?

Many BIOS'es really insist on booting IDE if they find one.

--
Brian Moore                       | "The Zen nature of a spammer resembles
      Sysadmin, C/Perl Hacker     |  a*roach, except that the*roach
      Usenet Vandal               |  is higher up on the evolutionary chain."
      Netscum, Bane of Elves.                 Peter Olson, Delphi Postmaster

The "L 01 01 01 ..." problem happened to me when I changed
my IDE Secondary master on my drive.
It disappeared when I was able to boot again (in my case to Dos
with a boot disk but it seems that's not your case),
run "FDISK /MBR", then to linux, and it was fixed after reinstalling
Lilo.

Hope this may be of some help to you, but I really don't know anything
about SCSI.

                Pedro Andrade