Boot weirdnesses

[bscott at ntisys.com]

  I encountered a couple of weird problems booting my primary home computer
today, and I wanted to share.

  Possibly relevant details:

- Red Hat Linux 7.3
- GRUB 0.91-4
- Epox EP-8K7A+ motherboard (w/ latest BIOS)
- AHA-2940UW SCSI host adapter (BIOS 2.20)
- IDE hard disk
- Multiple SCSI CD/DVD drives

Weirdness #1
------------

  All of a sudden, I couldn't boot Linux anymore.  I would still boot my
Wintendo partition, but not Linux.  GRUB appeared to process the "kernel"
and "initrd" commands properly, but when I issued a "boot" command, it would
simply hang -- no messages at all.  A three-finger-salute was required to
regain control.

  I figured my GRUB install must have been hosed somehow, so I tried booting
my GRUB floppy and loading the GRUB config file from the hard disk.  Same
problem.

  I could use the GRUB "root", "find", "cat", and "testload" commands to
explore my Linux boot partition without finding anything wrong.

  The "debug" command changed nothing -- still no messages after the "boot"
command.

  Eventually, I booted from CD (this led to the second weirdness; see
below).  I ran fsck on everything.  No problems.  So I mounted my hard disk
filesystems and started poking around.  Eventually, I did "cat grub.conf"
and that's when I found "it".

  My grub.conf file ended with a line with no trailing newline.  That is,
the file did not end with a newline character.  I added a newline to the
file, unounted, sync'ed, rebooted, and ta-da, it worked.

  There must be some kind of in-memory corruption that occurs when GRUB 
tries to process a config file without a trailing newline.  I say this 
because I have multiple stanzas in my config file for Linux (for various 
kernel choices), and *all* of them failed.  So it wasn't just the last 
stanza that was hosed.

  The reason my floppy boot didn't help is that I still loaded the config
file from the hard disk, which apparently triggered the same bug.

  This is an older version of GRUB; it may be fixed in newer releases.  
Regardless, though, it's something to be aware of.

Weirdness #2
------------

  In the process of trying to figure out weirdness #1, I attempted to boot
from CD.  I used a CD-R burn of White Box Enterprise Linux 3.0 Respin 1,
mainly because that was what I had on my (physical) desktop when weirdness
#1 started.

  I loaded the CD in my Plextor CD-RW drive and reset the machine.  The
Adaptec BIOS found the bootable CD and said it was going to boot it.  Then
the GRUB menu I usually see when booting from HD appeared.  (The broken
one.)

  "Huh?  What happened to the CD that the Adaptec card was supposedly going
to boot from?"

  I even went so far as to tell my mainboard BIOS Setup not to boot from
anything but my SCSI adapter, and the weirdness still occurred.  I know the
BIOS option did *something*, because if I set it that way and did not put a
CD in any drive, the machine failed to boot anyway.

  I eventually discovered that if I boot that same exact CD in my Pioneer
DVD reader, everything works as I would expect it.

  Haven't had a chance to pursue this to my satisfaction yet.  If I find out
more, I'll let the list know.

  -- Ben "I hope your computer problems are less insane then mine" Scott

-- 
Ben Scott <bscott at ntisys.com>
| The opinions expressed in this message are those of the author and do  |
| not represent the views or policy of any other person or organization. |
| All information is provided without warranty of any kind.              |