SCSI, IDE, and LILO, oh my!
I've been running a system with a 400MB IDE drive and a 4GB SCSI drive.
The hardware involved is as follows, from presumably least significant to
most significant:
ASUS P5A motherboard
128MB RAM
AMD K6-II/400
NCR 53c825 Fast/Wide SCSI-II controller
Ancient 400MB IDE drive (/dev/hda)
IBM Ultrastar 9ES 4GB SCSI drive (SCSI ID 0, /dev/sda)
To this, I have just added:
Seagate ST15150N 4GB SCSI drive (ID 3, /dev/sdb)
The IDE drive is a Win98 C: drive. The only Linux thing on it is LILO.
The IBM drive had a 2GB FAT32 partition that was a Win98 D: drive. Now,
I've added this Seagate, and put another 2GB FAT32 partition on that.
I copied over all the stuff from the IBM to the Seagate, and did a quick
fdisk and marked the IBM partition as type 93, "Amoeba", whatever that is.
I booted Windows, and it happily uses the Seagate as the D: drive and no
longer sees the partition on the IBM. I run a few tests, no problems.
Now we get into the meat of things. I want to yank out the IDE drive, and
use the 2GB partition on the IBM as the new C: drive. After all, 400MB is
just hopelessly inadequate to store an OS as great as Win98. :-< (Honest,
we will get to Linux-specific stuff shortly, I swear.) I copied all the
stuff from the IDE disk to the IBM drive, turned it back into a FAT32
partition, and then edited my /etc/lilo.conf to boot Windows from
/dev/sda4. That's a primary partition, marked bootable. (At this point,
LILO itself is still installed on the IDE drive. It's just trying to boot
Windows from the IBM drive instead of the old IDE drive.)
I rerun /sbin/lilo, reboot, and try to boot Windows. It just says
"Loading win..." and sits there. I reboot, and it boots Linux just
fine.
Next, I try editing LILO to install the LILO boot sector onto /dev/sda,
and not /dev/hda. I then reboot, go into the BIOS and tell it to boot from
SCSI and not IDE.
Now, I just get an endless stream of "LI"'s, which apparently means:
>The first stage boot loader was able to load the second stage boot
>loader, but has failed to execute it. This can either be caused by a
>geometry mismatch or by moving /boot/boot.b without running the map
>installer.
Okay, finally, now my questions:
1. Why can't I boot a LILO off of my SCSI drive, even to just boot
Linux? It should at least be able to do that. I made sure the
IBM was ID 0, just in case it mattered. Should I try yanking the
IDE drive completely from the system?
2. Even booting off the IDE drive, I can boot Linux but not Windows. I
can only think of two reasons:
a) Windows is that stupid, and I'll need to reinstall, or
b) that paritition is above the 1024-cylinder limit, but I thought
that wasn't really a problem on SCSI drives?
Anyway, I hope someone can help me sort this out.
Sincerely,
Ray Ingles (248) 377-7735 ray.ingles@fanucrobotics.com
"Sooner or later the first system is finished, and the architect, with
firm confidence and a demonstrated mastery of that class of system, is
ready to build a second system.
This second system is the most dangerous system a man ever designs."
- Fred Brooks, in "The Mythical Man-Month"
Reply to: