Re: Installing 2.0.2 onto IBM 7248-100

On 7/25/05, Andy Ruhl <[email protected]> wrote:
> On 7/25/05, Jochen Kunz <[email protected]> wrote:

> > Usually I do it this way:
> > save output of fdisk(8) and disklabel(8).
> > dd(1) kernel image like generic.fs to raw disk.
> >         ATTENTION! This overwrites the MBR partition table.
> > restore MBR partition table according to saved fdisk(8) output.
> > reboot and pray. ;-)
> >
> > Note: If you compile your own kernel you have to convert the ELF image
> > to a .fs image including bootloader and kernel with mkbootimage(8).
> 
> There is no man page for mkbootimage(8) for prep. There is one for
> some other archs, but not prep. All I found was when I built tools for
> 2.0.2, in my tools/bin directory there is a nbprep-mkbootimage binary.
> I did help on it and it gives me this:
> 
> # ./nbprep-mkbootimage -h
> usage: ./nbprep-mkbootimage <boot-prog> <boot-image> [<gzip'd kernel>]
> 
> I can't quite tell what that means. I'm going to try some stuff and report back.

By the way, thanks for your help so far Jochen! I always somehow
forget to say that.

Ok. I'm stumbling through now.

I noticed a /usr/mdec/mkbootimage binary on my prep system. In that
same directory also appears to be 2 bootloaders: boot and boot_com0

I assumed (correctly luckily) that boot is for VGA and boot_com0 is
for a serial console.

I cross built a kernel for this from another machine, so I grabbed
that and gzipped it.

The syntax is this:

./mkbootimage boot netbsd.fs netbsd.gz

where:

boot is the /usr/mdec/boot file
netbsd.fs is the name of the boot image you want to make (it doesn't
exist before you run this command)
netbsd.gz is a gzipped kernel you made.

I netbooted this thing and it works, that's wonderful.

Next, how to boot from the hardware?

I think I have something screwy going on, because I tried to do a dd
and got this:

# dd if=bugs1.fs of=/dev/sd0d seek=64
dd: /dev/sd0d: Device not configured

Where bugs1.fs is the name of the boot image I created and
successfully netbooted.

Also, following convention, the "d" partition in disklabel is
suppposed to cover the entire disk, right? Well, this isn't the case
for me:

# disklabel -r sd0
# /dev/rsd0c:
type: SCSI
disk: mydisk
label: fictitious
flags:
bytes/sector: 512
sectors/track: 309
tracks/cylinder: 5
sectors/cylinder: 1545
cylinders: 11474
total sectors: 17774160
rpm: 7200
interleave: 1
trackskew: 0
cylinderskew: 0
headswitch: 0           # microseconds
track-to-track seek: 0  # microseconds
drivedata: 0

7 partitions:
#        size    offset     fstype [fsize bsize cpg/sgs]
 a:  17231872         0     4.2BSD   2048 16384 28120  # (Cyl.      0 -  11153*)
 b:    542288  17231872       swap                     # (Cyl.  11153*-  11504*)
 c:  17758095     16065     unused      0     0        # (Cyl.     10*-  11504*)
 d:     16000        64     unused      0     0        # (Cyl.      0*-     10*)
 e:  17758095     16065     unused      0     0        # (Cyl.     10*-  11504*)

And here's fdisk:

# fdisk sd0
Disk: /dev/rsd0c
NetBSD disklabel disk geometry:
cylinders: 11474, heads: 5, sectors/track: 309 (1545 sectors/cylinder)
total sectors: 17774160

BIOS disk geometry:
cylinders: 1024, heads: 255, sectors/track: 63 (16065 sectors/cylinder)
total sectors: 17774160

Partition table:
0: Linux/MINIX (sharing disk with DRDOS) or Personal RISC boot (sysid 65)
    start 64, size 16000 (8 MB, Cyls 0/1/2-0/254/63), Active
1: NetBSD (sysid 169)
    start 16065, size 17758095 (8671 MB, Cyls 1-1106/99/34)
2: <UNUSED>
3: <UNUSED>

Any thoughts? Something doesn't look right in there... I think I might
have to re-do the disk...

Thanks!

Andy