Bug#238251: indigo2 mips install

[Thiemo Seufer <ica2_ts@csv.ica.uni-stuttgart.de>]

Maitland Bottoms wrote:
> Package: installation-reports
> 
> Debian-installer-version: beta3 netboot
> uname -a: Linux indigo2 2.4.22-r4k-ip22 #1 Mon Feb 9 21:48:42 CET 2004 mips unknown
> Date: Mon Mar 15 23:26:48 EST 2004
> Method: netboot from Woody i386 box
> 
> Machine: SGI Indigo2
> Processor: R4400SC V5.0  FPU V0.0
> Memory: 96M
> Root Device: SGI WD93 SCSI and ST15230N disk
> Root Size/partition table: 
> # fdisk /dev/scsi/host0/bus0/target0/lun0/disc 
> 
> Command (m for help): p
> 
> Disk /dev/scsi/host0/bus0/target0/lun0/disc (SGI disk label): 133 heads, 62 sectors, 1017 cylinders
> Units = cylinders of 8246 * 512 bytes
> 
> ----- partitions -----
> Pt#                                   Device  Info     Start       End   Sectors  Id  System
>  1:  /dev/scsi/host0/bus0/target0/lun0/part1               0         3     32769   0  SGI volhdr

The volume header has to be #9. fdisk handles this right, partman fails.
Note that mips switched back to fdisk with the most recent partman/partitioner
in unstable, so this should be fixed for the mirrors soon.

> 11:  /dev/scsi/host0/bus0/target0/lun0/part2               0      1017   8386733   6  SGI volume
> 12:  /dev/scsi/host0/bus0/target0/lun0/part3               5      1000   8213016   3  SGI raw

The type should be "Linux native". Again, fdisk handles this the right way.

[snip]
> Confusion of partition numbering:
> partconf: Creating swap on /dev/scsi/host0/bus0/target0/lun0/part13
> mkswap: /dev/scsi/host0/bus0/target0/lun0/part13: No such file or directory
> 
> should be
> mkswap  /dev/scsi/host0/bus0/target0/lun0/part4

Seems like partconf gets confused by the "hole" between #1 and #11.
It works with partition numbers ascending from #1 (the volume/volume
header partitions don't count WRT).


Thiemo