[Wolves] Chicken-and-egg situation regarding HPT372 RAID controller

[Dan]

Hi all

I have a problem.  BIG problem.

I'm installing a new server at work (Athlon 1800+ processor, 256MB RAM, 4
100GB Caviar HDs) and I want to install Debian 3.0 on it as RAID 0+1.

However, there seems to be a problem with the module.  It'll install OK - I
installed Debian 3.0 on a 'normal' drive, installed the 2.4.18 Kernel and
then the BF24 headers over the top, downloaded the module from
the Highpoint web site, compiled it, and a nice hpt37x2.o module came out.
Installed too with insmod(however it did moan about being a tainted 
kernel) - great, I thought.

So I copied the module to a floppy disk, recreated the RAID 0+1 array, and
during the install tried to install the module.  The system freezes and 
won't
budge.  I tried 'Install Foreign Modules' too but to no avail.

I finally tried dropping to a shell and doing the insmod trick.  AHA!
Success!  Recognized the 0+1 array as /dev/sda and I could install the base
system to that, although I could still see /dev/hde-h as disks.  Apparently
that's normal, though.

The problem is when I come to try and boot the base system.  LILO comes up
fine, and the first part of the boot sequence starts - but the system 
freezes
through not being able to mount the root filesystem.  It would appear that
the kernel recognises the RAID controller as HPT370, the driver for 
which can
only do RAID 0.  Is there any way make the hpt37x2 module live in the 
kernel?
  The problem is that it isn't a standard part of the kernel source tree
AFAIK.  I have NO IDEA how to do this.  HELP!

I have installed a regular 10GB drive onto the standard VIA controller 
on the
mainboard, and installed Debian to that.  It also has the 2.4.18-bf2.4 
source
installed, as I know full well that I'm going to have to *at least* 
recompile
the kernel.  I do know C so I can do source-hacking if required, but as 
I say
I just don't know where to start.  I have one week to get the thing fully
functional.  Eek!

The idea behind the server is that it will act as the home for the 
corporate
database, which is currently in MS Access, but I'm going to be transferring
to PostgreSQL.  The existing clients should be OK with this (minor
adjustments required) since PostgreSQL has windows ODBC drivers.  The 
server
should also act as a PDC (as it does now) but be internet-facing too with
poptop installed for VPN connections.

-- 
Dan