This card worked fine, and has become the kind of thing that
puts people off Linux :(

The megaraid drivers were split at 2.6.16 into old, and new.
Previously the code caught all megaraid drivers, but when
they split the driver, the "catch-all" died, so it seems if
your IDs don't match as expected you get no disks.

Understanding the various and duplicated bugs is hard,
especially as they overlap with an issue with using the
firmware post 6.61 on this hardware, versus other people who PCI ID just got dropped (mostly HP servers).

My understanding now, is that if you have 6.62 (or newer) firmware (which DELL shipped for a while) then when you go from Linux kernels 2.6.15 to 2.6.16 you have to downgrade the firmware on the controller to 6.61 so that the new Megaraid driver will recognize the card.

Alternative solutions include patching the legacy megaraid driver to recognize this card (as it seems to have worked fine for years!), a patch is given in 369378.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=395174
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=369378

This looks like a duplicate....
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=410817

http://fixunix.com/hardware/510693-lsi-logic-megaraid-cerc-ata-100-doesnt-detect-disks.html

Some insight into the root cause!?

http://www.mail-archive.com/linux-scsi@vger.kernel.org/msg11463.html

So he sent a patch...

http://bugzilla.kernel.org/show_bug.cgi?id=6695

So it'll be fixed in 2.6.26, having been broken for all of Etch, it will presumably be broken for Lenny as well.

I'm going to build a bespoke 2.6.26 kernel.

Update - Debian stock kernel 2.6.28-1-686 just booted fine - I'm getting some rather odd noises from the disks, and the boot time is markedly worse, but it is at least a stock kernel.