On Mon, Mar 27, 2000 at 04:09:19AM -0800, Todd Whitesel wrote:
> The "rules"
> are too complicated to adequately explain without getting drowned in
> minutiae, and I don't think any of us even knows all of them... It has
> taken quite a lot of head-banging and brainpower from NetBSD luminaries
> to get the BIOS installation logic even to the point it's at today.

It's not that complicated, but there just are some limitations that
you need to be aware of. The -current installer should get it right.
The 1.4.2 installer will too, with one unfortunate bug left: you
should not pick "cylinders" as unit during MBR partitioning, it will
make the partitions too small (this is documented in LAST_MINUTE for
1.4.2).

Tips for commonly made mistakes:

	1) Don't modify the notion that the installer has of the BIOS
	   geometry. It will get it right, just trust it. The option
	   to override it is only there for rare, buggy BIOSs that
	   totally mess up, but that will be easy to spot (you'll get
	   very weird values).

	2) While doing MBR partioning, the only number you need from
	   the "real geometry" is the total size in sectors. The
	   cyl/head/sector values of the real geometry are usually totally
	   unrelated these days. Forget about them, and don't let them
	   confuse you.

In this case, it seems that the SCSI BIOS fakes up a strange CHS
combination. 1023 * 63 * 32 will only get you at 1G through the
traditional BIOS interface (I'm not talking about the int13 extensions).
Since your disk 9G in size, it should have picked the maximum values
(1023 * 63 * 255). Seemingly, it does seem to be using that internally,
but it's telling the outside world something else, hence the problems
when you actually use it.

If the BIOS is not too old, you can try to enable "int13 extensions".
In any case, your partitioning looks ok, so try playing around
with your BIOS settings.

- Frank