> Hmm, I thought I had proposed something that would have improved the
> situation a bit, too.  Maybe, your proposal improves it on more machines
> than mine, but nonetheless...

So, I just reread all of your mail in this thread, and *nowhere* did
you suggest a mechanism for mapping BIOS geometries to real disks
(which is what we're talking about here).

> As an aside (one might look at this as an anecdote as well :-)), we've got
> here some machine with an LS120 drive, and with this machine, the BIOS, after
> having read the first sector of the disk, assumes, that that's the first
> sector of a FAT filesystem and patches the values for sectors/track,
> number of heads and a byte at offset 0x24 (which doesn't make sense to me).
> Actually, this patching doesn't make sense to me at all...

So, *if* that's true, then the boot block would have to be modified to
have those fields, in order to support `whole disk' installation.
Whatever.  I don't see how that has anything to do with the rest of
this discussion.

> > I don't believe that's possible.  In particular, unless you're doing a
> > `whole disk' installation on a newer machine (one with `large disk'
> > support -- again, assuming the boot block is fixed), then you *have*
> > to deal with the 1024 cylinder limit.
> 
> I'm not sure what you are talking about, when you are saying "`whole disk'
> installation on a newer machine".  If you've got a "newer machine", you
> could install NetBSD anywhere on the disk (given the proposed modified
> fdisk mbr, boot sector and boot program), provided you make sure that
> the MBR code stays the same as the one that NetBSD will write there.

I can only infer that you failed to parse my well-formed English
sentence, because you just reiterated the same thing I wrote (albeit
with less information).