At 18:26 Uhr +0200 08.07.1998, Frederick Bruckman wrote:
>On Wed, 8 Jul 1998, Dr. Bill Studenmund wrote:
>
>> On Wed, 8 Jul 1998, Frederick Bruckman wrote:
>>
>> > Why can't we have a disklabel in the NetBSD partition, the way the i386
>> > port does? That system is very flexible. You can craft partitions outside
>> > of the NetBSD partition, if you like, and it still coughs up a
>> > pseudo-label if there is no real one. In the past, we haven't seemed to
>> > need it as much as they do.

Thank goodness.

>> > DOS-like systems are limited to four
>> > "physical" partitions, but we can have (seven or) eight. That's more than
>> > enough for say, a 1G disk, but it starts to be restrictive with larger
>> > drives.

Let me remind you that this is not a limitation of the native Macintosh
partitioning scheme but of NetBSD.

>> What I'd think is cool is we just add a seperate MacOS partition which has
>> the disk label. It doesn't need to show up in the NetBSD disklabel, so we
>> don't need to loose a letter, and we can save it.

I agree that MacBSD should do native disklabels, even more so now that we
have a bi-endian FFS. But cascading partition tables the way the i386 port
is forced to do is one of the closest things to brain death that I know
of...

The main stumbling block in that area is NetBSD's limitation of eight
partitions per disk, effectively leaving you six partitions (currently
five, if your disk is larger than 2 GB  =8/ ).

>Yes, that might work. Not losing a letter is the whole point. There's no
>requirement that the disklabel cover the whole disk--you just omit any
>partition that you don't expect to mount, be it a driver, partition map,
>NetBSD disklabel, or MacOS partition. Another cool thing about the i386
>disklabel is that you can edit and "enable" it without rebooting. You
>could have a spare letter or two, and switch that to a different part of
>the disk on the fly. New style disklabels can have comment lines.

Cascading partition tables simply to overcome the eight partitions limit is
gross. You wouldn't need to multiplex letters if we had something like a
max of 32 partitions per disk.

>The
>biggest problem I can see is that people could possibly be confused by all
>the possibilities.

You might want to search the port-i386 mailing list archive for
disklabel/disk geometry related mails. Last I set up a NetBSD/i386
disklabel it literally took me hours to decide which of the various
geometry guesses I got was workable: Three tools give you four different
"disk geometries" which are all faked anyway. The NetBSD setup tool will
happily create overlapping partitions.

Again: The limitations we encounter are based in the crufty NetBSD
disklabel scheme, not in the (rather flexible) Macintosh partitioning
scheme.

	hauke


[N.B.: There was a longish thread about native disklabels on port-mac68k in
Oct 1996, and another one about partitioning issues ("wedges") on tech-kern
in Jan 1998.]


--
"It's never straight up and down"     (DEVO)