Subject: Re: default /usr partition
To: None <tech-install@netbsd.org>
From: Greg A. Woods <woods@most.weird.com>
List: tech-install
Date: 09/04/1999 15:55:06
[ On Sunday, August 22, 1999 at 23:41:30 (-0700), David Brownlee wrote: ]
> Subject: Re: default /usr partition 
>
> 	The 'bsd partion' idea is nice enough but how does it work with a
> 	disk arranged as:
> 
> 	+----------+
> 	| Non BSD  |
> 	+----------+
> 	| BSD      |
> 	+----------+
> 	| Non BSD  |
> 	+----------+
> 	| BSD	   |
> 	+----------+
> 	| Non BSD  |
> 	+----------+
> 
> 	What does the 'bsd partition' cover? Its not general enough to
> 	cover all cases. Maybe it covers from the start of the first BSD area
> 	to the end of the last, or maybe just the boot block and disklabel
> 	area.

If this arrangement of "slices" were possible then I'd expect there to
be two or four logical disk devices, each with separate BSD partitions
and sub-devices that just happened to reside on the same physical disk
(instead of on separate disks as would normally be the case with
separate groups of disk devices).

The original 3B2 port of AT&T UNIX used this concept of "slices" to
better differentiate low-level "fdisk-style" partitioning from
higher-level filesystem partitions.  I'm not sure any more if the 3B2
low-level partitions were actually "fdisk" compatible, but I believe
they were (there were a maximum of four slices, IIRC).

Note that this scheme would let one create a lot more partitions (eg. on
a really big physical disk) without increasing the number of possible
partitions per "disk", if one were inclined to do such things.

> 	If not, then maybe we do need another partition for 'boot block
> 	and disklabel area'...

That would be more flexible than the current 'd' nonsense, I think....

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>