At 23:30 Uhr +0100 24.10.1998, Dave Schmitt wrote:
>> On Sat, 24 Oct 1998, Dave Schmitt wrote:
>>
>> > I am trying to upgrade from a 700MB drive to a 3.2GB drive but am having
>> > some problems. I'm running NetBSD 1.3.1 on a MacIIci w/ a generic SBC
>> > kernel.
>
>Well, the current partitioning is as follows; most letters come from
>Installer.
>
>a 	NetBSD root	~300MB for / (with /tmp & /var)
>b	NetBSD swap	~40 MB (double RAM)
>c	[whole disk IIRC]
>d	an FWB driver partition (exact name escapes me at moment)
>e	HFS volume	~200MB for MacOS etc
>f	NetBSD usr 	~500MB for /home
>g	NetBSD usr 	~500MB for /usr
>h	NetBSD usr 	~500MB for /usr/local
>i	NetBSD usr 	~150MB for /mac/utils (*)
>j	NetBSD usr 	~1 GB  for /mac/netdist (*)
>
>(*) volumes that will be used for Mac files served over netatalk...
>
>The FAQ said something about keeping to 6 partitions, which is probably
>because of the maximum h partition you mentioned; I thought that meant
>NetBSD partitions, so that's what I did. Is the Installer just coming up
>with the HFS and driver partitions incorrectly? I.e., will netbsd actually
>be ignoring these partitions, thus shifting f-j down to d-h? If so, can I
>safely ignore the Installer's mistake or are things gonna get screwy?

For all I know, you cannot have more than a total of _7_ partitions on a
drive that has more than 2 GB - this includes sdXc (the whole disk) and the
driver partition as well as MacOS partitions. So you end up with a maximum
of sdX{a,b,c,e,f,g} plus the MacOS driver partition.

I have seen crashes with an IBM DCAS 4G drive and a 4.5 GB Seagate the very
moment I added an eight partition. Others have reported success with
different disks.

In principle, NetBSD/mac68k sees a total of 8 (eight) partitions per disk.
The others are ignored and are not accessible; the kernel's choice is
somewhat less than intuitive, especially if you change, remove and add a
few partitions with the MacOS formatter.
The whole thing sucks, if you ask me (I know, you haven't ;).

Scott Reynolds has made some changes in -current a month ago that probably
fix the problem but when I had a drive to test the changes with the kernel
sources wouldn't build, and now that they build again the drive has been
sold to a customer...

You could probably try a -current kernel from a binary snapshot or so.

	hauke


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