Hallo Leo,

> > 	real mem 26378240
> > 
> > I'm pretty sure there is 26 MB somewhere inside, so I wonder what
> > happened with those 864 KB which are not reported. (BTW the page
> > size is 8 KB).
> Some memory is reserved for video buffers (need to be in st-ram) and
> some is taken for the buffer cache.

OK, I assumed real mem would mean all memory detected. So what is the
other one, `avail mem', which says 24494080? (1840 KB less than `real')

> a: root, b: swap, c: whole disk

So a partition with ID `NBR' is assumed to contain a root fs, right?

> > For sd2 the list is even shorter, as their are no FAT fss on that disk.
> I think they modify the partition-id too, when I know their id's I can
> fix this.

There is no one-to-one correspondence between a partition ID and the type
of filesystem on that partition. Partitions with ID `RAW' for example may
contain either an ext2 or a minix fs (or even an msdos fs). The only ID
that has a special meaning is `XGM' (extended partition). But if NetBSD
really wants to use partitions IDs to decide the type of filesystem, we
could define `MX1' for minix fs, `MX2' for minix2 fs, `EX1' for ext fs
and `EX2' for ext2 fs (allthough none of those are currently supported
by NetBSD, and probably never will be). I would also suggest using `SWP'
for swap partitions, since there has been some discussion on those in
one of the Maus groups, and `SWP' seemed to be favoured by most.

That leaves me with the question which ID to use for partitions that
contain an ufs? (e.g. /usr/src, /var, /var/spool/ftp, /var/spool/news,
/usr/local etc.). Currently different IDs are used for root fs and
/usr fs, is it planned that each additional partition gets a unique
ID? (e.g. NBS, NBV, NBF, NBN, NBL...)

> > Never mind,
> > 	mount_msdos -o ro /dev/sd1d /mnt
> > would it...
> > 	dma-ready: code 2
> Probably a parity error...

Seems harmless, the same message is displayed 8 times in a row before
the ram fs image is read from floppy.

> > 	mount_msdos: mount: Invalid argument
> > oops, celibrated too soon.
> try: mount_msdos -r /dev/sd1d /mnt
> You only need '-o' when going through 'mount' I think (I don't have
> NetBSD at hand here).

It's the other way around. Mount understands -r, -w, -v etc., mount_msdos
understands -o <options>, -u <user>, -g <group> and -m <mode>. Passing -r
to mount_msdos directly, results in an `illegal option'. The problem is
not the `-o ro', I get the same result (`invalid argument') with just
`mount_msdos /dev/sd1d /mnt'. I can't figure out what is wrong here.

> > in writeable mode. Could someone tell me how this is done?
> Why would you want to mount it read/write? I never felt the need to do
> that.

Because I assumed that the mountpoint had to be writeable to actually
mount a filesystem on it. But in the meantime I discovered that the
install script doesn't care about that either. So now the question
has become, `how can I mount an msdos fs?'. Has anyone ever success-
fully mounted an msdos fs under NetBSD/Atari?

I'm curious what will happen if I try to run newfs, but I didn't do
that yet, as there's not much use in creating a root or /usr fs if
I cannot install the archives from an msdos fs.


Waldi