Larry Glamb <lg@tiki.theeddy.com> wrote:
> Olaf Seibert wrote:
> > MO drives that act like a disk work just fine as a disk. I am using
> > one right now, in fact. The 1.0 kernel (i386) reports it thusly:
> > 
> > ncr0 targ 4 lun 0: <FUJITSU ,M2512A          ,1408> SCSI2 direct removable
> > sd2 at scsibus4sd2(ncr0:4:0): asynchronous.
> > sd2(ncr0:4:0): illegal request
> > sd2: could not mode sense (4); using ficticious geometry
> > : 217MB, 217 cyl, 64 head, 32 sec, 512 bytes/sec
> > 
> > > The drive is reported by NetBSD as not to support a mode sense command.
> > 
> > but then it goes on using some geometry anyway.
> > 
>
> Did you get access to full capacity and have you tried reading the disk 
> on other
> machines or after upgrading to 1.1?

I must admit that the 117 megabytes are rounded down values.
However, the fake initial disk label that is produced gives the
partition size in blocks and that should be the full capacity.

This is slightly annoying, since disklabel then thinks that the
partition does not fit on the disk. Hence, when writing the label
I lied and pretended the disk was 118 cylinders.

I have used the same drive on another machine (a msdos box), but
obviously not with the same disks. I would not expect problems
though with that, or with 1.1, which I'll be installing shortly.
Unix, like AmigaOS, does not particularly care about geometry when
accessing the disk at low level. It is only the file system that
tries to do certain optimizations, and these days you don't want
those anyway. * Besides, modern disks have variable numbers of
sectors per track so no simple geometry can be right.


*: so, when newfs-ing a partition, set rotdelay to 0 and maxcontig to
8 or so, so that rotation delay "optimisations" are disabled.

> larry glamb (lg@theeddy.com)
-Olaf.
--
___ Olaf 'Rhialto' Seibert      D787B44DFC896063 4CBB95A5BD1DAA96 
\X/ There are no lemurs in this post    rhialto@polder.ubc.kun.nl