Subject: Re: Large disks on an IPX?
To: None <port-sparc@netbsd.org>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: port-sparc
Date: 08/18/2002 23:05:52
> I recently put a 9gb scsi drive into my IPX.  However, in trying to
> newfs it I've run into a problem.

> total sectors: 18591744
>   c: 18591744        0    unknown                        # (Cyl.    0 -
>   f:  1228800  8388608     4.2BSD     1024  8192    16   # (Cyl. 4096 -

> sd0(esp0:0:0):  Check Condition on CDB: 0x2a 00 00 92
> bf ff 00 00 01 00
>    INFO FIELD:  9617407
>      ASC/ASCQ:  Logical Block Address Out of Range

> So the end of the partition os out of range.  I haven't quite
> narrowed down the cut-off point exactly though.

Note that if the disk was sold as 9G, it is probably actually more like
8.3-8.4G; disk manufacturers routinely claim disk sizes several percent
higher than reality.  The c partition size above is some 8.86G.

However, the error you got was from more like 4.6G into the disk.  I
find it hard to believe even a disk manufacturer overquoted the disk
size that much; either you don't have the disk you think you do or it's
very weirdly broken.

What do the boot messages say the drive's capacity is?  It sounds to me
as though the disklabel says the drive is larger than the drive thinks
it is, and if you've got a 4G drive and think it's 9G, this is about
what I'd expect would happen.  Where did you get the disklabel that's
on there now?  If you generated it, what did you start with?  Was the
disk sold as intended for use on a Sun - or, if it wasn't new, did it
come from a Sun?  (That is, did it come with a Sun label on it?)

> So either I'm doing something seriously wrong with disklabel (and it
> has been a while), or my IPX doesn't support harddrives above a
> certain size.

I don't specifically recall trying a drive that large on a IPX.  But
from what I know of how the SCSI subsystem fits together, there isn't
much that could be wrong with the machine to produce a limit in the 4G
range.  If the SCSI interface didn't deal with 10-byte CDBs, things
would start breaking at the 1G point; since that's not happening, it
ought to work up to the 2T point.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B