Subject: Re: SCSI _is_ still broken on VS3100
To: maximum entropy <>
From: Marc Gutschner <Marc.Gutschner@Frankfurt.NetSurf.DE>
List: port-vax
Date: 01/07/1999 20:41:07
Maximum Entropy wrote:

>>From: Andrew Gillham <>
>>Date: Thu, 7 Jan 1999 12:04:54 -0500 (EST)
>>Is SCSI meant to work on the VS3100/M38?
>It's *meant* to work on all VAXstations, I think :-)
>>My bus/drives probe correctly,

Mhhm, but there seems to be still something strange in the SCSI support.
I've two busses (ncr0 and ncr1) in my config but only ncr0 is probed by the
kernel. ncr1 never shows up during the hardware scan :-( And yes, I have a
SCSI-SCSI daugther-board in my machine :-)

>>but I see the following:
>>vax# disklabel -r -w sd0 rz23
>>vax# newfs sd0a
>>write error: 102431
>>wtfs: Undefined error: 0
>>Any ideas anyone?
>I think you've labeled an rz22 disk with a disklabel for an rz23.

Doesn't seem to be the only possible cause. I have a pair of RZ23 in my
VS3100/m38. I did a 'disklabel sd0 > /tmp/rz23' to get a prototype label
with the view of the drive as the kernel sees it. Then labled the disk with
this Prototype and used 'disklabel -i' to create the partitions
interactively. This all goes really smooth but as soon as I try to 'newfs'
the created partitions I get - uhm - erratic behaviour. If I use 'newfs
sd0a' it immediately bombs out with a similar error as the one above. If I
then try 'newfs /dev/sd0a' chances are about 50-50 that a filesystem will be
created or it stops with a error like the above.
But things start to get _really_ weird as soon as something is written to
the discs. I want
'/', swap and '/var' on one RZ23 and '/usr' on the other, so I mounted them
under '/mnt'. As soon as I write to '/mnt/usr' the whole mountpoint seems to
vanish ! I.e. there is no '/mnt' any longer and a 'umount /mnt/var' (or
'/mnt/usr') just yields a 'No such file...' :-(

Any ideas anyone ?

MTIA for your patience,