I have a DECstation 5000/20 with a RZ23 disk that I thought had a
hardware problem (either the SCSI controller or the disk). After
reading on this list about SCSI problems that others were
experiencing I'm checking if it is a problem in the pmax-SCSI code or
perhaps the cabling.

When trying to do a install (base system only, X11 will be NFS
mounted) I netboot it and try a manual installation. After labling and
doing a newfs on rz2a I mount /dev/rz2a on /mnt and unpack base.tar.gz
to /mnt. After a while messages from kernel appears on the screen
interspersed with tar's output.

asc_last_dma_out: buflen 8192 dmalen 8192 tc 7680 fifo 16

These appear now and then, the difference being the tc-value which is
mostly 7680, but 5632 and 7168 also appear once. Finally the kernel
dies, outputting

mode = 0100040, inum = 2049, fs = /mnt
panic: ffs_valloc: dup alloc
syncing disks... 151 151 148 144 139 132 123 112 99 84 68 50 30 12
rebooting...

The filesystem is trashed beyond repair.

I've done some testing with the disk on an Amiga 3000 running AmigaOS,
and it seems to be ok (using the same cabling). A quick test on
another pmax running ultrix (different cabling) didn't show any
problems either.

Ideas?

-- 
Andreas Johansson			     andjo@lysator.liu.se
Studying Computer science and engineering
University of Linköping, Sweden  http://www.ctrl-c.liu.se/~andjo/