scsi tape quirks?
Ron Stanonik
stanonik at nprdc.navy.mil
Thu Aug 24 07:51:00 AEST 1989
We recently attached an exabyte to a sun4/280 running sunos4.0.1. While
becoming familiar with the exabyte, we ran into some quirks of scsi tape
handling; ie, the archive tape drive behaves the same.
1) The non-rewinding device seems to rewind if you close before
reading the filemark. For example, make a tape contain one file,
two blocks long
dd if=/etc/termcap of=/dev/rst0 count=2
Now try reading the tape one block, at a time
dd if=/dev/nrst0 count=1
dd if=/dev/nrst0 count=1
Both dd's returned the same (first) block.
2) You cannot read data past a filemark, unless you first read the
filemark and close. For example, make a tape containing two files
dd if=/etc/termcap of=/dev/nrst0 count=1
dd if=/etc/format.dat of=/dev/nrst0 count=1
Now read several blocks without closing. For this you need a little
program that opens /dev/nrst0 and does several read's. The first
read returns the termcap block, but successive reads only return 0.
On the other hand
dd if=/dev/nrst0
dd if=/dev/nrst0
will read first the termcap, then the format.dat.
Maybe this all is according to scsi tape semantics (if there is such a
time)? I'm not familiar enough with scsi to say whether this is correct
or not.
1/2" drives on our vax (4.3bsd) and suns (xt, not scsi) behave
differently. That is, the non-rewinding device doesn't rewind, and reads
will read past a filemark. That's what I would expect.
Ron Stanonik
stanonik at nprdc.navy.mil
More information about the Comp.sys.sun
mailing list