[comp.unix.ultrix] EOF problems on TK50 and TK70 tapes

raph@tigger.planet.bt.co.uk (Raphael Mankin) (08/01/89)

The problem with reading tape marks on TK50 and TK70 drives sounds
similar to that of reading ICL 7-track tape marks on IBM drives.
Essentially, the problem is that a tape mark is written as a block
that is shorter than the official minimum. In certain situations it
will be read as a noise record and not as a tape mark.

If my supposition is correct then the solution is to modify the device
driver so that the tape is always stopped before and after a tape mark
is written.  This will ensure that there is a longer gap before and
after the tape mark than there is when a tape mark is written on the
fly immediately after data.

The fudge of using 'mt' to write an extra tape mark works because it
happens when the tape is stopped, the gaps are longer, and the read
circuitry has more time to get itself sorted out.


Raphael Mankin				raph@planet.bt.co.uk