On Fri, 15 Jun 2001, J. Scott Berg wrote:
> say you are writing data, then you do a bsf. You must send a weof to
> the tape drive before executing the bsf, otherwise you have two
> overlapping streams
Why should the driver do this?
The scenario you describe should never happen. If you are writing data,
and a bsf command shows up on the queue, the first thing that should
happen is that the drive should _stop_ writing data when it's finished,
and write an EOF mark. The driver should never acknowledge a new command
while a write is going on; it should block and hold the command queue
until the write returns.
This problem would indicate that the write capabilities of the driver are
non-atomic, and can be interrupted before completing. You should be
unable to write data without writing EOF marks, and should be unable to
interrupt the write at the device level (events such as power loss
notwithstanding). Write interrupts should come from the application level
(stop sending data to the drive), and should, by the operation of the
driver, result in whatever data had been sent ending up on the tape, with
an EOF mark at the end of it.
-- To unsubscribe, send mail to: email@example.com --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2b29 : Fri Jun 15 2001 - 09:54:40 EDT