SCSI Tapes and Fibre Ch

Jim McGrath jmcgrath at QNTM.COM
Fri Jul 19 19:27:17 PDT 1996


* From the SCSI Reflector, posted by:
* "Jim McGrath" <jmcgrath at qntm.com>
*
        Reply to:   RE>>SCSI Tapes and Fibre Channel


I think some of the tape proposal is a bit too restrictive.  For instance,
I do not believe it is ever necessary to prevent the system from
sending down the next command's CDB information before status
is returned for the previous command.  Originally this discussion 
revolved around writes where it would be possible for a device to
write data for the second command to the media before the system
received status for the first command.  So if power was lost (or
bus traffic lost), then the system may end up being confused as to
the exact state of the media for commands 1 and 2, and in what order
the media was actually altered.

But note that this problem only occurs if the write data for command
2 is sent to the device before status for command 1 has been returned.
There is no reason I can see not to send the COMMAND information for
command 2 before status is returned.  This less restrictive usage would
reduce some overhead and potentially allow the device to use command
information to optimize its operations.

Jim






More information about the T10 mailing list