Proposal to add a new modifier to SCSI Write commands.
coughlan at star.zko.dec.com
Mon Dec 2 10:55:56 PST 1996
* From the SCSI Reflector (scsi at symbios.com), posted by:
* coughlan at star.zko.dec.com (Tom)
In response to several recent posts on this topic:
If we (as representative OS developers) had reasonable certainty that Read
Long and Write Long would remain in the Standard, and that they would be
widely implemented, then we would continue to use them, and even invest in
more sophisticated algorithms for their use, as suggested by David Cohen.
As it stands, however, we have seen indications that the drive vendors
would like to obsolete the Read Long and Write Long commands, and
indications that implementation of the Long commands may be becoming less
wide-spread. (My impression is that there are few RAID controllers, for
example, that implement the commands on their host interface.)
In this situation I have little interest in tailoring our software to deal
with the generality of the Long commands, and the need to dynamically
contend with per-drive variations in behavior. Maybe this debate will
result in a more certain future for the Long commands. If not, the forced
error proposal offers an alternative that is easier to implement and easier
to use, and so more likely to be implemented.
So, lets hear more from the drive vendors. Has the response to this topic
convinced you that the Long commands are here to stay?
Gene described a desire to be able to detect when a malicious utility has
used Write Long to create errors and make a drive look bad. Would a new
ASCQ that says "the last write was a Write Long" provide what is needed?
(This would be a qualifier for for ASC = "Unrecovered Read Error".)
With regard to the patent issue, I've obtained reasonable assurance from
within Digital that the proposal as stated does not infringe. It will take
more effort to determine Digital's willingness to licensing the patent. I
was hoping everyone would agree that any advantage obtained by altering the
proposal would be outweighed by the bureaucratic issues related to patent
infringement. (Ed already mentioned too much time spent with patent
lawyers). I can see that I'm not going to be that lucky, so I will attempt
to determine Digital's official position with regard to licensing this
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com
More information about the T10