An SBC-2 question regarding persistent reservation.

Tom Coughlan coughlan at missioncriticallinux.com
Mon Aug 21 14:06:58 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* Tom Coughlan <coughlan at missioncriticallinux.com>
*
Gene,

I think I understood the question.  In my view, the two requirements are
different.  The first requirement:

- only applies to commands that do not write the media, and,
- says you must check for reservation conflicts before the command
starts to execute.

The second requirement:

- only applies to commands that write the media, and,
- says you must check for reservation conflicts before the command
starts to modify the media (or cache)

The only unstated assumption is that there may be a logical unit
implementation that may benefit from being allowed to start executing a
write command, do some non-media-changing operations, then maybe even
leave and re-enter the current task state, then check for reservation
conflicts, then modify the media if allowed.  

Do you agree that such an implementation would be illegal if the second
requirement is removed?

Do you agree that making this implementation illegal this is
undesirable?

Tom

From: Gene.Milligan at seagate.com

> I am not certain that you got the thrust of the question. I questioned the
> second requirement (commands that write the media) since it is satisfied by
> the first requirement (commands that do not write the media). I questioned
> it since having two requirements infers that they should be different but a
> difference is not clear and certainly is not explicit.
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list