Selection with ATN in IU mode question

Elliott, Robert (Hou) Robert.Elliott at
Tue Sep 19 16:09:24 PDT 2000

* From the T10 Reflector (t10 at, posted by:
* "Elliott, Robert (Hou)" <Robert.Elliott at>
> * From the T10 Reflector (t10 at, posted by:
> * Ken Erickson <Ken.Erickson at>
> *
> If the initiator is hot-swapped with non-IU capable initiator 
> and the target attempted to respond with PPR message after
> selection w/ ATN in IU should not prevent the non-IU from 
> issuing a Message Reject message and allowing the target to 
> follow with a WDTR or SDTR message.  If the initiator fails
> the WDTR message the spec states to fall back to issuing a 
> SDTR message.
> What's the difference with starting with a PPR then WDTR?

PPR initiation from targets is banned, preventing this.

> Prohibiting targets from issuing a PPR message to "fall back" 
> to non-IU mode because it doesn't control domain validation 
> seem to me to be a moot point because we are falling back to 
> the lowest point of operation: narrow, slow, asynchronous.

Some expanders snoop negotiation messages.  If you managed
to use PPR to negotiate async narrow between two devices, a
snooping Ultra 2 expander won't know that async narrow was
selected.  This expander may trigger false errors or not 
forward data correctly.  That's why there's another rule 
stating that initiators shall only use PPR if PPR is necessary,
relying on WDTR/SDTR if no protocol options are needed.

> It sure would be nice if section was re-worded to
> remove the ambiguous stuff.  There is to many ways to 
> intrepret this section.

I hope to have a proposal rewriting the negotiation message 
sections (16.x) for the Halloween Parallel SCSI meeting.  
This rule from section may make it into the
rewrite; there are a few message rules scattered throughout
the document I'd like to collect into a new combined model

> Ken Erickson
> Quantum Corp.

PC: Robert.Elliott at
UNIX: relliott at

* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at

More information about the T10 mailing list