Description of rounding rules for PPR transfer agreements

Elliott, Robert Robert.Elliott at COMPAQ.com
Thu Dec 20 15:42:38 PST 2001


* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert" <Robert.Elliott at compaq.com>
*
The initiator has to be satisfied with whatever the target chooses to
return in PPR IN. This has to be a subset of the initiator's request,
and has to be a legal combination of fields.

QAS_REQ=1 and IU_REQ=0: this mode means the target participates in QAS
but does not generate QAS REQUEST messages itself.  This combination was
not allowed in SPI-3 but is allowed in SPI-4.  A target that doesn't
support it won't ask for it, so there's no SPI-3/4 interoperability
problem.

transfer period factor=8h and IU_REQ=0: This is an illegal combination.
The standard doesn't define every error case.  
a) If the target receives this combination in PPR OUT, it could do a
MESSAGE REJECT or return PPR IN with a valid subset of the values
(transfer period factor increased to 9h or higher).
b) If the initiator receives this combination in PPR IN, it is required
to reply with MESSAGE REJECT.  The target is then required to go to WDTR
IN and try to come to some agreement (falling back to SDTR then the
default agreement if every step fails).

IU_REQ=1 and transfer width exponent=0: This is an illegal combination.
The target can reply with IU_REQ=0 but cannot upgrade transfer width
exponent to 1.


---
Rob Elliott, Compaq Server Storage
Robert.Elliott at compaq.com



> -----Original Message-----
> From: Gerry.Houlder at seagate.com [mailto:Gerry.Houlder at seagate.com]
> Sent: Thursday, December 20, 2001 3:05 PM
> To: t10 at t10.org
> Subject: RE: Description of rounding rules for PPR transfer agreements
> 
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Gerry.Houlder at seagate.com
> *
> 
> I don't see wording in the standard (even rev 6) that expresses what I
> remember about this rule. I suspect the wording was removed at a much
> earlier revision or I might just be remembering a discussion 
> at a meeting
> and the result of the discussion (at that time) was never 
> written into the
> standard.
> 
> One discussion revolved around having a period of 8 or 9 
> (which requires DT
> data transfer) and having the DT_REQ bit cleared. This case looks
> "roundable" (i.e., just round the period value) so this was 
> not a problem.
> 
> Other cases were not so clear. What if QAS was set and IU_REQ 
> is not? What
> if period is 8 (320 MB paced transfer) and IU_REQ is not set? What if
> IU_REQ is set and width is set to 0? There may be more than one way to
> "round these down". There was talk of just reverting to the default
> agreement if these cases were encountered instead of 
> rounding. This seems
> to be an allowed rounding value for any case, but will 
> initiators depend on
> exact rounding behavior in the drives and could strange 
> things happen if
> different drive vendors do the rounding differently?
> 
> *
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
> 
*
* 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