SAS - OOB detection

[Larrie Carr]

* From the T10 Reflector (t10 at t10.org), posted by:
* Larrie Carr <Larrie_Carr at pmc-sierra.com>
*


> -----Original Message-----
> From: Elliott, Robert (Server Storage) [mailto:Elliott at hp.com]
> Sent: Saturday, May 24, 2003 8:24 AM
> To: t10 at t10.org
> Subject: RE: SAS - OOB detection 
> 
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Elliott, Robert (Server Storage)" <Elliott at hp.com>
> *
> > * From the T10 Reflector (t10 at t10.org), posted by:
> > * Larrie Carr <Larrie_Carr at pmc-sierra.com>
> > 
> > In Figure 58 - OOB Signal Detection of the SAS specification 
> > (originally Figure 47), there was a change between sas-r3f 
> > and sas-r3g of when an OOB event was detected.  The original 
> > version in sas-r3f agrees with my interpretation of the SATA 
> > specification while the new version does not.
> > 
> > SATA 1.0a states in section 6.7.4
> > 
> > "OOB signals are observed by detecting the temporal spacing 
> > between adjacent bursts of activity, on the differential pair."
> > 
> > The "temporal spacing" is the idle period and the "bursts of 
> > activity" is the ALIGN burst.  Due to this statement and the 
> > min/max specification for the idle periods, I would expect 
> > the OOB receiver to process the idle periods only.  Also, the 
> > requirement on an illegal idle time invalidating the event 
> > detection implies the ALIGN burst time is not important.
> > 
> > With this intent, the logical place to declare an OOB event 
> > (such as COMSAS) is at the end of a OOB idle period just 
> > after the ALIGN burst is detected.
> > 
> > With the current definition, it appears you may have to 
> > measure the ALIGN burst (which we don't have any 
> > specifications for) and then declare the OOB event when you 
> > just get the beginning of the next OOB idle period.  However, 
> > this declaration does not take into account the length of the 
> > OOB idle period (suggesting that any idle is valid).
> > 
> > A strange side effect of the sas-r3g specification is that an 
> > OOB event detection is trigger at the beginning of the OOB 
> > idle period which is the OOB event negation.
> > 
> > Can someone please add some color on this change?
> 
> The same logic should work for both SATA and SAS.  SAS is
> just pickier about how it specifies things - including a
> token minimum burst time of 100 ns and saying that the 
> repeating pattern is "idle/burst" rather than "burst/idle".
> 
> The key is to measure 4 idles of the proper length (and
> not be fooled if the first or last idle is too long).
> You can't do that without seeing a burst after the 4th
> idle.

True, you can't see the idle without seeing the burst - but you don't need to see the entire burst.  As the diagram shows, the detection point is at the end of the burst.  

I agree with the your comments.  However, the text below from the specification..

"A receiver shall detect an OOB signal after receiving four consecutive idle time/burst time pairs (see figure 58)."

..interacts with the diagram to indicate that the 4'th burst must be respected (because it's the burst of the 4th idle/burst time pairs).

> 
> > 
> > Thanks,
> >  
> > Larrie Carr
> > 
> > Principal Engineer
> > PMC-Sierra, Inc.
> > 8555 Baxter Place, Burnaby, BC, Canada  V5A 4V7
> 
> 
> --
> Rob Elliott, elliott at hp.com
> Hewlett-Packard Industry Standard Server Storage Advanced Technology
> https://ecardfile.com/id/RobElliott
> 
> 
> *
> * 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