Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: Remote Faults in RS




Hello Bob,
   Thank you for your input and advice.  I'd like to better understand the
intricacies of the 802.3 standards as this would be greatly beneficial in
ensuring a robust design.  You're correct in assuming our implementing of
the WIS and I'd like to get some clarification regarding the do's and don'ts
of Sequence ordered set deletion for rate control methods during WAN mode.

1)  If I have a stream of Sequence Ordered Sets, which of them can I delete,
assuming that I've counted 3 Idles/Ordered sets towards the minimum IPG?

For instance,      \I   \I   \I   \Q   \Q   \Q   \Q   \Q   \Q   \Q   \Q   \Q
\Q        
	       	       \I   \I   \I   \Q   \Q   \Q   \Q   \Q   \Q   \Q   \Q
\Q   \Q        
                         \I   \I   \I   \Q   \Q   \Q   \Q   \Q   \Q   \Q
\Q   \Q   \Q        
                         \I   \I   \I   \Q   \Q   \Q   \Q   \Q   \Q   \Q
\Q   \Q   \Q        

In the case above, I have 10 consecutive columns of Sequence ordered sets,
all of the same kind.  Which columns (#1-10) can I delete?  
   The 64b/66b PCS described in Clause 49, states in section 49.2.4.10 that
"deletion shall only occur when two consecutive sequence ordered sets have
been received and shall delete only one of the two."  Does that mean 50% of
the \Qs?  or does this mean that I can pass on the 1st \Q and delete the
remaining 9 \Qs that follow?  
 
2)  Referring to the RX Link Fault Signaling State Machine in figure 46-15,
it appears that in order to transition to the FAULT state it is required to
detect a minimum of 4 fault_sequences of the same type without not detecting
a fault_sequence over any interval of 128 consecutive columns.  If this is
correct, if a local fault is transmitted as a quick burst (very few  \Q's)
and is presented to this RX state machine after having had some of its
ordered sets deleted for rate control, is it detrimental if the link fault
is not detected due to the brevity of the failed link?  or is it the
obligation of the TX MAC/PCS to ensure that at least the minimum of 4
Sequence ordered sets are received by the RX RS?  Should the PCS consider
not deleting the first 4 columns of \Q's to guarantee notification of the
link fault to the RX RS?  
  
I'd appreciate any insight regarding these questions.

Thanks again,
Jennifer Sanati
jennifer.sanati@intel.com   


> -----Original Message-----
> From:	Grow, Bob 
> Sent:	Tuesday, June 05, 2001 5:55 PM
> To:	Sanati, Jennifer
> Subject:	RE: Remote Faults in RS
> 
> Jennifer:
> 
> The response posted by Sanjeev agrees with my understanding of correct
> draft 3.0 behavior.  I don't recall any changes from the working group
> ballot that would alter this in D3.1 either.  
> 
> From your message, it appears you might be implementing the WIS.  I hope
> you have access to the draft for definitive reference, the reflector often
> takes days to arrive at a correct answer to a question, and obviously the
> archive can't be relied on for current information.  The change to
> continuous Sequence ordered sets was in D2.1 (Feb 9, 2001).  Please let me
> know if I can be of any help in the future.
> 
> Bob Grow
> Editor, 802.3ae Clause 46
> 
> -----Original Message-----
> From: Sanati, Jennifer [mailto:jennifer.sanati@intel.com]
> Sent: Tuesday, June 05, 2001 3:15 PM
> To: 'stds-802-3-hssg@ieee.org'
> Subject: Remote Faults in RS
> 
> 
> 
> Hello everyone,
>    I would like some clarification regarding the format of remote faults
> that are generated during transmission in the reconciliation sublayer and
> presented to the 64b/66b 10GBase-R PCS.
> 
> Referring to an excerpt from a previous IEEE reflector discussion by
> Stephen
> Finch on Fri, 29 Dec 2000, he states...
> --------------------------------------------------------------------------
> --
> ---------------
> 3.  The RS layer is where the Local Fault Pulse Ordered Set is
>     processed.  The RS layer is the only place that a Remote
>     Fault Pulse Ordered Set can be generated.  If an RS receives
>     a Local Fault Pulse Ordered Set it must stop sending packets
>     and begin sending alternating columns of Idles and Remote
>     Fault Pulse Ordered Sets.  If an RS receives a Remote Fault
>     Pulse Ordered Set, it must stop sending packets and send
>     only Idles.
> 
> --------------------------------------------------------------------------
> --
> ---------------
> 
> Is the underlined statement still valid?  If a LF is received at the RS,
> is
> it then translated to a RF during TX and flagged by sending a stream of
> altermating columns of Idles and Sequence Ordered Sets?
> 
> such as,     /I   /Q   /I   /Q   /I
>                  /I   /Q   /I   /Q   /I
>            	     /I   /Q   /I   /Q   /I
>                  /I   /Q   /I   /Q   /I
> 
> If this is a correct assumption, for WIS rate control methods, is it
> correct
> that after the minimum IPG, it is only valid to delete Idles and pass on
> the
> Sequence Ordered Sets? 
> 
>    The previous question is pertinent only if the above data stream can
> occur as a possible input to the TX 64b/66b PCS.  If, instead, a RF
> triggers
> a "relatively long" string of consecutive Sequence Ordered Sets, 
> 
> such as,    /Q   /Q  . . . .   /Q   /Q
>       	    /Q   /Q  . . . .   /Q   /Q
>                 /Q   /Q  . . . .   /Q   /Q
>                 /Q   /Q  . . . .   /Q   /Q
> 
> then WIS rate control would be much easier to design for.
> 
> I would appreciate any input regarding these detailed matters.
> 
> Thanks,
> Jennifer Sanati
> jennifer.sanati@intel.com
>