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

Break Link (Re: headless chicken)




Howard, Rich,

As for LSS or Layer-1 signaling mechanism, I have the same perspective 
as that Rich has already stated in his following note and hence I will not 
discuss it again here.

At 17:43 00/06/30 -0700, Rich Taborek wrote:
http://grouper.ieee.org/groups/802/3/10G_study/email/msg02846.html

Instead, I would like to discuss the Howard's 'Break Link = Reset' here.
I think this Reset mechanism does the remote PHY reset by halves; 
it will not solve the remote PHY hung up if the remote PHY has been 
running into dead-lock before catching the byte-boundary in each lane.

At 11:43 00/06/30 -0700, Howard Frazier wrote:
http://grouper.ieee.org/groups/802/3/10G_study/email/msg02840.html
> Break Link = Reset.  You send Break Link when everything has gone south and
> your only recourse is to Reset and try again from square zero.  You send
> Break Link to let the remote end know that you have pushed the reset button,
> and the remote end should do the same.

My original proposing Break Link could be interpreted as Layer-2 Reset.
The Break Link transmission breaks the Duplex Link Up with valid MAC 
partner (Duplex Link Up) in the remote node.  This would be sensed by 
the remote MAC via STA as Duplex Link Down.  After 10 ms, stopping to 
send Break Link will recover the Duplex Link Up in the remote node 
and start the new Layer-2 communication.
http://grouper.ieee.org/groups/802/3/10G_study/email/msg02839.html

Layer 2 RESET
  Local node   -> Send Break Link for T ~= 10 mS
                  and reset the local Layer-2 port.

  Remote node     Duplex Link Down by receiving break link
                   -> reset the Layer-2 port
                  Duplex Link Up after 10 ms (by receiving not-BL)
                   -> accept data from MAC

I myself don't think we need Layer-1 Reset; 10GbE will have no 
Auto-Negotiation and hence have no possible hung-up state at least 
in Layer-1; PCS, PMA, and PMD.  If PHY has run into hung-up in Layer-1, 
what we should do is not pushing the reset button but replacing 
the PHY for repairing; it must be not compliant to the standard.  
I am afraid that allowing the Layer-1 reset introduces irresponsible 
implementation of PHY.

If you still want to have Layer-1 reset, you had better adopt much 
more primitive Break Layer-1 signaling, not depending on the Layer-1 
function such as byte Sync and deskew.  The most simple method would 
be stopping the output lightwave from the local transmitter.   However, 
to me, it is looking more like Auto-Negotiation; special start-up 
sequence for Layer-1 Link. 

Note that, in SONET, there is no in-band signaling to break frame Sync in 
the remote node.  I think this is obvious since in-band signaling is based 
on the frame sync; sending the Break Sync based on the Synchronization 
system itself sounds inconsistent to me.

Best Regards,
Osamu


-----------------------------------------
Osamu ISHIDA
NTT Network Innovation Laboratories
TEL +81-468-59-3263  FAX +81-468-55-1282