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

[EFM] Re: EFMWG: RE: MAC control Pause clarification




Frank,

I am not sure if the intention was to allow functional support for multiple 
queue clients such as VLAN tags.  At present, it is my understanding that 
802.3x does not distinguish pause based on VLAN tags, which some people are 
saying makes Flow Control in-operable.  As for using it on PON, I still do 
not believe that it should be above the PHY layer for a lot of reasons.

Thank you,
Roy Bynum

At 01:42 PM 6/7/01 -0400, FEffenberger@xxxxxxxxxxxxxxxxx wrote:

>Harry,
>Yes, the standard prohibits the transmission of unicast pause frames.
>It is quite clear on that.  However, since the reception of same is
>allowed, it appears as if the authors had future extensions in mind.
>
>I also got a possible suggestion for the reason why this might be,
>in that bridges must filter out pause frames, so that they don't
>propagate through the network.  This is easier to do when they all
>have the same (multicast) address.
>
>Also, use of pause is explicitly limited to point to point systems.
>(See note in section 31B.1.)  Hence use of Pause over PONs is
>prohibited de jure.
>
>I was hoping that the actual authors of the language could explain
>this themselves.  At the last meeting, some people were saying
>"Don't change the pause Opcode, get a new Opcode."  But given
>the existing standard, it looks like some future augmentation was
>contemplated when clause 31B was drafted.
>
>Regards,
>Frank E
>
>
>
>-----Original Message-----
>From: Harry Hvostov [mailto:HHvostov@xxxxxxxxxxxx]
>Sent: Thursday, June 07, 2001 1:13 PM
>To: 'FEffenberger@xxxxxxxxxxxxxxxxx'
>Subject: RE: MAC control Pause clarification
>
>
>Frank,
>
>Do you believe the standard precludes transmissions of the unicast pause
>frames? The frame format appears to be quite capable of supporting unicast
>DAs.
>
>Harry Hvostov
>R&D Director
>CTO Office
>
>Luminous Networks, Inc
>10460 Bubb Road
>Cupertino, CA 95014
>
>Tel: (408) 342-2512
>Fax: (408) 863-1107
>
>
>
>-----Original Message-----
>From: FEffenberger@xxxxxxxxxxxxxxxxx
>[mailto:FEffenberger@xxxxxxxxxxxxxxxxx]
>Sent: Wednesday, June 06, 2001 8:03 PM
>To: stds-802-3-efm@ieee.org
>Subject: MAC control Pause clarification
>
>
>
>Everybody,
>I got the impression from the discussion at the last meeting that the pause
>command was not defined for use with unicast MAC addresses.
>
>I checked in Annex 31B, and I found conflicting information:
>
>In section 31B.3.1, Transmit operation, it says in the first line item (a)
>that "The destinationParam ... is currently restricted to the value
>specified in 31B.1"
>That value is the globally assigned multicast address for pause.
>
>In section 31B.3.3, Receive operation, it says in the second paragraph that
>"Upon receipt of a valid MAC Control frame with ... the destination address
>indicating
>either: (1) the reserved multicast address specified in 31B.1 or (2) the
>unique physical
>address associated with this station,"
>
>So, it seems that the standard allows for the reception of unicast pause,
>but not
>the transmission of same.  Could any of the IEEE alumni explain why this is
>so?
>
>Thanks,
>Dr. Frank J. Effenberger
>Director, System Engineering
>Quantum Bridge Communications
>(978) 983-2532