[802.3_MAINT] Summary of maintenance adhoc call on MAC interface
1. Participants:
Wael Diab
Howard Frazier
Eric Lynskey
Jeff Mandin
Shimon Muller
Pete (UNH)
Pat Thaler
2. The link to the patent policy was mentioned.
3. Summary of issues discussed:
* Overview
Slides were presented describing the motivation and
specific changes to the MAC-to-MAC-Client interface made in 802.3as; as
well as a description of impact on MAC Control.
* Semantics of MA_UNITDATA.Request()
Pat Thaler has posted a summary of the views on this
(probably central) issue
* Clause 64
There was tentative consensus that the clause 64 state
diagrams can work with the understanding of MA_UNITDATA.Request()
(outlined by Pat) under which the transmission might be delayed. All
transmitted packets pass through the MPCP control multiplexer - and its
timed gaps between packets can assure that each time that
MA_UNITDATA.Request() is invoked there is never a packet in the pipe and
hence no delay.
* Clause 31
There was concern about figure 31B-1 (PAUSE operation
transmit state diagram) and whether it behaves properly under different
understandings of how MA_UNITDATA.Request operates.
* Correctness of figure 4-6
Various views were expressed regarding the interpretation
and correctness of the figure (4-6) that maps the MA_Unitdata.Request()
primitive to the Pascal TransmitFrame(). However there appeared to be
some basic consensus on how it SHOULD work ie. the process should stay
in GENERATE_TRANSMIT_FRAME state until all frame data has been
transmitted to the PHY.
4. Next Steps
There is as of yet no clear consensus on the two questions to be
addressed by the adhoc (ie. what fixes are necessary to MAC control?
Are any fixes needed within the MAC? ). Consequently we will take the
discussion to the reflector and have a second telecon on August 13 0800
PDT.
Thanks to all participants and best regards,
- Jeff