March 12,2010

Minutes of today's IEEE 1149.1 - Initialize Sub-Group Meeting

Attendees:
Roland Latvala
Francisco Russi
Adam Ley
John Braden
Carol Pyron

Minutes:

Today Carol provided the 'real world' BSDL example, and our meeting was dedicated to this discussion. Carol's example BSDL
included a 125 bit INIT_DATA register to support its Initialize requirements. Key point to note was that although this
device had BSR length of over 1000 bits, Init_Setup only required a 125 bit TDR for its INIT_DATA register. This was a very straight 
forward example, and leveraged similar constructs to what Ken's smaller example had emphasized. REGISTER_FIELDS, REGISTER_MNEMONICS, 
and REGISTER_ASSIGNMENTS. The TDR bits were grouped into logical sub-fields in the Register attributes section, each sub-field 
typically 5 bits (3 bits RX, and 2 bits TX Serdes channels.) These RX and TX fields were assigned mnemonics for the allowed values 
for any given Protocol (RX), and Drive Strength (TX).

Question came up about whether a dedicated 'comment field should be associated with each new attribute, or if the existing -- was sufficient. 
Special comments that a software tool could read in might be useful. Example (DDR term selection 2^n bit combinations of valid choices 
may reference a comment field anbout an app note in the Mnemonic pull down as opposed to providing hundreds of valid choices.

Question also came up on how to validate the INIT_DATA TDR bits relative to logical Port Mapping was correct, as well as how to validate these
TDR bit values themselves do what is intended. eg: TX 010 = 0.75x full swing. Since digital simulators don't care about levels, 
it would take Spice or other analog mixed mode simulations to validate the decodes. It is up to the Chip providers to ensure the 
BSDL is correctly coded and matches the design application notes and so on. This seems obvious enough but this does raise concerns 
about whether or not 3rd party tools could ever validate whether this information in the BSDL for the INIT_DATA register is correct 
or not.

Lastly it was again mentioned that since we are making good progress in the Initialize sub-group, we would like our work to be 
included in this next release of the 1149.1 Standard. It was generally agreed that industry would frown upon two successive 
releases 2010, 2011 of the standard if Initialze is not allowed to be included into this upcoming version, Using a normative annex may 
speed the integration process allowing us to work in parallel with the main body of the Standard as Ken mentioned last week.

Current Status:
Rules - informally done
BSDL Constructs - informally agreed upon
PDL - informally agreed upon as default side file format

Work still to be done:
Formalize Rules
Formailize BSDL constructs
Formailze PDL constructs
Incorporate into 1149.1 Std

Actions:
1. Carol to discuss release timeframe with CJ, and others.
2. Carol to finish up her BSDL example and distribute to the Initialize sub-group.

Next meeting date:
Same time next Friday March 19th.