Minutes of Weekly Meeting, 2007-10-19

Minutes from 10/29/2007 SJTAG ATCA Discussion Group Meeting

Call to order at 8:10am Eastern

Participants:

Brad Van Treuren,
Al Holliday,
Patrick Au,
Heiko Ehrenberg,
Peter Horwood,
Adam Ley

Minutes from 10/12/07 approved

Brad reviewed the slide set "ATCA Meeting Goals.pdf":

  • Ian McIntosh's comments via email:
    • I don't disagree on omitting JTAG where there is no device support - it's the reasoning for there being little or no JTAG-able devices in the first place that bothers me.
    • Within Selex, we have effectively mandated that all design with anything more than "trivial" digital content must provide means for effective JTAG testing of that digital content. Consequently, even our Power Supply Modules (obviously predominantly analogue) include JTAG as they feature microcontrollers and CPLDs to implement, e.g., control and protection features. What this means is that, while it may be possible for me to create a design that doesn't use JTAG, that doesn't mean that I should.
    • I think I can understand why, say, a particular Zone 3 AMC may not have anything worth JTAGing, but I struggle to see why a blade, as a whole, should not have JTAG support. After all, as I understand it, the fabric interconnects across the backplane are SerDes, which implies that each blade will have, e.g., FPGAs to implement those links, memory to buffer data, etc. and that implies "significant" digital content to me.
  • interoperability:
    • use only the 4 JTAG signals for test bus management, no other protocols
  • Access:
    • provide access to system JTAG infrastructure (single point) without the need to remove any blades or without the need to use extender cards; should be accessible via simple connector accessible when all blades are plugged in
    • do we want front panel access to blade level scan chain(s)?
      • probably not required and may be opposed by blade vendors (security, real estate) [Al mentioned that face plate real estate is already allocated for mission mode connectors]
    • allow multiple sources to control the JTAG infrastructure, e.g. one or the other JSM or an external controller, not just one exclusive access point; Peter had to leave the call at this time
  • Operation:
    • need to emphasize that carrier blades are really subsystem, requiring their local JTAG infrastructure to be immune to changes in the AMC blade configuration
    • need to be able to hot-swap blades that are "off-line", needs to be managed by system software, prohibit blades to be removed (put in hot-swap mode) that are currently being tested; slide 12, bullet 2 is a duplicate of bullet 3 on slide 11; slide 12, bullet 3 is a duplicate of bullet 1 on slide 12;
  • Patrick: is there a way we can prioritize these goals?

Adjourn at 9:29am Eastern

Next meeting 10/29/2007 at 8am EDT