From: Carl Barnhart
Sent: Friday, January 08, 2010 5:24 PM
Subject: 1149.1 INIT Meeting Notes, 8 Jan 2010
Attendees:
Carl Barnhart
C J Clark
Dave Dubberke
Ted Eaton
Herko Ehrenberg
Sivi Jaya
Roland Latvala
Adam Ley
Ken Parker
Carol Pyron
Francisco Russi
We first reviewed and made minor changes to a set of slides Carol
prepared to document our motivation. This is a start for some of the
descriptive text we will need. Carol has distributed the updated slides
separately.
Ken took an action item for himself to distribute our rules and the
slides to the IEEE 1532 Working Group, since he feels they will have a strong
interest in this issue and we have no input from the FPGA world at the moment.
(Done)
We then dove back into the rules (around 3.i and later).
- We discussed the rules for timing the INIT_RUN
instruction execution. There was a note proposing dropping the
criteria based on a system clock or on straight time. After
discussion, it was decided to keep these two specifications, but to add a
Recommendation to avoid the system clock requirement if possible, and that
the system clock could NOT be required after the completion of
INIT_RUN. This plugs an existing “hole” in 1149.1, which does not
now have a rule that operation of 1149.1 may not be dependent on a system
clock, only TCK. (CJ, take note.)
- We then had a lengthy and passionate discussion of
whether the results of INIT_RUN could or should persist through a
Test-Logic-Reset (TLR) state (Rule 3.k 3)). Ken fought for allowing
retention because multi-section boundary scan tests always start and stop
in TLR, and a lengthy INIT sequence could greatly extend test time ($).
Carol argued that her designers could not provide such persistence,
that TLR effectively reset the chip to mission mode and that mission mode
controls of I/Os, PLLs etc. would have to take over, destroying the
results of INIT_RUN.
NOTE: It would appear that there is a fundamental definitional
question here as to whether TLR puts the chip in mission mode or whether
TLR can be a simple waystation between test sections without asserting
mission mode. We may need the dot-1 WG to resolve this and perhaps
add words (rules?) governing the transitions between test sections
(require use of RTI instead of TLR for each test section?) and
test/mission mode. CJ? Adam? Ken, what about a rule that
says that if you want the INIT_RUN state to persist between test sections,
do not enter TLR and use RTI as the “parking” state instead?
A couple changes were made to the rules, and the new rules
distributed by Carol. Next week we will resume the discussion of TLR and
transition between test sections and mission mode.
Carl
"The secret is to ask the right question. Then the
answer takes care of itself."
Carl F. Barnhart
Senior DFT Architect / SiliconAid Solutions
Office: 512-535-1543 / Cell: 512-608-3280
Web: http://www.siliconaid.com