IEEE P1532 Meeting - Jan 18-19, 1999 - AMD

Back to Minutes page
Back to IEEE P1532 home page

Attendees

The following individuals attended the meeting 

   Neil Jacobson          Kurt Guntheroth
   Howard Tang            Dave Thompson
   Dave Bonnett           Steve Klinger
   Ted Eaton              Alan Herrmann
   Brad Ishihara          Yadvindra Dhami
   Ken Parker             Tapan J. Chakraborty
   Arthur Khu             Mark Moyer
   Carmy Yellin

Minutes

Monday 1/18/1999

Motion: Minutes from previous meeting

Ken Parker made a motion seconded by Kurt Guntheroth that the minutes from the previous meeting be approved unchanged.

The vote carried 6-0-0.

Action Item: Neil Jacobson and Ken Parker

Neil will provide Ken Parker with an updated working group member list and Ken will update page 2 of the draft.

Draft Review

Neil reiterated that a goal of this meeting was that we would be prepared to go to ballot on Tuesday evening.

Section 2

Neil solicited any additions and/or changes.

Neil noted the missing closing bracket on 2.2.3

Dave Thompson suggested that we add a description for all 1149.1 instructions referenced in the standard.

There were no other additions or changes suggested.

Action Item: Ken Parker

Ken will add a description of all 1149.1 instructions referenced in the standard.

Section 3:

The discussion centered on the idea of allowing 1149.1 instructions while in the ISC_ACCESSED state. This is a key to the idea of spinning plates that will be re-introduced on Tuesday.

It was decided that:

1149.1 Instructions will be allowed while in ISC_ACCESSED mode provided:

  1. The point of insertion is not during an ISC process(i.e. following the completion of an ISC_READ, ISC_ERASE, or ISC_PROGRAM command sequence)
  2. The ISC_DISCHARGE instruction must, if available, follow ISC_PROGRAM uninterrupted.
This was not officially motioned in and will be re-visited on Tuesday.

Action Item: Alan Herrmann

Alan will investigate any implementation limitations incurred by allowing 1149.1 instructions during ISC_ACCESSED mode.

Five other points of discussion were enumerated and discussed.

  1. Initial I/O state of an unprogrammed devices at power up - This was agreed to be tri-state
  2. Removal of the HIGHZ option for I/O during ISC_ACCESSED mode. After much discussion it was agreed that the HIGHZ option serves a specific purpose and that it will be left in. It was also decided that wording would be added to recommend clamp as the preferred I/O behavior during ISC_ACCESSED mode.
  3. Does the standard prohibit core integration - This was deemed solved by item 4 below.
  4. Does the standard require HIGHZ or CLAMP behavior on non-programmable I/O? It was decided that only I/O whose behavior may be directly programmed through ISC must exhibit HIGHZ or CLAMP during ISC_ACCESSED mode.
  5. Are FPGAs considered ISC devices - This was tabled for Tuesday

Action Item: Ken Parker

Ken will make all references to "ALL" I/O refer to all I/O whose behavior is determined through ISC control.

Action Item: Ken Parker

Page 31 section 5.1.1.C Note 4 will be moved to the architecture section and will be altered to require consistent I/O behavior between RUNBIST, INTEST, and ISC_ACCESSED mode

Action Item: Ken Parker

Ken will alter Figure 3.2 to include the "TEST" mode in order to illustrate the result 1149.1 instructions during ISC_ACCESSED mode.

Section 4

Action Item: Ken Parker

Pg. 28 (H) The Instruction register capture of the ISC_DONE bit will be guaranteed correct at power up as well as following an ISC_DISABLE instruction. An additional statement will be added to indicate that a "1" value indicates ISC_DONE is set and a "0" indicates ISC_DONE reset.

Action Item: Ken Parker

Section 4.2C will be removed and the note following will be moved to an appropriate section.

Action Item: Ken Parker

Section 4.2a will be reworded to indicate conformance to all 1149.1 mandatory provisions.

Section 5

The group discussed the recent submission of a standardized security methodology provided by Ken Parker. The group came to a tentative agreement that if:

  1. Wording was added the specifically indicated the minimization of the option to a single bit.
  2. The key field is allowed to not exist.
  3. An optional ISC_PSData register be accessed by the ISC_PROGRAM_SECURITY

Then this section will be adopted as the only possible implementation of 1532 security.

Action Item(s): Ken Parker

  1. Ken will reword Section 5.1.8 to incorporate the three conditions listed above.
  2. The instruction section will be updated to refer to the three security signals ISC_Disable_Read, ISC_Disable_Erase, and ISC_Disable_Program where appropriate.

Section 6.2.4

Action Item: Ken Parker

The third paragraph (lines 6 and 7) and 6.2.4.2 second sentence will be relocated to the general Instruction section and updated to indicate the possibility of 1149.1 instructions altering the I/O states present at the time of the ISC_ENABLE instruction.

Action Item: Ken Parker

Section 6.2.4.d will be altered to indicate the ISC_READ instruction should return the "blank" data state if the device security bit is set.

Tuesday 1/19/1999

Action Items: Ken Parker

  1. Add wording to indicate that the amnesia code of the device must be documented or that USERCODE of a blank device should be documented.
  2. Ken will modify figure 3.2 to indicate the transitions to and from test mode.

Motion:

All transitions to .1 instructions in the new state diagram are defined. Any disallowed transitions will be documented.

First: Ted Eaton
Second: Ken Parker
Carried: 5-0-2

Motion:

That aliased instructions be used to document allowable .1 instructions in ISC modes.

First: Brad Ishihara
Second: Kurt Guntheroth
Denied: 0-7-0

Action Item: BSDL Extension Tiger Team

A tiger team was formed to define the BSDL extension(s) required to describe the behavior of an ISC device. The definition process has a deadline of April 1999.

Tiger Team Members: Ken Parker, Dave Bonnett, and Ted Eaton

Action Item: Kurt Guntheroth

Kurt will develop a proposal for spinning plates.

Annex A: Programming Flow

Action Item(s): Ken Parker

  1. Add the ISC_DISCHARGE instruction to one of the examples.
  2. Split examples into program, verify, and security sections
  3. Add an indication on each DRScan to indicate if the data is real data or just filler.
  4. Device 3 of Section 8.5 needs to have an explicit write operation to set the security bit following the verify section.

Ken will add a note that the state of the ISC_DONE bit does not influence programming.

It was also agreed that a note would be added to state that security will take effect no later than the execution the ISC_DISABLE instruction.

Action Item: Neil Jacobson

Neil will begin to solicit a ballot group.

Motion:

ISC_READ optional for all devices

First: Kurt Guntheroth
Second: Brad Ishihara
Carried 6-1-0

The next meeting was scheduled for March 8th and 9th at Xilinx.

Back to Minutes page
Back to IEEE P1532 home page