Phone Conference P1450

1. The inclusion of ScanCellGroups into dot1 -

        ScanCellGroups (DOMAIN_NAME) {
                ( cellgroupname = cellref_expr; )*
                ( cellgroupname = cellref_expr {
                        ( Base < Hex | Dec > WAVEFORM_CHARACTER_LIST ; )
                        ( Alignment <MSB | LSB> ; )
                        ( DataBitCount DECIMAL_INTEGER; )
                } ) // end cellgroupname
        } // end ScanCellGroups

2. Name spaces for scan-chains, scan-cells, scan-groups

3. Relation (if any) between scan_cell_expr (as defined in 1450.1, clause 6.9) and CoreInstance as defined in CTL:

        ( CoreInstance core_type_name {
                ((core_instance_name)+;)
                (core_instance_name { PatternBurst (burst_name)+;})*
        })*

4. What is the problem (if any) with the X statement for use in fail feedback?

5. Multiple Inherit of the environment.

6. NameMaps has some problem with regard to Inheritance.... I

do not remember the details.

7. Allow multiple pattern Bursts at the top level in PatternExec with

PatList, ParallelPat and PatSet.

8. Naming of scan cells list of 1450.1

9. I believe you have already fixed the Variable Name and Group Names

to be in the same name space.

10. LockStep cannot be done if the patterns that use

the protocols are mismatched.

Opening question

Bruce had a question about pg 33, and the ability to specify a WFC

multiple times. Some tools will generate the SAME value multiple

times. Need to clarify that a single WFC value, but same value may be

specified multiple times...

Scsncell groups

This is a new proposed construct. Currently have defined

scanstructures and hierarchical scanchains, that contain scancells.

Purpose of 'groups' is to collect scancells into groups that may not

follow chains - i.e. functional grouping, not scan-chain-order

dependent.

One issue: since the block has a domain name it would appear to follow

domain-name conventions.

recommendation: if domain name present need to reference

scancellgroups by this name in patternburst, just like scanstructures.

some concern about whether this resolution process is appropriate. CTL

considers this more of a "design" issue, and have a perspective that

the design is entirely present even if the test is partitioned.

Since this information is design-oriented, the scoping issues are

netlist-oriented rather than test-pattern-oriented. Furthermore there

is not a need (at this time) to reference these groups from the

Patterns. Perhaps a different scoping/resolution/specification

methodology should be used here.

It has previously been established that scancells are in a single

global namespace -- all scancells from all scanstructures are in the

same namespace. This is necessary to allow multiple chain

organizations (where one cell is a member of several different chains)

to be supported without constraining the use of scanstructures.

There was some discussion about whether scanchain names are also in

this space. CTL expects that these chainnamess are in the same space,

to allow specification of a chain-name or a cell-name as part of the

ScanCell list and avoid ambiguities of whether a chain or a cell was

referenced (since if they are in the same space they need to all be

unique).

Greg to [AI1] to work table 3 of the document.

Proposed as part of the group expression, a cellref_expr for

scancellgroups to be general list with '+' and '-' operators

supported.

Some discussion about attribute behavior, and discussion about

whether the hierarchical constructs (particularly on scanchains in

dot1 at this time) are part of dot1 or dot6.