Re: Verilog++ 13th Committee Meeting - 11/26/01

Subject: Re: Verilog++ 13th Committee Meeting - 11/26/01
From: Peter Flake (flake@co-design.com)
Date: Mon Dec 10 2001 - 08:06:52 PST

Page 18 - Attribute syntax still required from PeterF. Peter is targeting
this week to complete this.
Action: Peter to complete Attribute Syntax.

Attributes in modports

<modport_port> ::= <attribute_instance> <modport_port>
| [<direction>] [<port_type>] <identifier>
                        | <identifier> '.' <identifier>
                        | <import_export> 'task' <named_task_proto>
                        | <import_export> 'function' <named_fn_proto>
| <import_export> <identifier> {',' <identifier>}



Page 30 - make sure SystemVerilog is fully substituted for Verilog-ACE.

Page 34 - Uncomfortable with putting in non-simulation information - should
be attributes. Example is the state encoding on this page. David S, Cliff
and others disagree. Alternative proposal is to have pre-defined
attributes. Cliff taking action to code up some examples. Peter still to
look at this item given everyone's comments. Everyone need to give this
some more thought for the next meeting.
Action: Cliff to come up with examples as discussed.

I am putting of the state encoding study until we agree whether state machines are to be included.

Alex Page 37 - minor modification required with example to be moved.

Page 40 - Issues related to scheduling semantics discussion. Compatibility
issues with Verilog2K1. Also issues getting this through IEEE 1364 committee.
Action: Peter Consider this further given the need to make this straight
forward for the IEEE1364 committee.

We can probably limit the requirement to ensuring that execution of a single statement (not a block) containing no user task or function call is uninterrupted.  This allows atomic test-and-set using assignment operators in an if statement.

Page 42 Simon - Fork strobe is an error from legacy SUPERLOG, and needs to
be removed. Stu has details.

Page 43 Peter F - Explanation still required.

The 'while' loop contains a delay of two clock cycles, from the 'repeat' statement, and this determines the pipeline throughput.
Each iteration spawns a process which lasts six clock cycles, the latency of the pipeline.  The variable 'processes' keeps a count of the number of currently active processes.  The pipeline flush is not complete until this count has fallen to zero.

Page 47 Simon - $root issue, as previously discussed. Stefen still has
issues with the code complexity that results from $root. Simon / Peter F
plus others think it is OK. Stefen willing to bow to majority to close
issue. Remove red comment from documentation. Agreement that there are
reasons to have $root but it should be constrained. General agreement to
have separate meeting to deal with it specifically.

Page 47/48 $root example require to show abstract model
Action: Simon to write an example.

Page 50 Extern module declaration needed?
Action: Kevin to provide proposal to group

Page 54 Peter needs to provide name space clarification.
Action: Peter to send email to Stu on this.

There is one name space hierarchy in SUPERLOG ESS.   A type name may be not be the same as an instance name.
Types include modules, interfaces, and data types.  Instances include tasks, functions, procedures, variables, constants and labels as well as module and interface instances.

Pre-defined (built-in) names begin with $.  For example $root is the name of the activation script.

Names are initially global.  A new scope is defined by:
a module or interface
a task or function
a sequential or parallel block
a structure or union

Tasks and function definitions cannot be nested within themselves, but can be defined in modules or interfaces.  Again the declaration in the closest enclosing scope is matched.

A data type can be defined in the global scope or in the first three local scopes using a typedef, and this name cannot be used for anything else within the same scope.

Page 57 Co-Design needs to add wires to example.
Action: Peter to provide.

Sorry, I have not had time to do this yet.

Regards,

Peter.

This archive was generated by hypermail 2b28 : Mon Dec 10 2001 - 08:03:13 PST