SV-ACMinutes2010_09_14 < P1800

P1800 Web>SystemVerilogAssertionCommittee>SVACMeetingMinutes>SV-ACMinutes2010_09_14 (2010-09-16, ErikSeligman) (raw view)EditAttach
Minutes from SV-AV Committee Meeting

Date: 2010-09-14

Time: 16:00 UTC (9:00 PDT)

Duration: 1.5 hours

Dial-in information:

--------------------

Meeting ID:            38198

Phone Number(s):

1-888-813-5316         Toll Free within North America

Agenda:

-------

- Reminder of IEEE patent policy.

See: http://standards.ieee.org/board/pat/pat-slideset.ppt

- Minutes approval

- F2F meeting (discussion)

- Email ballot results

2452: No vacuity information about synchronous aborts

- New issues

3206: Deferred assertions are sensitive to glitches

- Requesting authorization to work on additional enhancements

- Enhancement progress update

2476: Need clarification about system functions $onehot, etc

Checker usability enhancements

- Issue resolution/discussion

1853: BNF for calls to $rose and other sample value system functions.

2904: Clarify when disable iff condition must occur relative to starting

and ending of an attempt

3134: sequence and property range parameters are erroneously defined

3135: Verbal explanation of nexttime and always is misleading for

multiple clocks

1678: Clarify that rewriting algorithm doesn't replace name resolution

2571: confusing assertion clock inference rule

2386: Rename 16.9 to "Local variables"?

3117: make it clear that rewriting algorithm (F.4.1) applies to checker

and let

- Opens

Attendance Record:

------------------

         Legend:

                 x = attended

                 - = missed

                 r = represented

                 . = not yet a member

                 v = valid voter (2 out of last 3 or 3/4 overall)

                 n = not a valid voter

                 t = chair eligible to vote only to make or break a tie

Attendance re-initialized on 2010-07-06:

  v[-xxxxx--xxx] Laurence Bisht (Intel)

  v[xxxxxxxxxx-] Eduard Cerny (Synopsys)

  v[xxxx-xxxxxx] Ben Cohen

  v[-x--xxxxxxx] Surrendra Dudani (Synopsys)

  n[xx---x-xxxx] Dana Fisman (Synopsys)

  v[-x-xxxxxxxx] John Havlicek (Freescale)

  v[xxxxxxxxxxx] Tapan Kapoor (Cadence)

  t[xxxxxxxxxxx] Dmitry Korchemny (Intel ¿ Chair)

  v[xx-xxxxxxxx] Scott Little (Freescale)

  v[xxx-xxxxxxx] Manisha Kulshrestha (Mentor Graphics)

  v[xxxxxxxxxxx] Anupam Prabhakar (Mentor Graphics)

  v[xxxxxxx-xxx] Erik Seligman (Intel)

  v[xx-xxxxxxx.] Samik Sengupta (Synopsys)

  v[xxxxxxx-xxx] Tom Thatcher (Oracle ¿ Co-Chair)

    |- attendance on 2010-09-14

  |--- voting eligibility on 2010-09-14

Minutes:

--------

- Reminder of IEEE patent policy.

See: http://standards.ieee.org/board/pat/pat-slideset.ppt

Participants were reminded of the policy.

- Minutes approval

Erik: Move to approve minutes

Scott: Second

     Vote results: 11y, 0n, 0a

- Email ballot results

2452: No vacuity information about synchronous aborts

Dmitry: 2452 passed with 11y, 0n, 0a

     Note: added text should be in blue.

     Dana will update proposal to make new text blue, and Dmitry will

     update status.

Dmitry:    Even though the proposal passed, there were some questions raised

     about it.

- New issues

3206: Deferred assertions are sensitive to glitches

Dmitry:    If a deferred assertion references both a design variable and

     a test bench variable, then there could be false firing.

Tom: So you are saying that the deferred assertion will give its result

     in the Reactive region, but reactive region events could cause another

     pass through the active/NBA regions, resulting in a second

     evaluation being reported in the same time step.

Tom: Don't see how you are going to fix this. How would you know

     that any pass through the Reactive region is going to be the

     last for that time step?

Dmitry:    A very serious problem?

Ed: Could you use a concurrent assertion instead if you are concerned

     about this?

Dmitry:    Concurrent assertions are less friendly.

Tapan:     What is typical use model where a deferred assertion references both

     test bench variables and design variables?

Dmitry:    Imagine you are replacing part of design with a behavioral model

     written in SVTB. Now deferred assertions may reference both

     design and test bench variables.

Tapan:     Is this a standard usage described in UVM?

Dmitry: Will add an example to the Mantis item.

- Requesting authorization to work on additional enhancements

Dmitry:    Would like to request working group to work on the following

     enhancements: 3206 3191

     3206: Deferred assertions sensitive to glitches

     3191: Allow sequence methods with sequence expressions

Ben: What about 3195? "Local variable flow-in"

Dmitry:    If we have permission, then we can work on several enhancements in

     parallel.

- F2F meeting (discussion)

Dmitry:    Would like to have face-to-face meeting in November.

     Suggest Silicon Vally location.

Ben: Won't be able to attend.

Scott:     Doubt that company will pay for travel. Can attend by phone.

     Second week of November not good.

Dmitry:    Goal is to discuss more enhancements.

Erik: Suggest sending out an e-mail poll.

- Enhancement progress update

2476: Need clarification about system functions $onehot, et

Dana: What about sub-dividing the functions in the BNF according to type

     that they return

Erik: Didn't see the need. These BNF productions are not in Annex A.

Dana: Purpose of BNF is to show how things are used. Things that return

     different types may only be used in the context where that type makes

     sense.

Erik Is that the case? production "expression" could take on many types.

Erik Will look into modifying BNF.

Dmitry:    Should I appoint reviewers, or just vote?

Manisha: In the BNF, the names of all the functions should be in red.

Dmitry:    All terminals should be in red. Also parenthesis should be in red.

Erik: Will fix that.

Dmitry:    When updated proposal is ready, will call for vote.

Erik: Should we get sv-bc feedback before vote?

Dmitry:    We'll vote on it first, then send to sv-bc

Checker usability enhancements.

Dmitry:    Went through the checker usability document, explaining the differnt

     sections.

Dmitry:    The semantics of the $sampled function are re-defined for rand

     checker variables.

Erik: Are you shifting the meaning of sampling?

Erik: Why does the text say the value from the postponed region.

Dmitry:    Regular variable : sampled value is same throughout time step

     Randomly assigned variable $sampled changes at observed region

Tom: But you shouldn't be referencing rand checker variable outside a

     checker. The checker code all executes in Reactive region.

     However, if XMRs are allowed to rand checker variables, then you will

     have a big mess.

Dmitry:    Other enhancements in document: add always_ff and always_comb

Dmitry: Continuous assignments:

Tom: Should not have cont assign that could update in two different regions

     Cont assigns should always update in Reactive region. Even though

     a checker argument updates in Active, the continuous assignment inside

     the checker should always update in the Reactive region.

Manisha: What is the Re-Inactive region

Tom: No assignments are allowed in the Re-Inactive region. A continuous

     assignment should never update there.

Dmitry:    Checker output arguments

     Connection to checker output argument should imply a continuous

     assignment.

Tom: One more comment: We have talked about how the current checker

     sampling breaks the rewriting algorithm. We should explain in this

     document what the rewriting algorithm is, why it's important, and

     how it is broken now.

Dmitry:    Will try to put that into the document.

Meeting Adjourned.
Topic revision: r1 - 2010-09-16 - 22:24:25 - ErikSeligman
P1800