| Number | 276
|
| Category | errata
|
| Synopsis | A.6.4: BNF doesn't allow fork/join in functions
|
| State | lrmdraft
|
| Class | errata-simple
|
| Arrival-Date | Jan 28 2003
|
| Originator | sharp@cadence.com
|
| Release | 2001b: A.6.4
|
| Environment |
|
| Description |
The BNF for function_statement does not allow fork/join (par_block). However, fork/join is actually legal inside functions. It isn't really needed, since there can't be any delays, so a sequential block can be used instead. But Verilog-XL allows it, and so does NC-Verilog. |
| Fix |
In A.6.4, REPLACE function_statement WITH function_statement ::= statement plus a footnote that "Function statements are limited by the rules of 10.3.4." In A.6.2, REMOVE function_blocking_assignment. In A.6.3 and Syntax 9-13, REMOVE function_seq_block. In A.6.4, REMOVE function_statement_or_null. In A.6.6 and Syntax 9-4, REMOVE function_conditional_statement. In A.6.6 and Syntax 9-5, REMOVE function_if_else_if_statement. In A.6.7 and Syntax 9-6, REMOVE function_case_statement. In A.6.7 and Syntax 9-6, REMOVE function_case_item. In A.6.8 and Syntax 9-7, REMOVE function_loop_statement. In 10.3.4, REPLACE "a) A function definition shall not contain any time-controlled statements, that is, any statements introduced with #, @, or wait." WITH "a) A function definition shall not contain any time-controlled statements, that is, any statements containing #, @, or wait." and REPLACE "f) A function shall not have any nonblocking assignments" WITH "f) A function shall not have any nonblocking or procedural continuous assignments." and ADD the following rule -- "g) A function shall not have any event triggers." Note that this issue supercedes #229 (which was controversial). |
| Audit-Trail |
From: Shalom Bresticker <Shalom.Bresticker@motorola.com> To: Brad.Pierce@synopsys.com Cc: etf-bugs@boyd.com Subject: Re: errata/276: PROPOSAL - BNF A.6.4 doesn't allow fork/join in functions Date: Sun, 21 Sep 2003 15:58:32 +0300 While I like the general idea, a few more changes would be necessary: - REMOVE function_blocking_assignment in A.6.2 - Add the following restrictions in 10.3.4: * no intra-assignment delay and event controls (combine with 10.3.4(a)) * no event triggers (why not?) * no disable statements (why not?) * no procedural continuous assignments Shalom Brad.Pierce@synopsys.com wrote: > In A.6.4, > > REPLACE function_statement > WITH > > function_statement ::= statement > > plus a footnote that > > "Function statements are limited by the rules of 10.3.4." > > In A.6.3 and Syntax 9-13, REMOVE function_seq_block. > In A.6.4, REMOVE function_statement_or_null. > In A.6.6 and Syntax 9-4, REMOVE function_conditional_statement. > In A.6.6 and Syntax 9-5, REMOVE function_if_else_if_statement. > In A.6.7 and Syntax 9-6, REMOVE function_case_statement. > In A.6.7 and Syntax 9-6, REMOVE function_case_item. > In A.6.8 and Syntax 9-7, REMOVE function_loop_statement. > > http://boydtechinc.com/cgi-bin/issueproposal.pl?cmd=view&pr=276 -- Shalom Bresticker Shalom.Bresticker@motorola.com Design & Reuse Methodology Tel: +972 9 9522268 Motorola Semiconductor Israel, Ltd. Fax: +972 9 9522890 POB 2208, Herzlia 46120, ISRAEL Cell: +972 50 441478 From: Shalom Bresticker <Shalom.Bresticker@motorola.com> To: Brad.Pierce@synopsys.com Cc: etf-bugs@boyd.com Subject: Re: errata/276: PROPOSAL - BNF A.6.4 doesn't allow fork/join in functions Date: Tue, 30 Sep 2003 11:49:55 +0300 > [Note that intra-assignment delays and event controls > are not possible without @ and #, and these are already > disallowed by rule 10.3.4(a).] Strictly speaking, 10.3.4(a) disallows "any statements introduced with #, @, or wait". It would not be unreasonable to claim that an intra-assignment delay or event control does not INTRODUCE a statement, rather they are in the middle of a statement. Also, "disable" statements are not allowed in functions, though it is not clear to me why. > http://boydtechinc.com/cgi-bin/issueproposal.pl?cmd=view&pr=276 -- Shalom Bresticker Shalom.Bresticker@motorola.com Design & Reuse Methodology Tel: +972 9 9522268 Motorola Semiconductor Israel, Ltd. Fax: +972 9 9522890 POB 2208, Herzlia 46120, ISRAEL Cell: +972 50 441478 From: Shalom.Bresticker@motorola.com To: Brad.Pierce@synopsys.com Cc: etf-bugs@boyd.com Subject: Re: errata/276: PROPOSAL - BNF A.6.4 doesn't allow fork/join in functions Date: Tue, 30 Sep 2003 12:11:31 +0300 (IDT) > Also, "disable" statements are not allowed in functions, > though it is not clear to me why. My mistake: disables ARE allowed in functions. The BNF fooled me because the productions are ordered in "function_statement" differently than in "statement". -- Shalom Bresticker Shalom.Bresticker@motorola.com Design & Reuse Methodology Tel: +972 9 9522268 Motorola Semiconductor Israel, Ltd. Fax: +972 9 9522890 POB 2208, Herzlia 46120, ISRAEL Cell: +972 50 441478 From: Shalom.Bresticker@motorola.com To: Brad.Pierce@synopsys.com Cc: etf-bugs@boyd.com Subject: Re: errata/276: PROPOSAL - BNF A.6.4 doesn't allow fork/join in functions Date: Tue, 30 Sep 2003 12:31:20 +0300 (IDT) How about changing "introduced with" to "containing"? Shalom > Strictly speaking, 10.3.4(a) disallows "any statements introduced with #, @, or > wait". > > It would not be unreasonable to claim that an intra-assignment delay or event > control does not INTRODUCE a statement, rather they are in the middle of a > statement. -- Shalom Bresticker Shalom.Bresticker@motorola.com Design & Reuse Methodology Tel: +972 9 9522268 Motorola Semiconductor Israel, Ltd. Fax: +972 9 9522890 POB 2208, Herzlia 46120, ISRAEL Cell: +972 50 441478 From: Steven Sharp <sharp@cadence.com> To: etf-bugs@boyd.com, Shalom.Bresticker@motorola.com Cc: Subject: Re: errata/276: PROPOSAL - BNF A.6.4 doesn't allow fork/join in functions Date: Tue, 30 Sep 2003 14:49:21 -0400 (EDT) Just for reference, Verilog-XL does not allow intra-assignment event or delay controls in functions. It also does not allow nonblocking assignments, even without an explicit delay, since it treats them as having an implicit intra-assignment #0 (and they actually have a delay to the NBA queue, so if you disallow them with intra-assignment delays, it is consistent to disallow them entirely). Steven Sharp sharp@cadence.com |
| Unformatted |
|
Hosted by Boyd Technology