RE: [sv-bc] Semantic check inside generate

Surya.

 

The LRM states that generate regions are processed after parsing and
during elaboration. My interpretation of "parsing" means that the code
must at least comply with the BNF, in which both cases do not since 'x'
is not a constant_expression. But I don't know if any implementation
actually uses the published BNF as is to build their parsers.

 

Dave

 

 

 

________________________________

From: owner-sv-bc@server.eda.org [mailto:owner-sv-bc@server.eda.org] On
Behalf Of Surya Pratik Saha
Sent: Wednesday, February 21, 2007 8:44 PM
To: sv-bc@server.eda-stds.org; sv-ec@server.eda.org
Subject: [sv-bc] Semantic check inside generate

 

Hi,
Recently I have come across some designs where some invalid semantically
wrong verilog code is written in dead/unreachable portion of generate
region. LRM is not clear enough which semantic check should be done at
elaboration time and which in parsing time. And also different tools are
behaving differently in different scenario. I think there should be
clear guideline in the LRM to eliminate the whole confusion. 

Case 1:
module top;
    integer x;
    generate if (0) begin:b
        if (x) begin:c
        end
        end
    endgenerate
endmodule

Here 'x' has to be constant expression but placed in unreachable portion
of generate. Synopsys VCS is passing it, but MTI-Vlog is failing.

Consider another similar case:
module top;
    integer x;
    generate if (0) begin:b
        if (x) begin:c
            reg[x:0] r;
        end
        end
    endgenerate
endmodule

For the reg declaration, the 'x' has to be again constant but placed in
unreachable portion of generate. But here VCS and Vlog both are failing
here. So VCS is not consistent enough to skip all types of semantic
checks in generate region. That makes a lot of confusion in parser
writing as well as design writing. LRM has to have some words for that.



Regards
Surya.

 


-- 
This message has been scanned for viruses and 
dangerous content by MailScanner <http://www.mailscanner.info/> , and is

believed to be clean. 

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.