[sv-bc] RE: Mandated warnings -- was Re: [sv-ec] Mantis 2701, ballot ID #44 - Arturo's feedback

From: Steven Sharp <sharp_at_.....>
Date: Thu May 07 2009 - 11:44:05 PDT
>From: Brad Pierce <Brad.Pierce@synopsys.com>

>The no-op semantics of Verilog out-of-bounds writes (for known indices) does 
seem weird. By analogy with the x-pessimism of much of the language, I would 
have expected maybe the entire array to be x'ed out.


For known indices, this seems completely natural to me.  An array was
originally intended to represent a memory.  If the address is fully
decoded, an out-of-range address won't select this memory, so a write
will have no effect.  If the address is not fully decoded (i.e. the
upper bits that make it out-of-range are ignored), then it should have
been represented by only applying the decoded bits to the memory.  Then
it would have the same "wrap-around" effect as the real hardware.  So
I don't think this is at all weird.

For unknown indexes, I agree that a no-op write is serious x-optimism.
To avoid this optimism, it really should X the whole memory, or the
addresses for every possible address considering X to be 0 or 1.  This
presumably was not done because it would be so expensive.

Steven Sharp
sharp@cadence.com


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Thu May 7 11:45:47 2009

This archive was generated by hypermail 2.1.8 : Thu May 07 2009 - 11:48:01 PDT