Re: [sv-ec] [sv-bc] Semaphore question

Hi, All -

NOTE: it would be useful to have a semaphore.num() method to query how many 
keys are in the bucket.

Use models: Since processes can put more keys back into the bucket that 
they have removed, interesting use models include:

(1)     Do not "put" any semaphore keys into the bucket until after 
simulation has started to delay execution of all semaphore processes.

(2)     Half-way through a simulation, put more keys into the bucket to get 
more processes running simultaneously.

I have attached a few interpretations with examples. I now believe that 
semaphores were intended to behave the way Jamie outlined, but I think this 
needs to be quickly clarified and approved by the 1800 subcommittees and 
working groups.

I can support this behavior because it sounds like this was the behavior of 
Vera and I see no reason to surprise former-Vera users with a new behavior 
for a known syntax.

I believe I can also get the strict FIFO behavior by using the repeat 
command when I do a "get()".

We need to clarify this ASAP. I have already taught more than 100 engineers 
the "wrong" way that semaphore get() works (I hate it when I do that!)

Plus there is the interesting question that I do not believe has been 
answered yet, nor well documented, even though I think I know what the 
answer will be (the FIFO description can cause great confusion).

Question: What will happen if there are no keys, process-A requests two 
keys (is put on the requesting FIFO) and later process-B requests one key?

Is process-B put on the requesting FIFO behind process-A??

If one key is returned, does process-B come off the not-so-FIFO and become 
active??

I can live with Jamie, Arturo and Neil's interpretation of semaphores, but 
the LRM wording is clearly inadequate to describe this behavior.

Please review the attached examples for accuracy (examples are not tested).

Regards - Cliff

----------------------------------------------------
Cliff Cummings - Sunburst Design, Inc.
14314 SW Allen Blvd., PMB 501, Beaverton, OR 97005
Phone: 503-641-8446 / FAX: 503-641-8486
cliffc@sunburst-design.com / www.sunburst-design.com
Expert Verilog, SystemVerilog, Synthesis and Verification Training