RE: [sv-ec] Strict FIFO ordering of mailboxes try_put/try_get

But doughnuts are bad for you.  Shouldn't the language prevent you from
getting doughnuts at all?  Oh, my mistake.  That would be VHDL. :)

Stu
~~~~~~~~~~~~~~~~~~~~~~~~~
Stuart Sutherland
Sutherland HDL, Inc.
stuart@sutherland-hdl.com
503-692-0898
 

> -----Original Message-----
> From: owner-sv-ec@server.eda.org 
> [mailto:owner-sv-ec@server.eda.org] On Behalf Of Jonathan Bromley
> Sent: Wednesday, February 14, 2007 7:04 AM
> To: Mark Strickland (mastrick); Rich, Dave; sv-ec@server.eda-stds.org
> Subject: RE: [sv-ec] Strict FIFO ordering of mailboxes try_put/try_get
> 
> > Is there a big problem with the rule being that a try_get() 
> > succeeds if and only if BOTH the mailbox queue size is
> > greater than 0 AND the queue of processes suspended on
> > a blocking get has a size of 0?
> 
> Either try_get() jumps the queue of waiters if it 
> succeeds (Dave Rich's proposal), or it loses its 
> place in the queue if it fails (Mark's proposal).  
> Neither behaviour is likely to satisfy all users.
> 
> Consider a doughnut stall.  There is a FIFO queue
> of customers, and FIFO delivery of doughnuts to
> customers as they are prepared.  Now I arrive
> at the doughnut stall.  If I see no queue of 
> customers, and no doughnuts, I may choose to 
> go away and return later (try_get()).  If I see
> a queue of customers, I do not go away to return
> later, as I thereby lose my place in the queue. 
> Instead I put my friend in the queue as a proxy,
> and occasionally take a look to see how far down 
> the queue my friend has progressed.  When my
> friend is served, he will stand nearby and I
> can pick up my doughnut whenever I choose.  My
> try_get() waits on the proxy (my friend) and 
> not on the main queue of customers.
> 
> To implement this kind of mechanism we need some
> additional information that's not available in
> mailboxes right now.  The failing return value 
> from try_get() needs to distinguish between 
> "queue of customers" and "no doughnuts", so that 
> the caller can make a sensible choice of whether 
> to put its friend (proxy process) in the queue.
> If we follow Mark's suggestion, we *nearly* have
> this already - if num() is nonzero and try_get()
> fails, then there is a queue of customers - but
> obviously it would be better to have a single call
> to test both things atomically.
> 
> Mutatis mutandis for try_put(), of course.
> -- 
> Jonathan Bromley, Consultant
> 
> DOULOS - Developing Design Know-how
> VHDL * Verilog * SystemC * e * Perl * Tcl/Tk * Project Services
> 
> Doulos Ltd. Church Hatch, 22 Market Place, Ringwood, 
> Hampshire, BH24 1AW, UK
> Tel: +44 (0)1425 471223                   Email: 
> jonathan.bromley@doulos.com
> Fax: +44 (0)1425 471573                           Web: 
> http://www.doulos.com
> 
> The contents of this message may contain personal views which 
> are not the views of Doulos Ltd., unless specifically stated.
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 
> 
> 



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