Argonne National Laboratory

Experimental Physics and
Industrial Control System

2002  2003  2004  2005  <20062007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  Index 2002  2003  2004  2005  <20062007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020 
<== Date ==> <== Thread ==>

Subject: RE: seq debugger
From: "Jeff Hill" <johill@lanl.gov>
To: "'Andrew Johnson'" <anj@aps.anl.gov>, "'Laznovsky, Michael'" <lazmo@slac.stanford.edu>
Cc: "'EPICS core-talk'" <core-talk@aps.anl.gov>
Date: Thu, 22 Jun 2006 10:07:14 -0600
Also note that variable accesses that might appear to be atomic on single
processor architectures might need mutex locking on SMP systems. A person
maintaining a multi-threaded code should read up on the additional
synchronization constraints imposed by SMP architectures (and their cache
coherency issues). I expect that increasing ubiquity of SMP based
motherboards will be a trend.

Jeff

> -----Original Message-----
> From: Andrew Johnson [mailto:anj@aps.anl.gov]
> Sent: Thursday, June 22, 2006 9:21 AM
> To: Laznovsky, Michael
> Cc: EPICS core-talk
> Subject: Re: seq debugger
> 
> Hi Michael,
> 
> Laznovsky, Michael wrote:
> >
> > Since the snc compiler doesn't currently generate any code to protect
> > any part of the "action" functions (the body of each "when"), perhaps
> > the easiest way to go would be to add semaphores around those (only if
> > compiled for debug, to avoid the overhead otherwise).  Any other ideas?
> 
> I don't know enough about the internal workings of the sequencer to
> provide specifics, but I wonder if there is currently any protection
> between a sequence action running and a CA event callback updating its
> monitored variable.  This would need exactly the same protection that
> we're talking about for the local variable access API, so you might want
> to investigate how that is done (if at all).  Of course the problem here
> is that single SNL program may have multiple state sets, and if you add
> a mutex the tasks for each state set could end fighting for access to
> the same (set of) variable(s).
> 
> - Andrew
> --
> Not everything that can be counted counts,
> and not everything that counts can be counted.
>    -- Albert Einstein


References:
Re: seq debugger Andrew Johnson

Navigate by Date:
Prev: Re: seq debugger Andrew Johnson
Next: RE: alarm hook Jeff Hill
Index: 2002  2003  2004  2005  <20062007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020 
Navigate by Thread:
Prev: Re: seq debugger Andrew Johnson
Next: epics regression tests on AMD64 running RHEL Linux 4.0 Ernest L. Williams Jr.
Index: 2002  2003  2004  2005  <20062007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020 
ANJ, 02 Feb 2012 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·