EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  <20102011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024  Index 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  <20102011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: Re: do flnk-ed sequences of records get interrupted?
From: Andrew Johnson <[email protected]>
To: [email protected]
Cc: Dennis Nicklaus <[email protected]>
Date: Wed, 3 Feb 2010 10:05:31 -0600
Hi Dennis,

On Wednesday 03 February 2010 09:11:39 Dennis Nicklaus wrote:
> Can someone please explain a few more details about epics processing
> sequences of FLNK-ed records.  Once an flnk-ed sequence begins, can/will
> another different flnked sequence begin?

Yes it can, for a few reasons.  If a second chain is in a different lock-set 
and being executed by a different scan thread then it can be kicked off 
whenever it's ready and that thread gets CPU time (which may depend on 
relative thread priorities).  Even if the second chain is in the same lock-set 
and uses the same thread it can start processing if one of the records in the 
first chain starts asynchronous record processing (the sequence record 
processes its links this way, as does calcout if you set ODLY).

> Do I have to assume that unrelated (no shared PVs) flnk-ed series
> can/will run in parallel?

Yes, but (reading on) asynchronous device support usually queues up I/O 
requests to the same port, forcing the I/O from later chains to wait.  It can 
depend on that device support and the relative thread priorities what order it 
will actually execute multiple waiting operations.

> In our case, I have a series of records that implements several reads of
> a slow serial stream device, and associated other processing. I would
> like to protect against a user setting a PV that will cause a write to
> that same device before my series of reads is finished.  Do I need to
> implement some protection?  I do know about the statement in the
> StreamDevice manual that says, "The first out command in the protocol
> locks the device for exclusive access. That means that no other record
> can communicate with that device." But I think I'm looking for a
> different locking across a sequence of records.

Your question is really about whether streamDevice provides ways to control 
port locking, which someone else will have to respond since I've never used it 
myself.  However I think it's going to depend on whether your series of reads 
are all coming from the same message from the device or not; if each read is a 
separate write-read operation then I suspect you're going to have to implement 
your own protection mechanism in the database, but if all your reads are 
pulling data out of the same device response then I think you'll be safe.

- Andrew
-- 
The best FOSS code is written to be read by other humans -- Harald Welte


References:
do flnk-ed sequences of records get interrupted? Dennis Nicklaus

Navigate by Date:
Prev: do flnk-ed sequences of records get interrupted? Dennis Nicklaus
Next: Re: do flnk-ed sequences of records get interrupted? Tim Mooney
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  <20102011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: do flnk-ed sequences of records get interrupted? Dennis Nicklaus
Next: Re: do flnk-ed sequences of records get interrupted? Tim Mooney
Index: 1994  1995  1996  1997  1998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  <20102011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 02 Sep 2010 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·