EPICS Controls Argonne National Laboratory

Experimental Physics and
Industrial Control System

1994  1995  1996  <19971998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024  Index 1994  1995  1996  <19971998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
<== Date ==> <== Thread ==>

Subject: flaky IOC problems at Jefferson Lab
From: Marty Kraimer <[email protected]>
To: [email protected]
Date: Fri, 10 Jan 1997 08:09:22 -0600
This message discusses IOC CPU Priorities.

How about the following priorities, from highest to lowest.

IOCHIGH
IOCMEDIUM
CAHIGH
IOCLOW
CAMEDIUM
CALOW
CAUDP and CATCP
BACKGROUND

All scan tasks can be IOCHIGH, IOCMEDIUM, IOCLOW, or BACKGROUND.The
actual choices an application developer is given are: background, low,
medium, and high.

Connected CA clients can be CAHIGH, CAMEDIUM, CALOW, or BACKGROUND. The
actual choices a CA user is given are: background, low, medium, high.

CAUDP is the task listening for ca search requests. CATCP is the task
listening for ca TCP connection requests.

Discussion:

This allows the IOC application developer the ability to guarantee that
important IOC based needs are satisfied first. At the same time, for
heavily loaded IOCs, the developer can assign periodic, I/O interrupt,
and event scanned records to a priority of IOCLOW or BACKGROUND. Thus
she/he can assure that non-time critical IOC functions do not lock out
CA client requests.

For BACKGROUND tasks it would be nice to turn on round-robin scheduling.
The problem is that vxWorks requires round-robin scheduling for all
tasks or for none (I am looking at a vxWorks 5.1 manual). Note that this
can be used now by just inserting a call to kernelTimeSlice in a st.cmd
file. Note that round-robin scheduling was not always a feature of
vxWorks.

Another feature of vxWorks that was not always available is priority
inheritance semaphores. Currently EPICS never uses this feature. I don't
know what the overhead is for using this feature. This is extremely
important because EPICS uses semaphores extensively. If priority
inheritance slows down semTakes considerably it would severly degrade
ioc performance.

If anyone has performance figures for either round-robin scheduling
and/or priority inversion, please send a message to tech-talk.

Marty


Replies:
Re: flaky IOC problems at Jefferson Lab watson
Re: flaky IOC problems at Jefferson Lab Jeff Hill

Navigate by Date:
Prev: edd 2.2, Solaris 2.5 and Sun patch 103210-09 Andrew Johnson
Next: Re: flaky IOC problems at Jefferson Lab watson
Index: 1994  1995  1996  <19971998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
Navigate by Thread:
Prev: Re: flaky IOC problems at Jefferson Lab watson
Next: Re: flaky IOC problems at Jefferson Lab watson
Index: 1994  1995  1996  <19971998  1999  2000  2001  2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016  2017  2018  2019  2020  2021  2022  2023  2024 
ANJ, 10 Aug 2010 Valid HTML 4.01! · Home · News · About · Base · Modules · Extensions · Distributions · Download ·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing ·