Experimental Physics and
Industrial Control System

Eric Norum <[email protected]> · Thu, 10 Jan 2013 06:51:50 -0800

On Jan 10, 2013, at 6:26 AM, Jack Smith <[email protected]> wrote:

> Hi Eric, Andrew, et al:
> 
> Thanks for all your explanations. I tried Eric's codes and it works. I
> think Eric's algorithm is accurate enough for compensating the time
> spent on processing the records in its chain. Now, the delay in
> epicsEventWaitWithTimeout is (scan_period - time_on_scanList).
> 
> Since I have more questions in dbScan.c, I am starting a new thread ...
> 
> 1. In Eric's previous email, it says "It's the
> epicsEventWaitWithTimeout().  This allows the pause to be interrupted
> early in the case of IOC shutdown.". What does "This allows the pause
> to be interrupted early in the case of IOC shutdown" mean?

Think about what would happen if the periodTask loop just did an epicsThreadSleep at that point.  Things would work pretty much the same.  Well, they'd work the same until someone decided to shut down the IOC.  The SCAN="10 second" thread could take up to 10 seconds to reawaken and notice that ppsl->scanCtl == ctlExit and break out of the loop.
To avoid this problem the periodTask loop invokes epicsEventWaitWithTimeout to perform the delay.  When the IOC begins to shut down it sets  ppsl->scanCtl = ctlExit and raises the ppsl->loopEvent.  The epicsEventWaitWithTimeout then returns right away, the thread breaks out of the loop and terminates.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System