Experimental Physics and
Industrial Control System

On 01/31/2018 11:52 AM, Simon Reiter wrote:
> On 01/31/2018 10:15 AM, Ralph Lange wrote:
>> Have a look at
>> http://www-csr.bessy.de/control/SoftDist/sequencer/Tutorial.html#asynchronous-use-of-pvput
>>
>> I think the SYNC option of pvPut(), which turns it into a blocking
>> call, might help.
>>
>> Cheers,
>> ~Ralph
>>
> I tried this now with
>> #define SET_PV(pv_variable, value) \
>> { \
>>     pv_variable = value; \
>>     pvPut(pv_variable, SYNC);
>> }
> but I still observe the same behavior in EPICS3.16 and EPICS7.
> 
> Switching to ASYNC and using pvPutComplete into the when condition did
> not solve the problem
>> 2018-01-31 11:48:46.633 [fine   ] [sm] new state reached: 6 (STARTING)
>> 2018-01-31 11:48:46.634 [fine   ] [sm] new state reached: 3 (RUNNING)
> The log messages (ASYNC & pvPutComplete) in the IOC looks fine.
> 
> Further ideas?

I have tried to reproduce your problem by simplifying it to the bare
essentials: three states that do nothing but announce themselves through
a PV (using your macro and parts of your enum definition) and then
(unconditionally) go to the next state, looping the whole process for 10
iterations. I camonitor the PV state_current and expect to get a
sequence of values 1 2 3 1 2 3 etc. With normal pvPut this indeed fails:

franksen@tiber: ~ > camonitor state_current
state_current                  2018-01-31 12:54:53.663351 3
state_current                  2018-01-31 12:56:20.357273 1
state_current                  2018-01-31 12:56:20.357293 1
state_current                  2018-01-31 12:56:20.357348 2
state_current                  2018-01-31 12:56:20.357348 2
state_current                  2018-01-31 12:56:20.357418 3
state_current                  2018-01-31 12:56:20.357418 3

The reason it fails is that the fire-and-forget pvPut commands are
issued in a very tight loop here; the server receives the next ca_put
message before it can act (process, issue monitor events) on the first.

But with pvPut(pv_variable, SYNC) I get the expected

franksen@tiber: ~ > camonitor state_current
state_current                  2018-01-31 12:56:20.357418 3
state_current                  2018-01-31 12:57:35.429494 1
state_current                  2018-01-31 12:57:35.429605 2
state_current                  2018-01-31 12:57:35.429714 3
state_current                  2018-01-31 12:57:35.429801 1
state_current                  2018-01-31 12:57:35.429885 2
state_current                  2018-01-31 12:57:35.429965 3
state_current                  2018-01-31 12:57:35.430041 1
state_current                  2018-01-31 12:57:35.430120 2
state_current                  2018-01-31 12:57:35.430234 3
state_current                  2018-01-31 12:57:35.430343 1
state_current                  2018-01-31 12:57:35.430450 2
state_current                  2018-01-31 12:57:35.430590 3
state_current                  2018-01-31 12:57:35.430679 1
state_current                  2018-01-31 12:57:35.430771 2
state_current                  2018-01-31 12:57:35.430865 3
state_current                  2018-01-31 12:57:35.430962 1
state_current                  2018-01-31 12:57:35.431060 2
state_current                  2018-01-31 12:57:35.431153 3
state_current                  2018-01-31 12:57:35.431274 1
state_current                  2018-01-31 12:57:35.431381 2
state_current                  2018-01-31 12:57:35.431476 3
state_current                  2018-01-31 12:57:35.431559 1
state_current                  2018-01-31 12:57:35.431653 2
state_current                  2018-01-31 12:57:35.431745 3
state_current                  2018-01-31 12:57:35.431838 1
state_current                  2018-01-31 12:57:35.431933 2
state_current                  2018-01-31 12:57:35.432038 3
state_current                  2018-01-31 12:57:35.432123 1
state_current                  2018-01-31 12:57:35.432204 2
state_current                  2018-01-31 12:57:35.432281 3

I attached the test program and the db file.

Cheers
Ben
-- 
"Make it so they have to reboot after every typo." ― Scott Adams

record(mbbo,"state_current") {
}

program problem

option +s;

#define SET_PV(pv_variable, value) \
{ \
    pv_variable = value; \
    pvPut(pv_variable, SYNC); \
}
%%enum { State_UNKNOWN = 0, State_READY = 1, State_STARTING = 2, State_RUNNING = 3 };

int pv_state_current = State_UNKNOWN;
assign pv_state_current to "state_current";

ss problem {
  int counter=0;
  state S_READY {
    entry {
      SET_PV(pv_state_current, State_READY);
    }
    when() {
    } state S_STARTING
  }
  state S_STARTING {
    entry {
      SET_PV(pv_state_current, State_STARTING);
    }
    when() {
    } state S_RUNNING
  }
  state S_RUNNING {
    entry {
      SET_PV(pv_state_current, State_RUNNING);
    }
    when (++counter >= 10) {
    } exit
    when() {
    } state S_READY
  }
}

Attachment: signature.asc
Description: OpenPGP digital signature

Replies:

Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

References:

Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Re: Sequencer seem to skip states with EPICS 3.16.1 Ralph Lange

Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Re: Sequencer seem to skip states with EPICS 3.16.1 Ralph Lange

Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Navigate by Date:

Prev: Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Next: Re: EPICS database macro warning Mark Rivers

Index: 1994  1995  1996  1997  1998  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: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Next: Re: Sequencer seem to skip states with EPICS 3.16.1 Simon Reiter

Index: 1994  1995  1996  1997  1998  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