Experimental Physics and
Industrial Control System

"Hu, Yong via Tech-talk" <tech-talk at aps.anl.gov> · Thu, 6 May 2021 19:03:31 +0000

I just did a quick test. Randall's problem is repeatable. If a record (e.g. testAsyn) has DTYP "asynInt32" but no field(INP, "@..."), its FLNKed record will not be processed after a put to testAsyn.  However, the value of  testAsyn is changed although its alarm severity is INVALID.

[yhu@sps-diaglabs-dev-ibm example]$ caput testAsyn 1
Old : testAsyn                       0
New : testAsyn                       1
[yhu@sps-diaglabs-dev-ibm example]$ caget -a testAsyn
testAsyn                       <undefined> 1 UDF INVALID

And I get a waring message related to devAsynInt32 after iocInit. Probably Randall did not pay enough attention to that kind of warning. 

Starting iocInit
############################################################################
## EPICS R7.0.5
## Rev. R7.0.5-dirty
############################################################################
testAsyn devAsynInt32::initCommon  invalid INST_IO Must be asynMask(<port> <addr> <mask> <timeout>)userParams

BTW, "PP" is not needed in the field FLNK. This seems more straightforward to me: field(FLNK, "$(P):fsCalc")

Cheers,
-- Yong 

On 5/6/21, 2:38 PM, "Tech-talk on behalf of Johnson, Andrew N. via Tech-talk" <tech-talk-bounces at aps.anl.gov on behalf of tech-talk at aps.anl.gov> wrote:

    > On May 6, 2021, at 1:05 PM, Mark Rivers via Tech-talk <tech-talk at aps.anl.gov> wrote:
    > 
    >> But fsControl is also SCAN=Passive in the example. For the FLNK or the INP CP to work, something must trigger fsControl to process.
    > 
    > Randall's original message said:
    > 
    >> If I put to fsControl, it sets but nothing happens.
    > 
    > So my understanding is that he was doing a caput to fsControl, but the FLNK was not processing fsCalc, as it should have.

    That would make sense given that DTYP was set; the asynInt32 device support will have disabled the record (leaving PACT as 1) when it didn’t see an address in the INP field. In that state a put to the ai record will not cause it to be processed.

    - Andrew

    > ________________________________
    > From: Hartman, Steven <hartmansm at ornl.gov>
    > Sent: Thursday, May 6, 2021 12:57 PM
    > To: Mark Rivers
    > Cc: Arnold, Ned D.; Randall Cayford; tech-talk at aps.anl.gov
    > Subject: Re: [EXTERNAL] calcout record processing
    > 
    > 
    > 
    > On May 6, 2021, at 1:52 PM, Mark Rivers via Tech-talk <tech-talk at aps.anl.gov<mailto:tech-talk at aps.anl.gov>> wrote:
    > 
    > He did FLNK from fsControl to fsCalc.  If that FLNK is not actually processing fsCalc than I think it may be because of the DTYP/INP mismatch.
    > 
    > 
    > But fsControl is also SCAN=Passive in the example. For the FLNK or the INP CP to work, something must trigger fsControl to process.
    > 
    > --
    > Steven Hartman
    > hartmansm at ornl.gov<mailto:hartmansm at ornl.gov>
    > 
    > 
    > 

    -- 
    Complexity comes for free, simplicity you have to work for.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System