Re: Epics 3.14.8: faulty registrar statement in dbd causes memPartAlloc error on vxWorks

[Andrew Johnson <anj@aps.anl.gov>]

Hi Goetz,

On Thursday 27 May 2010 07:44:23 Goetz Pfeiffer wrote:
> I use Epics 3.14.8 for my application. I recently got rid of two of my
> sequencer programs due to various optimizations. But I still had a
> "registrar" statement in my dbd file, now for a module that no longer
> existed.

You should have seen a load failure complaining about missing symbols of the 
form pvar_func_<routine> for each of your missing registrar routines.  Those 
load failures should never be ignored, they indicate problems of this kind.

> I verified that any "registrar" statement for a non-existing module
> produces this error. But if you see this on the IOC console you have no
> idea whats wrong with your program. An error message like "invalid
> registrar [registrar-name]" would be much better.
>
> Does the newest Epics release still show this behavior and if it is so,
> are any plans to fix this ?

Please check whether you were getting those load errors or not.  The 
idcp_registerRecordDeviceDriver() subroutine that gets generated from your 
IOC's DBD file just calls every registrar function directly, it doesn't 
currently have the names of those routines as strings.  It could detect this 
case, but the load errors should be sufficient to tell you there's something 
wrong.

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