Experimental Physics and
Industrial Control System

"Mark Rivers" <[email protected]> · Fri, 24 Apr 2009 13:36:36 -0500

Thanks to Lewis for pointing our that the areaDetector main programs were not calling epicsExit(0).  They should have been, and that was causing a problem on Linux, where typing exit would NOT exit, it would hang.  The only way to kill the IOC was to type ^C.

I just added the epicsExit(0) call, and it fixes that problem on Linux.  I also checked that it does not introduce the problem Mark Clift is seeing on win32-x86, those IOCs continue to exit cleanly with no errors.  I am building everything statically on both win32-x86 and linux-x86.

Mark

________________________________

From: [email protected] on behalf of J. Lewis Muir
Sent: Fri 4/24/2009 11:01 AM
To: Jeff Hill
Cc: 'Mark Clift'; [email protected]
Subject: Re: Winsock error10004

On 4/24/09 11:42 AM, Jeff Hill wrote:
>> * Strangely, the ioc's that build under areaDetector 1-4 using
>>   win32-x86, do not have this error on exit and run fine.
>
> One possible difference could be systems with and without SMP
> (multi-core cpu or not). Alternatively, statically built or not might
> change the order of process rundown as executed by windows (i.e.
> threads stopped first versus shareable libraries unmapped from memory
> first).

Another possible difference is that the areaDetector IOCs don't invoke
epicsExit before returning from their main function; other IOCs do.

Lewis

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System