rts 115 Unexpected signal (Signal 22)

[Rita C Workman]

For me not a clue...so here is my Search for Cobol Errors and at least what explanations I found for the signals you mention...hope it at least helps
===========================
022 Illegal or impossible access mode for OPEN (Recoverable)

* The mode in which you are trying to open a file
violates the General Rule of COBOL for that type of
file; for example you might have opened a line
sequential file in the I-O mode.

* Open the file with a mode that is compatible with that
type of file.
=========================
115 Unexpected signal (Fatal)

* A signal the run-time system was not expecting has
been caught.
=========================
Rit

[Helen French]

Hi Jon,

This may help you:

http://us-support3.external.hp.com/cki/bin/doc.pl/sid=9ea2683e0ba02add8e/screen=ckiDisplayDocument?docId=200000013473776

HTH,
Shiju

[Jon McDermott]

Thanks for the responses so far. They are along the lines of what I am thinking - a SIGIO signal is recieved while the code is in the "C" routine and for some reason their is no handler declared for it, so everything unravels and we end up with the RTS error.

If the "C" routine is built as a stand alone program and run outside of the RTS, it runs fine. There is I/O to a pipe involved, so there has to be some signal handling.

The "C" routine called from within the COBOL program has been working fine up until we updated the code in some of the libraries linked with the RTS. There were some changes in semaphore handling and minor signal handling changes, so that's where we are looking right now. We think there may be some conflicts between the signal handling in the "C" routines and HP COBOL, introduced with the signal handling changes.

The link to the signal_regime note was for an older version of COBOL. My current documentation discusses the "set signal_regime(sig)=integer" run-time tuneable. I'm thinking perhaps an entry in $COBDIR/cobconfig of:

set signal_regime(22)=1

might resolve the problem. Any comments?