Generate an error in the event of memory corruption

We would like an enhancement to protect or ensure that the storage areas that are used
by DB2 are not corrupted by a cobol program or RPG program / or if
there is corruption that an error is generated ( rather than returning
an sql code like +100).


This is related to PMR 37880-122-000 in which we found out an RPG
subroutine corrupted the db2 memory area used by the application.
It was an RPG or a cobol program that calls an RPG subroutine and the RPG
was populating the field passed in the linkage area and using the incorrect length.
The other situation we had was a cobol program calling an cobol subroutine with
(subroutine had a larger linkage field receiving value being passed
and then moved data into that field to be sent back.
Since then we have came across other errors during system testing that
linkage areas between subroutines are incorrect, and other sql errors
have been generated.

We want the system to confirm that the sql
memory area was not damaged by a rpg/cobol program prior to running an
sql statement. For example a checksum digit or something to confirm the
integrity of this storage. We realize our applications are partially
at fault but it seems like this area of storage should be protected.

We are also concerned that possibly we could get a sqlcode +000 and the
wrong account was updated or our application will falsely report there
are no accounts with that criteria and there are ones out on the
database.

We have been seeing return codes such as -302 that we determined was
due to memory corruption, and I suspect -204's we have encountered are
also related to this.

Currently we are reviewing the linkage areas on all of
subroutines and trying to take as much corrective action as possible.
We believe the potential is there and it may be a very hard
problem to solve/track back to if it happens intermittently in
production during the processing of accounts.

We have this issue in a cobol program that calls an RPG subroutine and the RPG was populating the field passed in the linkage area and using the incorrect length.

The other situation we had was a cobol program calling an cobol subroutine with (subroutine had a larger linkage field receiving value being passed, and then moved data into that field to be sent back.