Weird problem with subfile in V5R4

All,
I have a client on V5R4.
Recently, I made a change to a program that had not been changed since V5R3.
The program displays a subfile and then chains from record 1 to the end processing the data.
What is happening now is that even though the record number used to chain to the subfile changes, the data returned is always from record 1. The program in debug mode indicates that the record number being returned does in fact match the record used to chain.
Tested with an old copy of the program from V5R3 and it works as it should.
The following is part of the DDS for the subfile, and the relevant part of the RPGLE code:
A R DETAIL
A*%%TS SD 20080319 155307 VAIJY REL-V5R4M0 5722-WDS
A SFL
A SBBOCD 3S 0H
A SBORD 9S 0H
A SBLINE 4S 0H
A SBITDC 3S 3H
A SBLOC 4A H
A HSEL 1A H
A SBPRIC R H REFFLD(##PRCE_SEL)
A SBFPRIC R H REFFLD(##PRCE_SEL)
A SBITD2Y R H REFFLD(##DESC_30)
A SBYNTX 1A H
A $BPRIC 11S 4H
A SVIN25 1A H
A SBITEM R O 5 1REFFLD(##ITEM)
A COLOR(WHT)
A SBITD1 R O 5 22REFFLD(##DESC_30)
A COLOR(WHT)
A SBITD2X 14A O 5 53TEXT('Description')
A COLOR(WHT)
A SBINV R Y O 5 68REFFLD(##INV)
A COLOR(WHT)


c z-add 1 rr#
c rr# cabeq 0 errorc move *off *in99
c dou *in99 = *on
c move *off *in34
c rr# chain detail 99
c if *in99 = *on
c leave
c endif

A breakpoint at the endif after the chain shows the same data being retreived regardless of the value in rr#.

I tried to duplicate this in a small sample program, but can not make it fail.
This is a very important program for the client, and also, this type of processing is common throughout the system.
I am very concerned that if there is a IBM issue here, they may start having other serious problems.

TIA,

Jeff Young
Sr. Programmer Analyst
IBM -e(logo) server Certified Systems Exper - iSeries Technical Solutions V5R2
IBM Certified Specialist- e(logo) server i5Series Technical Solutions Designer V5R3
IBM Certified Specialist- e(logo)server i5Series Technical Solutions Implementer V5R3