> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Thompson, Steve (SCI
TW)
>
> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Chase, John
>
> Hi, All,
>
> Got a CICS application program that branches out of itself
> (stays in the CICS region) somehow, and the SLIP
> documentation doesn't make it "obvious" how I should set a
> trap to catch it and SDUMP the region.
> I've looked mostly at IF and SBT scenarios, but can't seem to
> understand how to code the trap to catch a (branch to a)
> target address outside the current program. I've gotten a
> few IF dumps at branch instructions so far, but haven't
> caught the right one yet.
> <SNIP>
>
> Are you able to deduce the program doing the wild branch?
Yes; explicitly identified by the CEDF diagnostic utility of CICS.
> And
> do you know the language it is written in?
Yes; VS Cobol II.
> Have you checked
> the contents of the registers at the point you capture the
> dump to see if they give any indication of the culprit?
Yes; they do not.
> Have you looked at the system trace information (hopefully
> you captured it in your dump)?
No; I haven't trapped the "right" branch yet.
> Do you have any idea where the branches are going to (a
> particular range of storage or just all over)?
Yes and No: It's an ISV application, and is presently running in three
different CICSes on two different CPUs: one a z9-BC and the other a
z9-EC. In ONE of the CICSes on the z9-BC the program runs correctly.
In the other two it branches to "the twilight zone" where CEDF next
intercepts an EXEC CICS ENDBR command with a filename of all blanks.
Problem is that the program never issues a STARTBR command in the first
place. The function of the program (on first invocation) is to display
the application's sign-on screen.
Now, in the instance on the z9-EC, CEDF gives an offset of x'61E192'
into the x'11108'-byte-long program (thus my conclusion that it's a
"wild branch"). In the z9-BC instance, CEDF does not show an offset
value. Yet in both cases the "wild branch" goes somewhere where the
next CICS command is ENDBR. And CEDF doesn't show "current" registers
or PSW when it intercepts an EXEC CICS command.... :-(
-jc-
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
