I've tried specifying lots of different parms in the CEEOPTS DD without success.
Today I tried something completely different. My code is very simple, I define a bunch of variables in working storage and then invoke an ICSF API. I do a couple of displays before and after the call, but the call is abending with the S0C4. I added 4 instructions to the code. The first three simply define 3 new variables, one of which is initialized to 0. Immediately before the ICSF API call, I perfrom a DIVIDE by 0, which results in a S0C9 AND THE DUMP BEING GENERATED in the CEEDUMP DD. Commenting out the DIVIDE by 0 instruction, the job abends with the S0C4 and no dump. The only difference between the two runs is the DIVIDE by 0 instruction. Interestingly, with the Abend0C9, the joblog includes IEA995I, which displays 'USER COMPLETION CODE=4039'. I believe the U4039 is a result of LE percolating the S0C9. With the S0C4, the job log shows only IEF450I, which simply displays 'ABEND=S0C4'. So, I'm guessing that LE is getting control after the S0C9 and generating the dump with the U4039, but with the S0C4, LE is being bypassed and no dump is created. I also tried specifying the ABTERMENC option in the CEEOPTS input. Specifying ABEND did not generate a dump. Specifying ABTERMENC(RETCODE) did change the Abend S0C4 to return code=3000, but there still was no dump. Does this provide any more clues to why the dump is not being generated? Is there another option to tell LE to process the abend and generate the dump for the S0C4? Thanks! Greg Boyd www.mainframecrypto.com On Mon, 13 Feb 2017 13:29:28 +0100, Peter Hunkeler <[email protected]> wrote: >Have you tried >//CEEOPTS DD * TERMTHDACT(UAIMM,,96)/* > >The system will use the *last* dump dd statement out of //SYSABEND, //SYSUDUMP >and //SYSMDUMP found in the step to decide how to format (or not) and where to >write the dump. With the above option, you should get s system dump but not a >CEEDUMP. > >Note that the system might suppress a dump to //SYSMDUMP if DAE thinks this is >a dulicate. You should have a corresponding message in the jobs log it this is >the case. > > >-- >Peter Hunkeler > >---------------------------------------------------------------------- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
