TRAP(OFF) in the PARM parameter did the trick, the 0C1 is caused by a CLFI instruction which is obviously not supported by a z850. I had sent the customer a version of the product compiled with XL/C at ARCH(7) but that was not low enough! It looks like ARCH(5) is necessary.
Thanks to everyone who helped. Robin -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Peter Hunkeler Sent: 12 July 2017 01:32 To: IBM-MAIN@LISTSERV.UA.EDU Subject: AW: Re: LE strikes again >>SLIP SET,C=0C1,J=jobname,ML=1,END >> >>Then have them load the dump into IPCS, select option 2.2 and send you the results. > > >I don't think you can SLIP trap an 001 program check as long as LE is running with TRAP(ON,SPIE). LE's ESPIE exit will gain control before SLIP; and LE will do error handling and at the end isseue SVC 0D (ABEND) with abend code S0C1. > > >This is why the failing instruction in the symptom dump points to 0A0D, and not the original failing instruction address. It occurred to be that my statement above was not accurate. A SLIP for S0C1 will of course also match when TRAP(ON,SPIE) is in effect, but only for the ABEND LE issues at the end of its error handling. Thus you will see the SVC D instruction as the failing instruction. When run with TRAP(OFF), the SLIP will match on the original S0C1 that RTM issues (lack of any ESPIE) for the 001 program check. And this dump will show PSW and registers at time of the original 001 program check. -- Peter Hunkeler ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN