FWIW, I can duplicate the problem with a call from a trivial C++ program. Whether or not the program actually is authorized does not seem to make a difference. Here's the entire program:
#ifdef __MVS__ // Same pragmas! #pragma runopts( STACK(64K,16K),ANYHEAP(192K,8K),HEAP(1024K),PLIST(HOST),POSIX(ON),TRAP(ON,NO SPIE),NOEXECOPS ) #endif #include <stdio.h> #include <stdlib.h> extern "OS" {bool ISAUTH();} int main(int argc, char* argv[]) { printf("ISAUTH() test sandbox entered\n"); printf("Headed off to ISAUTH()\n"); bool isauth = ISAUTH(); printf("Back from ISAUTH(), isauth=%d\n", isauth); // make sure call does not get optimized out printf("Adios muchachos!\n"); return 0; } Output is ISAUTH() test sandbox entered Headed off to ISAUTH() CEE3798I ATTEMPTING TO TAKE A DUMP FOR ABEND U4088 TO DATA SET: ... Charles -----Original Message----- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Charles Mills Sent: Thursday, May 28, 2015 6:34 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: Re: Mysterious U4088-63 from RPTSTG(ON) Well, I moved ISAUTH() unchanged to its own assembler module. No change in the error. I removed the IEABRCX DEFINE and bingo! It works. BTW, the RPTSTG(ON) output shows nothing unusual. I can post it if anyone thinks they would see something there. FWIW, here is the complete assembled code of the new ISAUTH. The addresses are of course different. CDSALEN is also a lot smaller because ISAUTH basically needs none of the work area used by some of the other functions. ***That's not the key difference -- the smaller CDSALEN still fails if IEABRCX DEFINE.*** Other than that, the only difference I can see relative to the old code is the base/displacement branches rather than branch relatives. (The EXTRN CEESTART is generated once per assembly by EDCPRLG. It's earlier in the listing in the big module.) I am *totally* mystified. Even if you posit a (HIGHLY unlikely!) bug in TESTAUTH or LE, I don't see how substituting two BRC's for BC's could expose it. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN