>At the end of task termination, we free the storage containing the TCB/STCB under which we are running, and set PSATOLD to the address of the WAIT TCB (which is in the nucleus). That gets traced in some SSRV 78 entries for some more FREEMAINs, and then we enter the dispatcher.
Hi Jim, Many thanks for the above details. Much appreciated! It helped. I think I understand what the root cause of the different ABENDs is: The ECB from an ATTACH has been overlaid. I just have to find the place in the code, now.... It took me a while to understand what the system trace was telling me; never before have I had the need to dig that deep into system trace, and to analyze a trace from a multitasking application. When I started to write this post -- actually the text below -- I still was under the impression that there will be questions I'd have to ask to be able to understand. Now, after finishing it, it seems all is clear. Instead of deleting the summary, I'll leave in the post. Maybe it is of interest to someone. I do have the relevant trace entries with some comments of mine in a PDF. If anyone is interested, just drop me a note. -- Peter Hunkeler This is on z/OS V2.1. Here is what I understand from the system trace : o TCB A (job step task) DETACHes subtask TCB B, which has not yet completed. o DETACH SVC code (under TCB A) initiates abnormal termination of subtask (S13E), then puts TCB A into a WAIT until subtask termination is complete. o TCB B gets dispatched and termination is starting via SVC D. o Many, many trace entries later, TCB B, branches to POST (from within DETACH SVC code) to post the ECB TCB A is waiting for. o The TCB/STCB are freed next, as explained by Jim. The TCB address passed to DETACH must have been correct, and the TCB and other structures not overlaid, otherwise we would not have come that far, right? o Next trace entries show TCB A being dispatched, still in DETACH SVC code. Just next entry, DETACH is branching to POST (SSRV 129) from IGC062+02D8. Is this to POST the ECB from the ATTACH of the task just terminated? o Next trace entry is the PGM 011 which IEAVEPST is suffering from offset 99E. The storage the TEA address points to is not allocated. This must be either the address from TCBECB or the address of the RB from the ECB. o Next entries, all for TCB A, are -- RCVY PROG for the 0C4-11 -- RCVY FRR when POST's FRR is entered. The 0C4-11 is changed to an S202-00, then percolation occurs -- RCVY FRR when DETACH's FRR is entered. The S202-00 is changed to S23E-00, then percolation occurs again -- RCVY PERC, when the S23E-0 is percolated to RTM2. -- SVC D, when RTM2 initiates termination of TCB A (which is the JSTCB). -- Peter Hunkeler ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
