Deciding when to request an SVC dump
Generally, the programs that request dumps of virtual storage are
recovery routines, which run in a recovery environment. Under certain
circumstances, a recovery routine might not need to request a dump
of any storage. For example:
- Another recovery routine might have already requested an SVC dump for this error. If more than one recovery routine might receive control, each routine should determine whether another routine already requested a dump by checking the SDWAEAS bit of the system diagnostic work area (SDWA).
- Some errors do not require a dump for diagnosis. For example, a system completion code of X'913' results when an operator or user does not supply a correct password.
Note: If your program calls a system service (by issuing
a macro or callable service), that system service might encounter
a user-induced error and end abnormally. Generally, the system does
not take dumps for user-induced errors. If you require such a dump,
then it is your responsibility to request one in your recovery routine.
See Providing recovery for information about writing recovery
routines.
When a dump is required for diagnosis, an authorized program can
request either an SVC dump or an ABEND dump. Requesting an SVC dump
rather than an ABEND dump has several advantages:
- SVC dumps are delivered to the installation's system programmer rather than to the person who ran the abnormally terminating job. Usually, the system programmer wants to see the dumps requested by authorized programs immediately.
- The system processes an SVC dump whether or not the recovery routine percolates or requests a retry. In contrast, the system processes an ABEND dump only if all functional recovery routines (FRRs) percolate, or if no ESTAE-type recovery routine suppresses the dump.
- The system processes an SVC dump closer to the time of error, so the SVC dump might contain more useful data for diagnosis. The system handles an SVC dump request right after the program issues the SDUMPX macro. In contrast, the system handles an ABEND dump request either just before a retry, or after all of the recovery routines have percolated. During percolation, the recovery routines might have altered storage contents or dump options.