Programming considerations

The exits must follow standard linkage conventions. They must save the registers on entry and restore the registers when they return. The exit must be reentrant, refreshable, reusable, and reside in an APF-authorized library. It is inadvisable for the exit to have APF authorization unless it is designed also to be called as the entry point for a job step program.

The exits can use any of the TSO/E service routines. For a description of the service routines, see .

In some cases, you may not need to write a termination exit. This depends on:

Whether the processing that the initialization exit performs requires a termination exit to perform clean-up activities
How you use the exits to customize PARMLIB processing

If the initialization exit obtains a system resource, you must write a termination exit to free the resource. For example, if the initialization exit obtains storage to return a new command buffer to the PARMLIB command processor, you must provide a termination exit to free the storage. If the initialization exit obtains storage, for example, for a new command buffer, it must obtain the storage from subpool 1.

The initialization exit can change PARMLIB operands using the command buffer. The exit checks the command buffer it receives and determines whether to change any operands. To change the operands, the exit must:

Obtain storage from subpool 1 for a new command buffer
Build the new command buffer
Update the key, length, and data fields for the new command buffer (parameter entry 2)
Set return code 0 and return control to PARMLIB.

For more information about the parameter entries, see TSO/E standard exit parameter list. For more information about the format of the command buffer, see Command buffer.