3.4 Procedural Abstraction

Description

Procedural Abstraction is essentially a reverse inlining process and is performed on identical assembly code sequences that appear more than once. Identical code sequences are abstracted into a callable routine, and each instance of the original code sequence is replaced with a call to that routine. This optimization reduces code size considerably, but will impose a small reduction in execution speed. This optimization can, however, adversely impact debugging.

When using MPLAB XC8, Procedural Abstraction optimizations are automatically turned on with level s optimizations (-Os). Any abstracted assembly sequences will be placed in a separate section to the original code. If there are linker options or linker scripts being used to locate code at specific addresses, these might not correctly handle the sections containing the abstracted code.

Migration

Procedural Abstraction optimizations can be turned off if they interfere with the linking requirements of a project. Completely disable Procedural Abstraction for the entire program using the -mno-pa option. The content of individual files or functions can be protected from Procedural Abstraction by using the -mno-pa-on-file=filename or -mno-pa-on-function=function options respectively. There is also a nopa attribute (or __nopa specifier) that can be used to disable Procedural Abstraction for a particular function.