Robert mentioned that it might be looking into where instrumentation is placed in a program in order to optimise it. The motivating (overly simple) example is that you wouldn't want the instrumentation for a locking function to go in the lock function itself as not every lock is instrumented, and the overhead tends to stack up. Instead, you'd want to put it at the caller site.
If possible, can I build an example demonstrating this as a problem (and show an analysis that fixes it)?
Robert mentioned that it might be looking into where instrumentation is placed in a program in order to optimise it. The motivating (overly simple) example is that you wouldn't want the instrumentation for a locking function to go in the lock function itself as not every lock is instrumented, and the overhead tends to stack up. Instead, you'd want to put it at the caller site.
If possible, can I build an example demonstrating this as a problem (and show an analysis that fixes it)?