You, of course, write open-source software. They didn’t. And for the sake of your mobile users, you both need to be friends. Enter Frida, the diplomat (she’s really only a library, but don’t tell anyone). She has coaxing superpowers that allow you to expose the innards of binary-only software, be it other libraries, operating systems, or other OS processes you must deal with. You can program Frida to infiltrate closed-source software, and expose their internals into abstractions you can use for testing the interoperability of your software. Want to lift some of their logic into your mock? Or replace a few functions in their binary code with your mocks? Hopefully, you want to do that using high-level languages such as JavaScript and/or Python, because those are the ones Frida likes the most.
In this talk, we use Frida, the scriptable, dynamic instrumentation toolkit, to expose internal functionality from binary-only software. By exposing internal functions and data structures, tightly integrating software often becomes easier to test at fine granularity. What previously had to be larger integration tests dependent upon several running subsystems, may, with a little effort, become isolated test fixtures that are easier to reason about. We show you unfortunate souls who have to deal with this level of interoperability how to program Frida to identify and expose functions in remote processes, and how to combine these exposed functions into small test fixtures in a unit-testing style.
