Zephyr Developer Summit 2022

What makes for a good test?
It's a simple, it produces reliable results and in the end it runs on a wide range of platforms with various ISAs, compilers, target speeds and memory layouts.
There are lots of assumptions in existing tests, which may not be entirely correct for all possible platforms. Good examples are: test involves interaction between threads (are we aware of SMP platforms?); there's a wait for something (time period can be too long or short for target); test exploits undefined behavior (good luck with reproduction in a different environment). Ok, all above checked? Now meet compiler optimizations...

During this talk we'll look at various kinds of issues getting in the way of writing a robust test. It'll be based on real experience with debugging issues for various ARC platforms and it'll be helpful for both, those writing new tests and those debugging existing ones.