Kernel Testing with KUnit: Bridging the Gap
summary
Compares KUnit and kselftest for when to test functions versus system behavior, explains that KUnit adds kunit_tool QEMU support and cross‑compile flags so tests run on real architectures, standardizes output with the new KTAP standard and a hardened parser, adds SKIP and run‑isolated modes, and reports numbers: test suites double from 20 to 40 and individual tests grow from 204 to 324, with tests built into the kernel or loadable as modules and working in CI.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:59Why (and How) Should You Test the Kemel?
- 4:22What test framework should I use?
- 7:24Other Testing Tools
- 9:29The Challenges Faced in 2021
- 11:38A Test Result Format
- 17:42Architectures and Tooling
- 18:55Architecture support
- 20:02Since Last Year's LCA
- 23:35New Tests!
- 25:32What's coming soon?
- 26:36What do you want?
related talks
44mKUnit - Unit Testing for the Linux Kernel
16mHow do we make KUnit work for us?
45mKUnit: Past, Present, and Future
24mKUnit Testing Insufficiencies- 45m"KUnit: Past, Present, and Future"
28mTesting Drivers with KUnit (Does hardware have to be hard?)
19mChecking your work: Linux kernel testing and CI
16mKselftests augmented: running kernelspace tests from userspace