Explore the complex world of Linux kernel testing in this 41-minute conference talk from linux.conf.au 2020. Dive into the challenges of maintaining a rapidly growing operating system that runs on diverse hardware and performs numerous functions. Learn about various automated testing projects, including per-patch CI, kernel selftests, performance testing, regression testing, and hardware testing across different scales. Discover the push for more unit testing and the importance of avoiding code duplication. Gain insights into the kernel development process, including mailing list interactions, automated replies, and the role of maintainers. Understand the significance of projects like Linux Next and various testing frameworks. Examine workflow problems, the need for a common test result format, and the importance of publishing test results. Ultimately, grasp the ongoing pursuit of a more efficient and bug-free Linux kernel development ecosystem.
Overview
Syllabus
Introduction
The problem of scale
Maintaining errs
Kernel versions
Whats the fault
Hardware
Userspace
The matrix
What are we testing
Testing for regression
We have to compromise
Lets break things down
How much is someone tested
What can a developer do
Selftests
Unit tests
Code on the mailing list
The first email you get
The attack of the robots
Automated replies
What robots do
Support the robot uprising
I wrote that
So the code has satisfied
What happens now
The persona of a reviewer
What is a review of run
The maintainer
Linux Next
Kernel Testing
The Problem
CoCo
Whats happening
Release candidates
Will someone use it
LTP
Fuzzing
Distribution
Bugs
Shifting responsibility
Workflow problems
No cheating
No OKO
Takeaways
What do we do
What happens next
Maintaining
A common test result format
Kernel Coorg
Publishing test results
QA
Taught by
linux.conf.au
