0-Day is a validation infrastructure to automate Linux kernel performance regression testing that immediately detects, isolates, and reports issues to help improve Linux on IA quality.
Here’s how it works: 0-Day monitors the mainline kernel and hundreds of Linux developer trees around the clock. As soon as a change is made, 0-Day pulls the patch set and immediately tests several builds, progressing through boot, functional, and finally power and performance tests.
When a test fails or shows performance regression, 0-Day automatically pinpoints the specific patch that caused the regression. The key is Code Bisection, a method to identify changes that cause specific behavior. 0-Day bisects patch sets and retests the code sections until it pinpoints the exact problem patch, and then sends a defect report to the developer who wrote the code, often in less than a day. 0-Day is unique in that it not only reports the failure, but also the specific patch that caused the failure, and the information to reproduce the problem.
0-Day (and its predecessor, the Linux Kernel Performance Project) essentially shifts left formal regression testing to the developers, where it’s faster, easier, and less costly to fix problems.