As announced today at the Open Source Summit in Lyon (see the official press release), the KernelCI project which powers kernelci.org with automated testing for the upstream Linux kernel has now joined the Linux Foundation. This is the new home the project has finally found after sailing through uncharted waters for over five years.
What does it actually mean for the project?
Given the huge scale at which the Linux kernel is being used, achieving comprehensive test coverage for it is an incredibly challenging goal. Based on the open source philosophy principles, KernelCI's distributed architecture makes this possible by enabling the whole kernel community to collaborate around a single upstream CI system. Becoming part of the Linux Foundation will let the project flourish and become in turn integral part of the Linux kernel development workflow.
Some actual results of this move can already be seen with the new common database for storing test results from KernelCI and other related projects that share a common goal such as Red Hat's CKI and Linaro's LKFT. It is an experimental step towards expanding KernelCI in a modular way to interact with other existing systems. There are as many different ways to test the kernel as there are use-cases for it, and many types of specialised systems to cover: a CPU architecture such as 0-day, a Linux distribution such as CKI, a range of reference platforms such as LKFT...
What happens next?
This is only a new beginning, many things are yet to come with many decisions to be made: how to use the budget available from the membership scheme, how to make the infrastructure more sustainable and scalable, how to compromise and address the needs of all the members joining the project... Answers to all these questions are likely to appear as the coming months and years unfold.
One thing we can be sure of is that there is no reason for the current development plans to stop or be impacted in any way - on the contrary. There is indeed a strong need to extend test coverage and the capabilities of KernelCI at large, with a huge potential to improve upstream kernel development as a direct result. Becoming part of the Linux Foundation should be about facilitating progress in this direction above all.
What does Collabora plan to do?
Collabora has been involved with KernelCI since almost the beginning, providing some of the servers running the kernelci.org services and a large LAVA test lab. We've also become the biggest contributor to KernelCI development with myself as de facto maintainer for the core repositories on Github and in charge of the weekly kernelci.org updates.
Among all the things we have done in the last couple of years, these are probably the most significant ones:
Enabled automated bisection for boot regressions
Added v4l2-compliance test suite and a few more minor ones (suspend/resume...)
Enabled "Depthcharge" bootloader in LAVA to run KernelCI tests on Chromebooks
Portable command-line tools and YAML-based configuration to perform all the KernelCI steps in any environment
Going forward, here are some clear items on the project's current roadmap which will carry on full steam ahead with a growing team of developers:
Further improve handling of test results: web interface, email reports, regression tracking
Deploy new bisection tool to cover complex test suite results
Improve modular core architecture to enable inter-operability between the various elements that form the overall KernelCI pipeline
Then like with every open source project, other contributors will also add more features, enable new platforms in the global test hardware pool or bring new ideas. The combined efforts of all parties is what makes the project unique and mature enought to reach its ultimate goal. We will make sure we continue to be among the top active players within the growing KernelCI community, as a founding member of the Linux Foundation project as well as with myself being both a maintainer and on the technical steering committee.