A tale from one of the busier places (by patch churn at least) in the kernel, and how to make it all work well. Since about one year the drm/i915 kernel driver is maintained by two maintainers and a team of roughly 15 people who can apply patches directly to the main driver branch – fairly normal in open source, but rather unheard of in the Linux kernel. It all started as an experiment, with the expectation that worst case I’ll burry it quietly again. Which hasn’t happened, at least yet.

I’ll talk about why I considered trying this out, what I feared could happen, what’s actually happened and what I think, with a bit of hindsight, are the key ingredients to make this work, and where it could be useful model in other places.