On the licensing of the kernel

This proposal has been rejected.


One Line Summary

You thought you knew the licensing of the kernel? You don't! Join me to find out why.


Together with stats, I will present some interesting and surprising facts on the licenses found in a Kernel. Is there 500+ ways to state a file is under the GPL? Check! Some weird licenses? Check! Witty comments and easter eggs? Check! Some GPLv3? Check! Proprietary-like licenses? Check!

How did I find out about this? I maintain the scancode-toolkit which is a new FLOSS tool to accurately detect licenses, copyrights and other interesting origin and licensing clues in sources and binaries.

I will present how the tool works and can reasonably perform an exhaustive pair-wise multi-diff between the ~45K to 50K files of a kernel and a set of ~5000 license texts and notices to accurately pinpoint all the licenses found in a kernel. And how this is possible without doing 100 millions of diffs that would otherwise take days or weeks to complete.

And I will then present a summary of the fun, weird and sometimes conflicted licenses found the a kernel together with extensive stats.

Finally I will make some suggestions on how the licensing documentation of the kernel could be improved and made simpler for maintainers and contributors using simple conventions, the help of the SPDX licenses list and the help of the scanner that I will demonstrate.


legal, licensing, GPL


  • Phil2

    Philippe Ombredanne

    AboutCode.org and nexB Inc.


    I am an avid open source hacker, a contributor to strace, a co-founder of SPDX at the Linux Foundation, an active mentor to the Google summer of Code, contributor to Python packaging tools and several other.