I'm not a specialist at all of these questions but I would say that kernel modifications are not worth taking such a big risk. And if they wanted to avoid publi

I'm not a specialist at all of these questions but I would say that kernel modifications are not worth taking such a big risk. And if they wanted to avoid publishing the code of some parts of their kernel implementation maybe they could instead make a kernel module and in this case I believe it is accepted of not releasing the source code of the module. But in any case as mentionned in other comments I believe the interesting code is more high level, or server side.