Reading LWN I don't get the impression that the kernel is immune to "Your patch works and provides a feature some people might like, but I don’t like it, go away" -- even big corporate contributors sometimes wind up in year long efforts to try to get something they and others find useful upstream if the upstream maintainer is unconvinced about the worth of the feature or that it won't have net downsides for the codebase as a whole. The "extensible scheduler" is one prominent recent example.
