An end to implicit fall-throughs in the kernel

Furthermore, often switch-case is used to have multiple values land execution at the same statement. The syntax for this, i.e. consecutive case labels with no statements in between, is indistinguishable from unannounced fall-through.

The underlying problem here is that switch-case is, with seeming regularity, learned as "what C has for select-case in Visual Basic". That, like many common assumptions regarding C-family languages, is very wrong.

Thirdly, the compiler-oriented solution implies that review will actually compile submitted code and has both care and cojones to require warnings fixed. If that's the case, nice; but I'm not even convinced that any of these "fix fallthru warning" patches' cases were hit in a specific test first to determine that fallthru is actually the correct behaviour. Done wrong, that could be worse than changing nothing.