On Tue, Dec 07, 2021 at 08:10:02AM -0800, Linus Torvalds wrote:

Can you really not understand my dislike of a data type that is
fundamentally a lazy shortcut and intentionally hides error cases with
leaks?

I see the distinction; I just don't see it as hiding. refcount_t will
very much generate a splat indicating things need fixing, the leak
simply ensures that it doesn't get worse from there on out.

Anyway, if the code already needs to handle that failure case, such as
inc_not_zero() usage, then there is indeed no additional complexity.