On Fri, Jun 19, 2020 at 01:08:53PM +0200, Bastien Nocera wrote: > On Fri, 2020-06-19 at 12:53 +0200, Hans de Goede wrote: > > A note for future reference, not sure what you mean with driver > > > > API here. If you mean the in kernel API, the kernel rules are > > > > that we are always free to change that (Linux does not have a > > > > stable driver API). > > > > > > > > So if a header does not sit under include/uapi (indicating that > > > > it is an userspace API) then a change like this is fine. > > I meant the internal driver API, which might break out-of-tree drivers. There is no such thing. Well, there might be, but we don't care and have to act as if there are no such thing otherwise we would never be able to change anything :) > I know that this API is fluid, and that there are no stability > guarantees, but I'd still expect at least a note in the commit message > to that effect. Why? Who cares? Anyone who lives outside of the kernel already knows they have to dig in the kernel if their code breaks, that's the cost of living outside of the kernel. thanks, greg k-h
