On Tue, 11 Dec 2018 20:13:21 +1100 Andrew Donnellan <andrew.donnellan@xxxxxxxxxxx> wrote: > On 11/12/18 7:57 pm, Greg Kurz wrote: > > I now realize that I should have mentioned the real motivation for this > > change. I'm working on refactoring the code so that we can use ocxl in a > > KVM guest. The concept of link can be shared by both powernv and pseries > > variants but the SPA is definitely a powernv only thingy. The benefit > > of this patch is hence to kick 'struct link' out of free_spa() so that > > it can be utimately moved to powernv specific code. > > > > The initial version of this change was just moving the link->spa check > > and link->spa = NULL to the callers, where it was quite obvious they're > > not needed... Should I re-post this as two patches for clarity ? > > Ah, that makes much more sense. > > If that's the case, then why not go all the way and change the function > signature while you're at it? > Sure, will do.