[llvm] [llvm][aarch64] Apple A16 & A17 had adrp-add fusion, but A14 did not (PR #81325)

[David Green via llvm-commits]

davemgreen wrote:

> I was assuming Apple `ld`/`ld-prime`, and only thinking about the effect of this on instruction scheduling / macro-fusion + knowledge of the respective uarch's.
> 
> Can you tell me more about how that linker relaxation works? Is it an optimization that keys off of `FeatureAdrpAdd` itself, or is it merely enabled by placing the `adrp` next to the `add`?

As far as I understand inside the linker it is done via adjacent instructions. It doesn't have much other information, other than the relocations. I think so long as the linker used will most likely be apple ld using LoH's, then it's fine to be more accurate with the fusion and not "mis-use" the feature to get adjacent instructions.

> > > Changes like this should ideally be accompanied by a test.
> > 
> > 
> > I'll add one.
> 
> [ffab5a0](https://github.com/llvm/llvm-project/commit/ffab5a089b1e94b3305fbdfdf1547b751121c090)

Thanks!

https://github.com/llvm/llvm-project/pull/81325