[PATCH] [AArch64] Handle weak references to variables in small model when using the static relocation model

[Renato Golin]

On 10 September 2014 11:15, Asiri Rathnayake <asiri.rathnayake at arm.com> wrote:
> I've made Phab review here: http://reviews.llvm.org/D5285

Thanks! I'll do the review there...


> Hmmm, I count exactly 80 columns here.

Ha! Of course! I was editing the diff, which had the extra + at the
beginning... :)


> I think you are referring to the semantics of the assembly instruction (how
> the CPU interprets the instruction), whereas the assembly mnemonics are
> setup this way. I've checked with the old ARM64 backend (+ armcc) and this
> is the correct way to setup the operation.

Yes, I was. It did sound right, I just wanted to make sure there were
no assumptions here.



> My understanding of the AArch64 backend (and llvm for that matter) is little
> limited, but this is how the old ARM64 backend used to load weak references
> in the absence of a GOT. This is a regression from the backend merge.

That's ok, I think the code is correct, and the "problem", as you say,
is so remote that is almost a non-problem.

Maybe a short comment describing the range would be good measure.

cheers,
--renato