[PATCH] D14489: [AArch64] Applying load pair optimization for volatile load/store

[Tim Northover via llvm-commits]

On 15 November 2015 at 17:45, Junmo Park <junmoz.park at samsung.com> wrote:
> If we can't merge two volatile load, I think this commit is not worth for non-volatile and volatile load merge. How do you think about this?

I agree. There are just so many warnings about messing with this in
the ARM ARM that I wouldn't be happy even if we could guarantee we
were implementing exactly what I believed was the correct
interpretation in LLVM.

Tim.