[llvm-dev] Atomic LL/SC loops in llvm

Tue May 10 13:53:53 PDT 2016

On 10 May 2016 at 13:50, Matthias Braun <mbraun at apple.com> wrote:
> One possibility would be to present the sequence as an instruction bundle. That will avoid any instructions/spills/reloads getting moved in the middle of it. Which on the other hand doesn't sound that different from just expand a pseudo instruction...

I wondered about that, but can bundles span multiple basic blocks (or,
I suppose the real question is, contain terminators)? It might also
conflict with the targets' own use of bundles. But certainly worth
thinking about more.

Tim.