RFC: C2 LRB self-fixup for testing to sh/jdk

[Aleksey Shipilev]

On 6/6/19 10:34 PM, Roman Kennke wrote:
> This looks good to me.
> 
> As discussed on IRC, we need to clarify (post-push) with Roland:
> - What's up with EA there. Why it fails without the change, and why the
> change fixes it, and if it's correct.

Yes, here:
  https://mail.openjdk.java.net/pipermail/shenandoah-dev/2019-June/009898.html

> - If we can ditch the whole implicit-null-checking fluff, because the
> load of mark-word is gone, which was -afaik- the whole reason why we
> needed it to begin with. That would be a nice simplification and one
> lesser pain in the ...

Right, that would be a follow-up.

-Aleksey