[Bug tree-optimization/63537] [4.8/4.9/5 Regression] Missed optimization: Loop unrolling adds extra copy when returning aggregate

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63537

--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #1)
> Eventually a simple pass could handle
> 
>   <retval> = ret;
>   ret ={v} {CLOBBER};
>   return <retval>;
> 
> and back-propagate <retval> into all stores/loads of ret.

Shouldn't tree-nrv.c already handle this, except that it currently bails out if
TREE_ADDRESSABLE (found)? (strangely enough it has (dead) code to handle the
addressable case further in the same function)