RFC: PATCH to gimplify_modify_expr_rhs to make return slot explicit

[Diego Novillo]

Jason Merrill wrote:

> Any thoughts?
> 
Not an objection to the patch, but this will also cause TBAA to 
introduce more alias relations with the variable in the return slot. 
The aliasing code will not only see it as an addressable variable, but 
will also consider it call-clobbered.

I'm not sure how much of a problem this is, but it should be easy to 
measure by checking the .alias dumps before/after the patch.  We 
obviously need to make the aliaser smarter (law and dberlin are working 
on enhancements that should help, though they may not be ready for 4.0).


Diego.