[PATCH] SimplifyCFG: turn recursive GatherConstantCompares into iterative

[Chad Rosier]

> As a side question, the existing code assumes that the "icmp" are always
> taking the constant as the second operand (it never tries to match the
> first operand). Is it something that is canonicalized in the IR and can be
> assumed here?

AFAIK, it's the preferred form in IR and I've seen a lot of code that make
this assumption.  However, you should investigate on a case by case basis.
 I'm not aware of any pass or verifier that strictly enforces this
standard across all optimizations/passes.

As an aside, the reassociate pass now canonicalizes constants of binary
operators to the RHS.  This doesn't include icmp instructions, however. 
Later passes may obliterate this canonicalization.

 HTH, Chad