[PATCH]: Fix latent bug in tailcall

[Daniel Berlin]

Diego Novillo wrote:
> Daniel Berlin wrote:
> 
>> However, due to laziness, etc, they simply call update_ssa, which is an
>> incredibly large hammer for what they are doing, and does a ton more work.
>>
> Yes, of course.  But this is not news and we discussed it quite
> thoroughly in the past.
> 
> I thought this was about the invalid SSA left after tailcall?  The
> over-reliance on update_ssa is something we still need to address.
> 
It started out that way, and i said "There is a way to do this without
calling update_ssa, and i tried it, ....."

and that is what started this tangent.