[PATCH] D11437: Correct x86_64 fp128 mangled name and return/varargs types

[Reid Kleckner]

rnk added a comment.

We really should not use the general LLVM IR type conversion machinery to decide how we are classifying our arguments. There are existing instances of us doing this, but we should strive to eliminate them. I think the right approach is probably to check `getLongDoubleFormat()`, see if it is `APFloat::IEEEquad` or `APFloat::x86DoubleExtended`, and pick memory, sse, or x87 classifications based on that.


http://reviews.llvm.org/D11437