[llvm] [X86][CodeGen] Emit float128 libcalls for math functions (PR #79611)

[Pranav Kant via llvm-commits]

pranavk wrote:

Also worth noting that x86 codegen for fp128 in clang right now seems completely broken because LLVM lowers to fFOOl variants without proper casting.

https://godbolt.org/z/drP5Wrr47

https://github.com/llvm/llvm-project/pull/79611