[llvm] [RISCV] Undo fneg (fmul x, y) -> fmul x, (fneg y) transform (PR #157388)

[Piotr Fusik via llvm-commits]

pfusik wrote:

> This also causes some more fnegs to be sunk after a bitcast to integer, so they're now done as xor.

Why is that, if the transform is for a single use of fneg in fmul?

> From glancing at some of the schedules for WriteFSGN my guess is that this is also profitable.

It should be, especially with BINVI.

https://github.com/llvm/llvm-project/pull/157388