[llvm] [LLVM][AArch64] Correctly lower funnel shifts by zero. (PR #140058)

[Simon Pilgrim via llvm-commits]

RKSimon wrote:

If it wasn't constant and a later combine manages to fold the node to a constant then it might not be regenerated/combined and technically you could see it at lowering - DAG combines still aren't properly topologically ordered :(

But I'm not entirely clear what you're trying to handle here - does arm64 only support FSHR nodes with constant shift amounts? What happens with zero / modulo amounts with that instruction that is so bad?

https://github.com/llvm/llvm-project/pull/140058