[llvm] [AArch64] Optimize two large shifts and a combine into a single combine and shift (PR #99480)

[David Green via llvm-commits]

https://github.com/davemgreen approved this pull request.

The code LGTM, thanks. The combines can see all kinds of types and operands, but I think with VLSHR nodes it should constrained to only things that are legal for the uzps.

https://github.com/llvm/llvm-project/pull/99480