llvm-project

38cb7ddc - [VectorCombine] foldPermuteOfIntrinsic - support multiple uses of shuffled ops (#175299)