[llvm] [AArch64] Avoid single-element vector fp converts in streaming[-compatible] functions (PR #112213)

[Sander de Smalen via llvm-commits]

================
@@ -6237,7 +6237,8 @@ def : Pat<(v2f64 (AArch64frsqrts (v2f64 FPR128:$Rn), (v2f64 FPR128:$Rm))),
 // Some float -> int -> float conversion patterns for which we want to keep the
 // int values in FP registers using the corresponding NEON instructions to
 // avoid more costly int <-> fp register transfers.
-let Predicates = [HasNEONandIsStreamingSafe] in {
+// TODO: Allow these in streaming[-compatible] functions with +sme2p2.
----------------
sdesmalen-arm wrote:

nit: could you replicate this comment on the other two uses too?

https://github.com/llvm/llvm-project/pull/112213