[clang] [llvm] [CLANG][LLVM][AArch64]Add SME2.1 intrinsics for MOVAZ tile to vector,… (PR #88499)

[Momchil Velikov via llvm-commits]

================
@@ -2832,6 +2832,23 @@ AArch64TargetLowering::EmitTileLoad(unsigned Opc, unsigned BaseReg,
   return BB;
 }
 
+MachineBasicBlock *
+AArch64TargetLowering::EmitTileMovaz(unsigned Opc, unsigned BaseReg,
----------------
momchil-velikov wrote:

This function looks almost identical to `EmitZAInstr`. It looks to me you can reuse `EmitZAInstr` (with a couple of small modifications) and then employ the `SMEPseudo2Instr<name, N>` technique.
Then you won't need the switch cases starting at line 3012.

https://github.com/llvm/llvm-project/pull/88499