[clang] [llvm] [WIP][RISCV] Support for Zvabd fast-track proposal (PR #124239)

[Craig Topper via llvm-commits]

================
@@ -575,15 +575,16 @@ multiclass VALU_IV_X<string opcodestr, bits<6> funct6> {
            SchedBinaryMC<"WriteVIALUX", "ReadVIALUV", "ReadVIALUX">;
 }
 
-multiclass VALU_IV_I<string opcodestr, bits<6> funct6> {
-  def I  : VALUVI<funct6, opcodestr # ".vi">,
-           SchedUnaryMC<"WriteVIALUI", "ReadVIALUV">;
+multiclass VALU_IV_I<string opcodestr, bits<6> funct6, Operand optype = simm5> {
+  def I : VALUVI<funct6, opcodestr#".vi", optype>,
+          SchedUnaryMC<"WriteVIALUI", "ReadVIALUV">;
 }
 
-multiclass VALU_IV_V_X_I<string opcodestr, bits<6> funct6>
+multiclass VALU_IV_V_X_I<string opcodestr, bits<6> funct6,
----------------
topperc wrote:

These classes don't appear to be referenced elsewhere in the patch. Why do they need to be chaned?

https://github.com/llvm/llvm-project/pull/124239