[llvm] [AMDGPU][True16][MC] true16 for v_alignbit_b32 (PR #119409)

[Ivan Kosarev via llvm-commits]

================
@@ -2506,6 +2507,42 @@ def : GCNPat<(i32 (trunc (srl i64:$src0, (and i32:$src1, (i32 31))))),
 def : GCNPat<(i32 (trunc (srl i64:$src0, (i32 ShiftAmt32Imm:$src1)))),
           (V_ALIGNBIT_B32_e64 (i32 (EXTRACT_SUBREG (i64 $src0), sub1)),
                           (i32 (EXTRACT_SUBREG (i64 $src0), sub0)), $src1)>;
+} //end True16Predicate = NotHasTrue16BitInsts
+
+let True16Predicate = UseFakeTrue16Insts in {
+def ROTRPattern_fake16 : GCNPat <
----------------
kosarev wrote:

Do we need the `ROTRPattern_fake16` bit? Seems obvious that it's a rotr pattern? (I guess it's supposed to work as a reference to the `ROTRPattern` class, but then it's only used once, so probably should be removed as well.)

https://github.com/llvm/llvm-project/pull/119409