[llvm] [GlobalISel] Migrate remaining generic wip_match_opcode combines to MIR-pattern. (PR #222851)

[Yaxun Liu via llvm-commits]

================
@@ -776,25 +863,31 @@ def same_val_zero: GICombineRule<
   (apply (COPY $dst, (GITypeOf<"$x"> 0)))
 >;
 
-// Fold (0 op x) - > 0
+// Fold (0 op x) -> 0. Use matchConstantOp so splat-zero LHS matches.
----------------
yxsamliu wrote:

This now also matches splat-zero vectors and changes AArch64 output. Could you list and test this as a functional change, or move it to a separate patch?

https://github.com/llvm/llvm-project/pull/222851