[llvm] [AArch64][GlobalISel] Avoid running the shl(zext(a), C) -> zext(shl(a, C)) combine. (PR #67045)

[Jon Roelofs via llvm-commits]

================
@@ -690,6 +690,10 @@ class AArch64TargetLowering : public TargetLowering {
   bool isDesirableToCommuteWithShift(const SDNode *N,
                                      CombineLevel Level) const override;
 
+  bool isDesirableToPullExtFromShl(const MachineInstr &MI) const override {
+    return false;
----------------
jroelofs wrote:

A quick comment on the "why" would be good. Could be just the second sentence from the commit message.

https://github.com/llvm/llvm-project/pull/67045