[llvm] [X86] EmitCmp - always use cmpw with foldable loads (PR #92251)

[Phoebe Wang via llvm-commits]

================
@@ -22694,9 +22694,10 @@ static SDValue EmitCmp(SDValue Op0, SDValue Op1, unsigned X86CC,
   // Only promote the compare up to I32 if it is a 16 bit operation
   // with an immediate.  16 bit immediates are to be avoided.
----------------
phoebewang wrote:

Update comments?

https://github.com/llvm/llvm-project/pull/92251