[llvm] [AArch64] Use isKnownNonZero to optimize eligible compares to cmn (PR #96349)

[David Green via llvm-commits]

================
@@ -258,3 +258,34 @@ define i32 @neg_range_int(i32 %a, i32 %b, i32 %c) {
   ret i32 %retval.0
 }
 
+; (b > -3 || a < -(c | 1))
+define i32 @neg_range_int_cmn(i32 %a, i32 %b, i32 %c) {
----------------
davemgreen wrote:

I think this case (the ccmn one) is still incorrect, unfortunately.

https://github.com/llvm/llvm-project/pull/96349