[llvm] [InstCombine] Fold xored one-complemented operand comparisons (PR #69882)

[Yingwei Zheng via llvm-commits]

================
@@ -4,6 +4,118 @@
 declare void @llvm.assume(i1)
 declare void @barrier()
 
+define i32 @test_slt_xor(i32 %0, i32 %1) {
----------------
dtcxzyw wrote:

`zext` is unnecessary.
```
define i1 @test_slt_xor(i32 %0, i32 %1) {
  %3 = xor i32 %0, -1
  %4 = xor i32 %3, %1
  %5 = icmp slt i32 %4, %3
  ret i1 %5
}
```

https://github.com/llvm/llvm-project/pull/69882