[Mlir-commits] [mlir] [mlir][arith] Fix integer-to-float comparison canonicalization (PR #233631)

[Jakub Kuderski]

================
@@ -2849,7 +2876,8 @@ class CmpFIntToFPConst final : public OpRewritePattern<CmpFOp> {
       if (exponent == APFloat::IEK_Inf) {
         int maxExponent = ilogb(APFloat::getLargest(rhs.getSemantics()));
         if (maxExponent < (int)valueBits) {
-          // Conversion could create infinity.
+          // Conversion could create infinity. For example, sitofp(65536 : i32)
----------------
kuhar wrote:

also here

https://github.com/llvm/llvm-project/pull/233631