[Mlir-commits] [mlir] [mlir][arith] Add rounding mode flags to binary arithmetic operations (PR #188458)

[Renato Golin]

================
@@ -157,6 +157,11 @@ class AttrConverterConstrainedFPToLLVM {
       convertedAttr.set(TargetOp::getRoundingModeAttrName(),
                         convertArithRoundingModeAttrToLLVM(arithAttr));
     }
+    // Constrained intrinsics do not support fastmath flags. Remove the
+    // arith fastmath attribute if present.
+    if constexpr (SourceOp::template hasTrait<
----------------
rengolin wrote:

It's not clear what `constrained` means here, so it's not clear what's the lowering requirement. Perhaps just note what you said above into the comment would help.

https://github.com/llvm/llvm-project/pull/188458