[Mlir-commits] [mlir] [MLIR][Arith] Fix BitcastOp fold crashing on unhandled constant attributes (PR #212483)

[Tobias Gysi]

================
@@ -2270,16 +2270,23 @@ OpFoldResult arith::BitcastOp::fold(FoldAdaptor adaptor) {
     return ub::PoisonAttr::get(getContext());
 
   /// Bitcast integer or float to integer or float.
-  APInt bits = llvm::isa<FloatAttr>(operand)
-                   ? llvm::cast<FloatAttr>(operand).getValue().bitcastToAPInt()
-                   : llvm::cast<IntegerAttr>(operand).getValue();
+  APInt bits;
+  if (auto floatAttr = dyn_cast<FloatAttr>(operand))
+    bits = floatAttr.getValue().bitcastToAPInt();
+  else if (auto intAttr = dyn_cast<IntegerAttr>(operand))
+    bits = intAttr.getValue();
+  else
+    return {};
+
   assert(resType.getIntOrFloatBitWidth() == bits.getBitWidth() &&
          "trying to fold on broken IR: operands have incompatible types");
 
   if (auto resFloatType = dyn_cast<FloatType>(resType))
     return FloatAttr::get(resType,
                           APFloat(resFloatType.getFloatSemantics(), bits));
-  return IntegerAttr::get(resType, bits);
+  if (auto resIntType = dyn_cast<IntegerType>(resType))
----------------
gysit wrote:

nit: I would probably drop this change since the `getIntOrFloatBitWidth` asserts if the type is not an integer or float. The check here would thus be too late.


https://github.com/llvm/llvm-project/pull/212483