REGR: fillna on f32 column raising for f64

doc/source/whatsnew/v1.3.3.rst

@@ -26,6 +26,8 @@ Fixed regressions
 - Fixed regression in :func:`is_list_like` where objects with ``__iter__`` set to ``None`` would be identified as iterable (:issue:`43373`)
 - Fixed regression in :meth:`.Resampler.aggregate` when used after column selection would raise if ``func`` is a list of aggregation functions (:issue:`42905`)
 - Fixed regression in :meth:`DataFrame.corr` where Kendall correlation would produce incorrect results for columns with repeated values (:issue:`43401`)
+- Fixed regression in :meth:`Series.fillna` raising ``TypeError`` when filling ``float`` ``Series`` with list-like fill value having a dtype which couldn't cast lostlessly (like ``float32`` filled with ``float64``) (:issue:`43424`)
+-
 
 .. ---------------------------------------------------------------------------
 

pandas/core/array_algos/putmask.py

@@ -41,8 +41,14 @@ def putmask_inplace(values: ArrayLike, mask: np.ndarray, value: Any) -> None:
     if lib.is_scalar(value) and isinstance(values, np.ndarray):
         value = convert_scalar_for_putitemlike(value, values.dtype)
 
-    if not isinstance(values, np.ndarray) or (
-        values.dtype == object and not lib.is_scalar(value)
+    if (
+        not isinstance(values, np.ndarray)
+        or (values.dtype == object and not lib.is_scalar(value))
+        # GH#43424: np.putmask raises TypeError if we cannot cast between types with
+        # rule = "safe", a stricter guarantee we may not have here
+        or (
+            isinstance(value, np.ndarray) and not np.can_cast(value.dtype, values.dtype)
+        )
     ):
         # GH#19266 using np.putmask gives unexpected results with listlike value
         if is_list_like(value) and len(value) == len(values):

pandas/tests/series/methods/test_fillna.py

@@ -702,6 +702,23 @@ def test_fillna_categorical_raises(self):
         with pytest.raises(TypeError, match=msg):
             ser.fillna(DataFrame({1: ["a"], 3: ["b"]}))
 
+    @pytest.mark.parametrize("dtype", [float, "float32", "float64"])
+    @pytest.mark.parametrize("fill_type", tm.ALL_REAL_NUMPY_DTYPES)
+    def test_fillna_float_casting(self, dtype, fill_type):
+        # GH-43424
+        ser = Series([np.nan, 1.2], dtype=dtype)
+        fill_values = Series([2, 2], dtype=fill_type)
+        result = ser.fillna(fill_values)
+        expected = Series([2.0, 1.2], dtype=dtype)
+        tm.assert_series_equal(result, expected)
+
+    def test_fillna_f32_upcast_with_dict(self):
+        # GH-43424
+        ser = Series([np.nan, 1.2], dtype=np.float32)
+        result = ser.fillna({0: 1})
+        expected = Series([1.0, 1.2], dtype=np.float32)
+        tm.assert_series_equal(result, expected)
+
     # ---------------------------------------------------------------
     # Invalid Usages