BUG: Add self-overlap check and tests

Previsouly we unnecessarily required contiguous inputs for
1-D "trivially iterable".  That requirement was unnecessary, but
also guarded against certain self-overlaps.

This was not visible because a self-overlapping array is rarely
aligned (but for complex this happens typically).

2 files changed, 11 insertions, 0 deletions

diff --git a/numpy/core/src/umath/ufunc_object.c b/numpy/core/src/umath/ufunc_object.c
index bf8bd1a8c..b0654d7b0 100644
--- a/numpy/core/src/umath/ufunc_object.c
+++ b/numpy/core/src/umath/ufunc_object.c
@@ -1212,6 +1212,12 @@ try_trivial_single_output_loop(PyUFuncObject *ufunc,
                 return -2;
             }
         }
+        /* Check self-overlap (non 1-D are contiguous, perfect overlap is OK) */
+        if (operation_ndim == 1 &&
+                PyArray_STRIDES(op[nin])[0] < PyArray_ITEMSIZE(op[nin]) &&
+                PyArray_STRIDES(op[nin])[0] != 0) {
+            return -2;
+        }
     }
 
     /* Call the __prepare_array__ if necessary */
diff --git a/numpy/core/tests/test_mem_overlap.py b/numpy/core/tests/test_mem_overlap.py
index ba73dad62..24bdf477f 100644
--- a/numpy/core/tests/test_mem_overlap.py
+++ b/numpy/core/tests/test_mem_overlap.py
@@ -666,6 +666,11 @@ class TestUFunc:
     def test_unary_ufunc_call_fuzz(self):
         self.check_unary_fuzz(np.invert, None, np.int16)
 
+    @pytest.mark.slow
+    def test_unary_ufunc_call_complex_fuzz(self):
+        # Complex typically has a smaller alignment than itemsize
+        self.check_unary_fuzz(np.negative, None, np.complex128, count=500)
+
     def test_binary_ufunc_accumulate_fuzz(self):
         def get_out_axis_size(a, b, axis):
             if axis is None: