Describe the issue:

Hi,

For a research paper, we carried out a large-scale benchmark of Pynguin, an Automatic Unit Test Generation Tool for Python, to test its new feature that can find Python interpreter crashes. In this benchmark, we found a potential bug in numpy, and we are reporting this issue.

The example was run with the command python -X faulthandler <test> to dump the traceback.

Reproduce the code example:

import numpy.polynomial.legendre as legendre

legendre.legder(("0", "0"), scl=20)

Error message:

Fatal Python error: Segmentation fault

Current thread 0x00007f3e28300bc0 [python] (most recent call first):
  Garbage-collecting
  <no Python frame>

Current thread's C stack trace (most recent call first):
  Binary file "/lib64/libpython3.14.so.1.0", at _Py_DumpStack+0x4c [0x7f3e27d213fa]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x121e7e [0x7f3e27d21e7e]
  Binary file "/lib64/libc.so.6", at +0x1a2c0 [0x7f3e27a262c0]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x1facd8 [0x7f3e27dfacd8]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x183bf9 [0x7f3e27d83bf9]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x1fab68 [0x7f3e27dfab68]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x1fa12a [0x7f3e27dfa12a]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x25ed69 [0x7f3e27e5ed69]
  Binary file "/lib64/libpython3.14.so.1.0", at +0x27fcd1 [0x7f3e27e7fcd1]
  Binary file "/lib64/libpython3.14.so.1.0", at Py_RunMain+0x231 [0x7f3e27e52401]
  Binary file "/lib64/libpython3.14.so.1.0", at Py_BytesMain+0x3b [0x7f3e27e4c75b]
  Binary file "/lib64/libc.so.6", at +0x35b5 [0x7f3e27a0f5b5]
  Binary file "/lib64/libc.so.6", at __libc_start_main+0x88 [0x7f3e27a0f668]
  Binary file "python", at _start+0x25 [0x55c7863453d5]

Extension modules: numpy._core._multiarray_umath, numpy.linalg._umath_linalg (total: 2)
Segmentation fault         (core dumped)

Python and NumPy Versions:

2.4.0
3.14.0 (main, Oct 17 2025, 00:00:00) [GCC 15.2.1 20251022 (Red Hat 15.2.1-3)]

Runtime Environment:

[{'numpy_version': '2.4.0',
'python': '3.14.0 (main, Oct 17 2025, 00:00:00) [GCC 15.2.1 20251022 (Red '
'Hat 15.2.1-3)]',
'uname': uname_result(system='Linux', node='lucas-hp', release='6.17.10-300.fc43.x86_64', version='#1 SMP PREEMPT_DYNAMIC Mon Dec 1 14:59:36 UTC 2025', machine='x86_64')},
{'simd_extensions': {'baseline': ['X86_V2'],
'found': ['X86_V3'],
'not_found': ['X86_V4', 'AVX512_ICL', 'AVX512_SPR']}},
{'ignore_floating_point_errors_in_matmul': False},
{'architecture': 'Haswell',
'filepath': '/home/lucas/Documents/crashes/.venv/lib/python3.14/site-packages/numpy.libs/libscipy_openblas64_-fdde5778.so',
'internal_api': 'openblas',
'num_threads': 12,
'prefix': 'libscipy_openblas',
'threading_layer': 'pthreads',
'user_api': 'blas',
'version': '0.3.30'}]

How does this issue affect you or how did you find it:

It's not a serious problem, but we thought it was important to at least warn you that this behaviour exists, so that you can take the action that suits you best.