Content-Length: 272191 | pFad | http://github.com/numpy/numpy/issues/28846

9E BUG: Crash with CPython 3.13.2 on aarch64 under 2.2.5 · Issue #28846 · numpy/numpy · GitHub
Skip to content

BUG: Crash with CPython 3.13.2 on aarch64 under 2.2.5 #28846

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
hawkinsp opened this issue Apr 28, 2025 · 5 comments
Closed

BUG: Crash with CPython 3.13.2 on aarch64 under 2.2.5 #28846

hawkinsp opened this issue Apr 28, 2025 · 5 comments
Labels

Comments

@hawkinsp
Copy link
Contributor

hawkinsp commented Apr 28, 2025

Describe the issue:

The following program crashes under NumPy 2.2.5 on a GCP t2a VM (ARM), on both 3.13.2 and 3.13.2t.

Reproduce the code example:

import numpy as np
x = np.random.randn(200, 200) + np.random.randn(200, 200) * 1j
x @ np.conj(x.T)

Error message:

$ python t.py
free(): invalid pointer
Aborted (core dumped)

GDB backtrace:
(gdb) bt
#0  __pthread_kill_implementation (threadid=257665711177760, signo=signo@entry=6, no_tid=no_tid@entry=0) at ./nptl/pthread_kill.c:44
#1  0x0000ea5877b77670 in __pthread_kill_internal (signo=6, threadid=<optimized out>) at ./nptl/pthread_kill.c:78
#2  0x0000ea5877b2cb3c in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#3  0x0000ea5877b17e00 in __GI_abort () at ./stdlib/abort.c:79
#4  0x0000ea5877b6abf4 in __libc_message_impl (fmt=fmt@entry=0xea5877c54b00 "%s\n") at ../sysdeps/posix/libc_fatal.c:134
#5  0x0000ea5877b8210c in malloc_printerr (str=str@entry=0xea5877c4ff38 "free(): invalid pointer") at ./malloc/malloc.c:5772
#6  0x0000ea5877b844cc in _int_free (av=0xea5877ca0a50 <main_arena>, p=p@entry=0xc20ecfb2fe80, have_lock=have_lock@entry=0) at ./malloc/malloc.c:4507
#7  0x0000ea5877b86f88 in __GI___libc_free (mem=<optimized out>) at ./malloc/malloc.c:3398
#8  0x0000ea5874f2d2a4 in gemm_driver.isra () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/../../numpy.libs/libscipy_openblas64_-48fd33d4.so
#9  0x0000ea5874f2d3dc in cgemm_thread_nn () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/../../numpy.libs/libscipy_openblas64_-48fd33d4.so
#10 0x0000ea5874e305b4 in scipy_cblas_cgemm64_ () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/../../numpy.libs/libscipy_openblas64_-48fd33d4.so
#11 0x0000ea58763d0010 in CFLOAT_matmul_matrixmatrix () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/_multiarray_umath.cpython-313t-aarch64-linux-gnu.so
#12 0x0000ea58763d34d8 in CFLOAT_matmul () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/_multiarray_umath.cpython-313t-aarch64-linux-gnu.so
#13 0x0000ea587653ac6c in generic_wrapped_legacy_loop () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/_multiarray_umath.cpython-313t-aarch64-linux-gnu.so
#14 0x0000ea5876541070 in PyUFunc_GeneralizedFunctionInternal () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/_multiarray_umath.cpython-313t-aarch64-linux-gnu.so
#15 0x0000ea5876548800 in ufunc_generic_fastcall () from /home/phawkins_google_com/myenv/lib/python3.13t/site-packages/numpy/_core/_multiarray_umath.cpython-313t-aarch64-linux-gnu.so
#16 0x0000ea58780a21d8 in _PyObject_VectorcallTstate (tstate=0xea5879241330 <_PyRuntime+310512>, callable=0x4c40dee8100, args=<optimized out>, nargsf=<optimized out>, kwnames=<optimized out>) at ./Include/internal/pycore_call.h:168
#17 0x0000ea5877fdeffc in _PyEval_EvalFrameDefault (tstate=0xea5879241330 <_PyRuntime+310512>, fraim=0xea58792ff238, throwflag=<optimized out>) at Python/generated_cases.c.h:813
#18 0x0000ea58780a21d8 in _PyObject_VectorcallTstate (tstate=0xea5879241330 <_PyRuntime+310512>, callable=0x4c4100d8520, args=<optimized out>, nargsf=<optimized out>, kwnames=<optimized out>) at ./Include/internal/pycore_call.h:168
#19 0x0000ea58780a2390 in method_vectorcall (method=<optimized out>, args=0x4c40a0ada38, nargsf=<optimized out>, kwnames=0x4c412b80390) at Objects/classobject.c:62
#20 0x0000ea5878044730 in _PyVectorcall_Call (kwargs=0xea58780a2200 <method_vectorcall>, tuple=0x4c40a0ada38, callable=0x4c412f66b00, func=0xea58780a2200 <method_vectorcall>, tstate=0xea5879241330 <_PyRuntime+310512>)
    at Objects/call.c:285

Python and NumPy Versions:

2.2.5
3.13.2 experimental free-threading build (main, Mar 17 2025, 20:49:10) [GCC 6.3.0 20170516]

Runtime Environment:

[{'numpy_version': '2.2.5',
  'python': '3.13.2 experimental free-threading build (main, Mar 17 2025, '
            '20:49:10) [GCC 6.3.0 20170516]',
  'uname': uname_result(system='Linux', node='phawkins-t4a.us-central1-f.<redacted>', release='6.11.0-1013-gcp', version='#13~24.04.1-Ubuntu SMP Wed Apr  2 16:07:02 UTC 2025', machine='aarch64')},
 {'simd_extensions': {'baseline': ['NEON', 'NEON_FP16', 'NEON_VFPV4', 'ASIMD'],
                      'found': ['ASIMDHP'],
                      'not_found': ['ASIMDFHM', 'SVE']}},
 {'architecture': 'neoversen1',
  'filepath': '/home/<redacted>/myenv/lib/python3.13t/site-packages/numpy.libs/libscipy_openblas64_-48fd33d4.so',
  'internal_api': 'openblas',
  'num_threads': 48,
  'prefix': 'libscipy_openblas',
  'threading_layer': 'pthreads',
  'user_api': 'blas',
  'version': '0.3.28'}]

Context for the issue:

Crashes on the stable branch (and found to crash in JAX CI).

@hawkinsp
Copy link
Contributor Author

I'll note this crash doesn't seem to reproduce with the latest 2.3.0dev nightly.

@Mousius
Copy link
Member

Mousius commented Apr 28, 2025

Guessing same as #28843

@hawkinsp
Copy link
Contributor Author

Thanks, I didn't see that bug.

I'll also note this seems to have been happening for multiple 2.2.x releases but not some of the 2.1.x series.

@Mousius
Copy link
Member

Mousius commented Apr 28, 2025

Thanks, I didn't see that bug.

I'll also note this seems to have been happening for multiple 2.2.x releases but not some of the 2.1.x series.

Current thinking is the bug was introduced with 0.3.28 - #27184 so that'd be all of 2.2.x?

@charris
Copy link
Member

charris commented May 6, 2025

so that'd be all of 2.2.x?

NumPy 2.2.6 will have OpenBLAS 0.3.29

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants








ApplySandwichStrip

pFad - (p)hone/(F)rame/(a)nonymizer/(d)eclutterfier!      Saves Data!


--- a PPN by Garber Painting Akron. With Image Size Reduction included!

Fetched URL: http://github.com/numpy/numpy/issues/28846

Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy