Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] #21361

aperezlebel · 2021-10-18T14:54:01Z

Describe the bug

I followed the contributing code guidelines and ran into a segmentation fault when running the test sklearn/tests/test_common.py. It appears to come from libsvm.fit(...).

I am not sure whether it is a problem related to my device or not.

Steps/Code to Reproduce

git clone https://github.com/scikit-learn/scikit-learn.git
cd scikit-learn
python3 -m venv venv
source venv/bin/activate
pip install pip --upgrade
pip install numpy scipy matplotlib pytest sphinx cython ipykernel
pip install --no-build-isolation --editable .
pytest sklearn/tests/test_common.py

Expected Results

No error is thrown.

Actual Results

Fatal Python error: Segmentation fault

Current thread 0x00007fe406b46280 (most recent call first):
  File "/home/aperez/dev/sandbox/scikit-learn/sklearn/svm/_base.py", line 315 in _dense_fit
  File "/home/aperez/dev/sandbox/scikit-learn/sklearn/svm/_base.py", line 255 in fit
  File "/home/aperez/dev/sandbox/scikit-learn/sklearn/utils/estimator_checks.py", line 2091 in check_classifiers_train
  File "/home/aperez/dev/sandbox/scikit-learn/sklearn/utils/_testing.py", line 313 in wrapper
  File "/home/aperez/dev/sandbox/scikit-learn/sklearn/tests/test_common.py", line 109 in test_estimators
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/python.py", line 183 in pytest_pyfunc_call
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_callers.py", line 39 in _multicall
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_manager.py", line 80 in _hookexec
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_hooks.py", line 265 in __call__
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/python.py", line 1641 in runtest
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 162 in pytest_runtest_call
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_callers.py", line 39 in _multicall
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_manager.py", line 80 in _hookexec
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_hooks.py", line 265 in __call__
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 255 in <lambda>
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 311 in from_call
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 254 in call_runtest_hook
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 215 in call_and_report
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 126 in runtestprotocol
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/runner.py", line 109 in pytest_runtest_protocol
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_callers.py", line 39 in _multicall
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_manager.py", line 80 in _hookexec
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_hooks.py", line 265 in __call__
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/main.py", line 348 in pytest_runtestloop
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_callers.py", line 39 in _multicall
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_manager.py", line 80 in _hookexec
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_hooks.py", line 265 in __call__
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/main.py", line 323 in _main
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/main.py", line 269 in wrap_session
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/main.py", line 316 in pytest_cmdline_main
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_callers.py", line 39 in _multicall
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_manager.py", line 80 in _hookexec
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/pluggy/_hooks.py", line 265 in __call__
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/config/__init__.py", line 162 in main
  File "/home/aperez/dev/sandbox/scikit-learn/venv/lib/python3.8/site-packages/_pytest/config/__init__.py", line 185 in console_main
  File "/home/aperez/dev/sandbox/scikit-learn/venv/bin/pytest", line 8 in <module>
Erreur de segmentation (core dumped)

Versions

System:
    python: 3.8.10 (default, May 27 2021, 17:54:13)  [GCC 9.3.0]
executable: /home/aperez/dev/sandbox/scikit-learn/venv/bin/python
   machine: Linux-5.13.0-1014-oem-x86_64-with-glibc2.29

Python dependencies:
          pip: 21.3
   setuptools: 56.0.0
      sklearn: 1.1.dev0
        numpy: 1.21.2
        scipy: 1.7.1
       Cython: 0.29.24
       pandas: None
   matplotlib: 3.4.3
       joblib: 1.1.0
threadpoolctl: 3.0.0

Built with OpenMP: True

The text was updated successfully, but these errors were encountered:

ogrisel · 2021-10-18T16:19:43Z

Thanks for the report. Do you happen to know which specific test is causing the segfault?

For instance is the following enough to reproduce?

pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

You can then refine the test name pattern passed to -k to find out the culprit.

ogrisel · 2021-10-18T16:21:22Z

For the record, I cannot reproduce locally with the same versions of numpy and scipy and our CI is green.

Which compiler version are you using? Assuming you are using the system's compiler on Linux, the following should do:

gcc --version

aperezlebel · 2021-10-18T17:12:01Z

For instance is the following enough to reproduce?
pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

Yes, it reproduces the same error:

sklearn/tests/test_common.py::test_estimators[LinearSVC()-check_classifiers_train] PASSED                                                           [ 11%]
sklearn/tests/test_common.py::test_estimators[LinearSVC()-check_classifiers_train(readonly_memmap=True)] PASSED                                     [ 22%]
sklearn/tests/test_common.py::test_estimators[LinearSVC()-check_classifiers_train(readonly_memmap=True,X_dtype=float32)] PASSED                     [ 33%]
sklearn/tests/test_common.py::test_estimators[NuSVC()-check_classifiers_train] PASSED                                                               [ 44%]
sklearn/tests/test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] Fatal Python error: Segmentation fault

[... same traceback]

It seems to come from test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)].

Which compiler version are you using?

gcc --version gives me:

gcc (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0

jeremiedbb · 2021-10-19T08:14:42Z

I have the exact same version of gcc and I don't reproduce the failure locally

ogrisel · 2021-10-19T09:14:38Z

@alexrz can you please try to install clang / clang++ from the llvm project (either with apt install from ubuntu or using conda-forge) and rebuild scikit-learn with this this compiler using the following command:

CC="clang" CXX="clang++" LDSHARED="clang -shared" make clean inplace
pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

aperezlebel · 2021-10-19T09:50:19Z

After running commands described in "Steps/Code to Reproduce", I ran:

sudo apt install clang
CC="clang" CXX="clang++" LDSHARED="clang -shared" make clean inplace
pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

and got the same error on the same test.

ogrisel · 2021-10-19T14:56:55Z

This might be the same as #20430 and #21336.

Next step would be to try to reproduce with gdb and then output a post mortem backtrace.

ogrisel · 2021-10-19T14:58:18Z

For instance:

gdb -ex r --args python -m pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

and then, type bt to get a backtrace of the C-level call stack at the time of the crash.

aperezlebel · 2021-10-19T15:09:50Z

I got:

(gdb) bt
#0  0x00007fffdbe31c31 in ddot_k_PRESCOTT () from /home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/scipy/spatial/../../scipy.libs/libopenblasp-r0-085ca80a.3.9.so
#1  0x00007fffc893b39a in ?? () from /home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/scipy/linalg/cython_blas.cpython-38-x86_64-linux-gnu.so
#2  0x00007fffc89226e0 in ?? () from /home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/scipy/linalg/cython_blas.cpython-38-x86_64-linux-gnu.so
#3  0x00007fffbe547aca in __pyx_fuse_1__pyx_f_7sklearn_5utils_12_cython_blas__dot (__pyx_v_n=<optimized out>, __pyx_v_x=<optimized out>, __pyx_v_incx=<optimized out>, __pyx_v_y=<optimized out>, 
    __pyx_v_incy=<optimized out>) at sklearn/utils/_cython_blas.c:2861
#4  0x00007fffbe258327 in svm::Kernel::Kernel (this=0x7fffffff6030, l=<optimized out>, x_=<optimized out>, param=..., blas_functions=0x7fffffff64e0) at sklearn/svm/src/libsvm/svm.cpp:394
#5  0x00007fffbe25b394 in svm::SVC_Q::SVC_Q (blas_functions=0x7fffffff64e0, y_=0x555556ff0fd0 '\001' <repeats 100 times>, '\377' <repeats 100 times>..., param=..., prob=..., this=0x7fffffff6030)
    at sklearn/svm/src/libsvm/svm.cpp:1682
#6  svm::solve_nu_svc (blas_functions=0x7fffffff64e0, si=0x7fffffff6000, alpha=0x555557450160, param=0x7fffffff69a0, prob=0x7fffffff6360) at sklearn/svm/src/libsvm/svm.cpp:1682
#7  svm::svm_train_one (prob=0x7fffffff6360, param=0x7fffffff69a0, Cp=<optimized out>, Cn=<optimized out>, status=0x7fffffff64d4, blas_functions=0x7fffffff64e0)
    at sklearn/svm/src/libsvm/svm.cpp:1856
#8  0x00007fffbe264cde in svm_train (prob=0x7fffffff6340, prob@entry=0x7fffffff6500, param=param@entry=0x7fffffff69a0, status=status@entry=0x7fffffff64d4, 
    blas_functions=blas_functions@entry=0x7fffffff64e0) at sklearn/svm/src/libsvm/svm.cpp:2504
#9  0x00007fffbe23e921 in __pyx_pf_7sklearn_3svm_7_libsvm_fit (__pyx_v_X=__pyx_v_X@entry=0x7fffbb041db0, __pyx_v_Y=__pyx_v_Y@entry=0x7fffbb041c90, __pyx_v_svm_type=__pyx_v_svm_type@entry=1, 
    __pyx_v_kernel=__pyx_v_kernel@entry=0x7fffc3052730, __pyx_v_degree=__pyx_v_degree@entry=3, __pyx_v_gamma=__pyx_v_gamma@entry=0.53188777537536391, __pyx_v_coef0=__pyx_v_coef0@entry=0, 
    __pyx_v_tol=__pyx_v_tol@entry=0.001, __pyx_v_C=__pyx_v_C@entry=0, __pyx_v_nu=0.5, __pyx_v_epsilon=0, __pyx_v_class_weight=__pyx_v_class_weight@entry=0x7fffbb041f30, 
    __pyx_v_sample_weight=0x7fffbb041e10, __pyx_v_shrinking=1, __pyx_v_probability=0, __pyx_v_cache_size=200, __pyx_v_max_iter=-1, __pyx_v_random_seed=209652396, __pyx_self=<optimized out>)

ogrisel · 2021-10-19T15:36:19Z

OpenBLAS 0.3.9 seems very old.

Can you please include the output of:

python -m threadpoolctl -i sklearn

in the environment where you can reproduce the crash?

aperezlebel · 2021-10-19T15:45:48Z

python -m threadpoolctl -i sklearn

[
  {
    "user_api": "openmp",
    "internal_api": "openmp",
    "prefix": "libgomp",
    "filepath": "/home/aperez/mambaforge/envs/xxx/lib/libgomp.so.1.0.0",
    "version": null,
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/numpy.libs/libopenblasp-r0-2d23e62b.3.17.so",
    "version": "0.3.17",
    "threading_layer": "pthreads",
    "architecture": "SkylakeX",
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/scipy.libs/libopenblasp-r0-085ca80a.3.9.so",
    "version": "0.3.9",
    "threading_layer": "pthreads",
    "architecture": "Prescott",
    "num_threads": 8
  }
]

aperezlebel · 2021-10-19T15:56:45Z

When installing packages with condaforge, I don't get the error.

python -m threadpoolctl -i sklearn

[
  {
    "user_api": "openmp",
    "internal_api": "openmp",
    "prefix": "libgomp",
    "filepath": "/home/aperez/mambaforge/envs/scikit-learn/lib/libgomp.so.1.0.0",
    "version": null,
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/home/aperez/mambaforge/envs/scikit-learn/lib/libopenblasp-r0.3.18.so",
    "version": "0.3.18",
    "threading_layer": "pthreads",
    "architecture": "SkylakeX",
    "num_threads": 8
  }
]

ogrisel · 2021-10-19T16:02:50Z

It's possible that openblas 0.3.9 on CPU architectures detected as Prescott has a bug with readonly memory buffer that causes the segfault.

ogrisel · 2021-10-19T16:05:47Z

@AlexPrz maybe you can try to do:

OPENBLAS_CORETYPE="Haswell" pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

and also:

OPENBLAS_CORETYPE="Haswell" python -m threadpoolctl -i sklearn

in the venv that has the problem.

aperezlebel · 2021-10-19T16:12:49Z

When running:

OPENBLAS_CORETYPE="Haswell" pytest -v sklearn/tests/test_common.py -k "check_classifiers_train and svc"

in the venv that has the problem, I don't get the error anymore.

And OPENBLAS_CORETYPE="Haswell" python -m threadpoolctl -i sklearn returns:

[
  {
    "user_api": "openmp",
    "internal_api": "openmp",
    "prefix": "libgomp",
    "filepath": "/home/aperez/mambaforge/envs/scikit-learn/lib/libgomp.so.1.0.0",
    "version": null,
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/numpy.libs/libopenblasp-r0-2d23e62b.3.17.so",
    "version": "0.3.17",
    "threading_layer": "pthreads",
    "architecture": "Haswell",
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/home/aperez/dev/sandbox/test/scikit-learn/venv/lib/python3.8/site-packages/scipy.libs/libopenblasp-r0-085ca80a.3.9.so",
    "version": "0.3.9",
    "threading_layer": "pthreads",
    "architecture": "Haswell",
    "num_threads": 8
  }
]

jeremiedbb · 2021-10-19T16:15:32Z

On my side, I confirm that forcing the Prescott architecture triggers the segfault.

ogrisel · 2021-10-19T16:31:31Z

I think the only fix is to wait for scipy to upgrade it's dependency to openblas.

ogrisel · 2021-10-19T16:44:41Z

I reported the bug upstream in scipy.

The workaround, in the mean time, is to install scipy from conda-forge instead of pypi.org.

conda create -n cf-env -c conda-forge numpy scipy cython
conda activate cf-env
pip install --no-build-isolation --editable .

amueller · 2021-10-23T17:27:31Z

So should this no longer be a blocker?

jjerphan · 2021-11-16T12:11:21Z

scipy/scipy#14316 seems to have fixed this issue indirectly by upgrading OpenBLAS for the PyPI distribution of scipy:

/tmp/scikit-learn   ± ● main $ source venv/bin/activate
/tmp/scikit-learn   ± ● main $ where python            
/tmp/scikit-learn/venv/bin/python
/usr/bin/python
/tmp/scikit-learn   ± ● main $ python -m threadpoolctl -i sklearn
[
  {
    "user_api": "openmp",
    "internal_api": "openmp",
    "prefix": "libomp",
    "filepath": "/home/jjerphan/.local/share/miniconda3/lib/libomp.so",
    "version": null,
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/tmp/scikit-learn/venv/lib/python3.9/site-packages/numpy.libs/libopenblasp-r0-2d23e62b.3.17.so",
    "version": "0.3.17",
    "threading_layer": "pthreads",
    "architecture": "SkylakeX",
    "num_threads": 8
  },
  {
    "user_api": "blas",
    "internal_api": "openblas",
    "prefix": "libopenblas",
    "filepath": "/tmp/scikit-learn/venv/lib/python3.9/site-packages/scipy.libs/libopenblasp-r0-8b9e111f.3.17.so",
    "version": "0.3.17",
    "threading_layer": "pthreads",
    "architecture": "SkylakeX",
    "num_threads": 8
  }
]

@AlexPrz: can you confirm?

aperezlebel · 2021-11-16T13:42:42Z

I do confirm that I don't get the segfault anymore with scipy 1.7.2 whereas I get it with scipy 1.7.1.

ogrisel · 2021-11-18T08:57:09Z

For information we observed this crash on the CI on the unrelated PR #21697. The CI entry with the crash is also using the old version of openblas: 0.3.13 that detects the Prescott kernel.

threadpoolctl info:
       filepath: /usr/share/miniconda/envs/testvenv/lib/libgomp.so.1.0.0
         prefix: libgomp
       user_api: openmp
   internal_api: openmp
        version: None
    num_threads: 2

       filepath: /usr/share/miniconda/envs/testvenv/lib/libopenblasp-r0.3.13.so
         prefix: libopenblas
       user_api: blas
   internal_api: openblas
        version: 0.3.13
    num_threads: 2
threading_layer: pthreads
   architecture: Prescott

We should probably update our 6D40 CI tests to XFAIL or skip those tests on the Prescott architecture.

jjerphan · 2021-11-18T09:41:26Z

As discussed with @glemaitre, I am taking this as I already did some openblas xfail fixtures for faulty tests.

Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

* TST XFAIL test when unstable openblas configuration * Adapt versions parsing Co-authored-By: Adrin Jalali <adrin.jalali@gmail.com> * DOC Prefer ifskip and reference #21361 Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Fix Julien's scheduler incorrect dispatch * Simplify Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Do not remove blank line * fixup! Do not remove blank line * Retrigger CI * Prudently assume Prescott might be the architecture if it is unknown Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> Co-authored-by: Adrin Jalali <adrin.jalali@gmail.com> Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

…rn#21702) * TST XFAIL test when unstable openblas configuration * Adapt versions parsing Co-authored-By: Adrin Jalali <adrin.jalali@gmail.com> * DOC Prefer ifskip and reference scikit-learn#21361 Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Fix Julien's scheduler incorrect dispatch * Simplify Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Do not remove blank line * fixup! Do not remove blank line * Retrigger CI * Prudently assume Prescott might be the architecture if it is unknown Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> Co-authored-by: Adrin Jalali <adrin.jalali@gmail.com> Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

* TST XFAIL test when unstable openblas configuration * Adapt versions parsing Co-authored-By: Adrin Jalali <adrin.jalali@gmail.com> * DOC Prefer ifskip and reference #21361 Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Fix Julien's scheduler incorrect dispatch * Simplify Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> * Do not remove blank line * fixup! Do not remove blank line * Retrigger CI * Prudently assume Prescott might be the architecture if it is unknown Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org> Co-authored-by: Adrin Jalali <adrin.jalali@gmail.com> Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

aperezlebel added the Bug: triage label Oct 18, 2021

ogrisel added Blocker Bug segfault and removed Bug: triage labels Oct 18, 2021

ogrisel changed the title ~~Segmentation fault when running test_common.py~~ Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] Oct 19, 2021

ogrisel mentioned this issue Oct 19, 2021

BUG: openblas 0.3.9 shipped with the scipy 1.7.1 wheel can cause segfaults when it fails to detect the correct CPU architecture (Prescott instead o Haswell) scipy/scipy#14886

Closed

glemaitre mentioned this issue Oct 25, 2021

Ridge with cholesky solver returns NaN #21447

Open

jeremiedbb removed the Blocker label Oct 25, 2021

glemaitre mentioned this issue Nov 17, 2021

TST check the legend instead of label names in CalibrationDisplay #21697

Merged

jjerphan mentioned this issue Nov 18, 2021

TST Skip test when unstable openblas configuration #21702

Merged

jjerphan added a commit to jjerphan/scikit-learn that referenced this issue Nov 18, 2021

DOC Prefer ifskip and reference scikit-learn#21361

4c8374e

Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

jjerphan added a commit that referenced this issue Nov 18, 2021

DOC Prefer ifskip and reference #21361

247c2db

Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

jjerphan added a commit to jjerphan/scikit-learn that referenced this issue Nov 18, 2021

DOC Prefer ifskip and reference scikit-learn#21361

5ca918a

Co-authored-by: Olivier Grisel <olivier.grisel@ensta.org>

This was referenced Nov 18, 2021

[MRG] FIX segmentation fault on memory mapped contiguous memoryview #21654

Merged

ddot_k_PRESCOTT segfaults when called on non-memory aligned data OpenMathLib/OpenBLAS#3453

Closed

adrinjalali closed this as completed in #21702 Nov 18, 2021

This was referenced Jul 25, 2022

⚠️ CI failed on Linux_nogil.pylatest_pip_nogil ⚠️ #23786

Closed

TST Make sure memmap are aligned when OpenBLAS detects Prescott architecture #23994

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] #21361

Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] #21361

Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] #21361

Segmentation fault when running test_common.py::test_estimators[NuSVC()-check_classifiers_train(readonly_memmap=True)] #21361

Comments

Describe the bug

Steps/Code to Reproduce

Expected Results

Actual Results

Versions