| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
Avoid tests being killed due to OOM on Linux if a system is configured with 'ulimit -s unlimited' by skipping tests relying on infinite recursion. While unclear if Python should support 'ulimit -s unlimited', we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set. Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
Sorry, something went wrong.
Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
There was a problem hiding this comment.
This LGTM. Thanks!
Sorry, something went wrong.
There was a problem hiding this comment.
LGTM
Sorry, something went wrong.
Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
|
Thanks @Thyre for the PR, and @Fidget-Spinner for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14. |
Sorry, something went wrong.
…pythonGH-143606) Avoid tests being killed due to OOM on Linux if a system is configured with 'ulimit -s unlimited' by skipping tests relying on infinite recursion. While unclear if Python should support 'ulimit -s unlimited', we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set. (cherry picked from commit 61e0366) Co-authored-by: Jan André Reuter <jan.andre.reuter@hotmail.de> Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
|
GH-143619 is a backport of this pull request to the 3.14 branch. |
Sorry, something went wrong.
GH-143606) (#143619) gh-143460: Skip infinite recusion tests for infinite stack size (GH-143606) Avoid tests being killed due to OOM on Linux if a system is configured with 'ulimit -s unlimited' by skipping tests relying on infinite recursion. While unclear if Python should support 'ulimit -s unlimited', we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set. (cherry picked from commit 61e0366) Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de> Co-authored-by: Jan André Reuter <jan.andre.reuter@hotmail.de>
…python#143606) Avoid tests being killed due to OOM on Linux if a system is configured with 'ulimit -s unlimited' by skipping tests relying on infinite recursion. While unclear if Python should support 'ulimit -s unlimited', we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set. Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
…python#143606) Avoid tests being killed due to OOM on Linux if a system is configured with 'ulimit -s unlimited' by skipping tests relying on infinite recursion. While unclear if Python should support 'ulimit -s unlimited', we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set. Signed-off-by: Jan André Reuter <j.reuter@fz-juelich.de>
| Back | FazBrowse Home | New Git URL |
Avoid tests being killed due to OOM on Linux if a system is configured with ulimit -s unlimited by skipping tests relying on infinite recursion via infinite_recursion with a high passed number.
While unclear if Python should support ulimit -s unlimited, we should at least try to avoid failing a PGO build running tests due to an unlimited stack size being set.
Note: While this lets make pass on the systems I've tested on (with PGO enabled), make test will fail with ulimit -s unlimited.
On a system with sufficient amount of memory, I saw several of these errors:
Traceback (most recent call last): File "/dev/shm/reuter1/jupiter/Python/3.14.2/GCCcore-15.2.0/Python-3.14.2/Lib/test/test_dictviews.py", line 286, in test_deeply_nested_repr self.assertRaises(RecursionError, repr, d) ~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^ AssertionError: RecursionError not raised by reprwhereas on my personal machine, I saw some tests with typical recursion (not using infinite_recursion) still filling up the memory due to the machine having substantially less RAM. There were also two or three timeouts, but without triggering OOM.
Based on #143460 (comment), I'd consider this acceptable, but I can also do another round of going through the tests to exclude any test that could yield issues with ulimit -s unlimited.