[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

STINNER Victor  added the comment:


New changeset 531f2ebd60a662111c78107935d249d3d52f9a4f by Miss Islington (bot) 
in branch '3.9':
bpo-43406: Fix test_signal.test_stress_modifying_handlers() (GH-24815) 
(GH-24817)
https://github.com/python/cpython/commit/531f2ebd60a662111c78107935d249d3d52f9a4f


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

miss-islington  added the comment:


New changeset ac5e23c540f5ec20fc390c72e097e0fdaf8b7ac9 by Miss Islington (bot) 
in branch '3.8':
bpo-43406: Fix test_signal.test_stress_modifying_handlers() (GH-24815)
https://github.com/python/cpython/commit/ac5e23c540f5ec20fc390c72e097e0fdaf8b7ac9


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

Change by STINNER Victor :


--
resolution:  -> fixed
stage: patch review -> resolved
status: open -> closed

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

Change by miss-islington :


--
pull_requests: +23582
pull_request: https://github.com/python/cpython/pull/24816

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

Change by miss-islington :


--
pull_requests: +23583
pull_request: https://github.com/python/cpython/pull/24817

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

STINNER Victor  added the comment:


New changeset 1fa17e8cc62775a2e34b158135ce8589f9394f03 by Victor Stinner in 
branch 'master':
bpo-43406: Fix test_signal.test_stress_modifying_handlers() (GH-24815)
https://github.com/python/cpython/commit/1fa17e8cc62775a2e34b158135ce8589f9394f03


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

STINNER Victor  added the comment:

> The new functional test has two race conditions.

I don't think that it's a bug in Python, but more race conditions in the test.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

Change by STINNER Victor :


--
pull_requests: +23581
stage: resolved -> patch review
pull_request: https://github.com/python/cpython/pull/24815

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[STINNER Victor]

STINNER Victor  added the comment:

The change added a new functional test. As I expected, it failed on "AMD64 
Debian root". I expected it because the previously added functional test failed 
failed on "AMD64 Debian root", I reported the race condition in 2017 and it's 
still not fixed: bpo-30849.

The new functional test has two race conditions.

I reopen the issue.


== Race condition 1 ==

AMD64 Debian root 3.x:
https://buildbot.python.org/all/#/builders/345/builds/903

0:15:05 load avg: 2.58 [137/427/1] test_signal failed (env changed) (1 min 13 
sec)
Warning -- Unraisable exception
Traceback (most recent call last):
  File "/root/buildarea/3.x.angelico-debian-amd64/build/Lib/threading.py", line 
1067, in _wait_for_tstate_lock
elif lock.acquire(block, timeout):
OSError: Signal 10 ignored due to race condition


== Race condition 2 ==

By the way, the test also fails when I stress my laptop:

$ ./python -m test --fail-env-changed test_signal -F -j40 -m 
test_stress_modifying_handlers -v
== CPython 3.10.0a6+ (heads/master:a9c03d7fb7, Mar 10 2021, 12:41:26) [GCC 
10.2.1 20201125 (Red Hat 10.2.1-9)]
== Linux-5.10.15-200.fc33.x86_64-x86_64-with-glibc2.32 little-endian
== cwd: /home/vstinner/python/master/build/test_python_223315æ
== CPU count: 8
== encodings: locale=UTF-8, FS=utf-8
0:00:00 load avg: 4.17 Run tests in parallel using 40 child processes
(...)
0:00:05 load avg: 7.52 [  5/1] test_signal failed
test_stress_modifying_handlers (test.test_signal.StressTest) ... FAIL

==
FAIL: test_stress_modifying_handlers (test.test_signal.StressTest)
--
Traceback (most recent call last):
  File "/home/vstinner/python/master/Lib/test/test_signal.py", line 1297, in 
test_stress_modifying_handlers
self.assertGreater(num_received_signals, 0)
AssertionError: 0 not greater than 0

--

Ran 1 test in 0.039s

FAILED (failures=1)
test test_signal failed
(...)

--
nosy: +vstinner
resolution: fixed -> 
status: closed -> open

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Change by Antoine Pitrou :


--
pull_requests: +23546
pull_request: https://github.com/python/cpython/pull/24755

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Change by Antoine Pitrou :


--
resolution:  -> fixed
stage: patch review -> resolved
status: open -> closed

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

miss-islington  added the comment:


New changeset 1385f8355a036fd65aaf9c7e7ab48467ca922bcf by Miss Islington (bot) 
in branch '3.9':
[3.9] bpo-43406: Fix possible race condition where ``PyErr_CheckSignals`` tries 
to execute a non-Python signal handler (GH-24756) (GH-24761)
https://github.com/python/cpython/commit/1385f8355a036fd65aaf9c7e7ab48467ca922bcf


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

miss-islington  added the comment:


New changeset 4715be8a4384159e47fb09e5f3bd077734842655 by Antoine Pitrou in 
branch '3.8':
[3.8] bpo-43406: Fix possible race condition where ``PyErr_CheckSignals`` tries 
to execute a non-Python signal handler (GH-24756) (GH-24762)
https://github.com/python/cpython/commit/4715be8a4384159e47fb09e5f3bd077734842655


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Change by Antoine Pitrou :


--
pull_requests: +23534
pull_request: https://github.com/python/cpython/pull/24762

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Antoine Pitrou  added the comment:


New changeset 68245b7a1030287294c65c298975ab9026543fd2 by Antoine Pitrou in 
branch 'master':
bpo-43406: Fix possible race condition where ``PyErr_CheckSignals`` tries to 
execute a non-Python signal handler (GH-24756)
https://github.com/python/cpython/commit/68245b7a1030287294c65c298975ab9026543fd2


--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[miss-islington]

Change by miss-islington :


--
nosy: +miss-islington
nosy_count: 2.0 -> 3.0
pull_requests: +23533
pull_request: https://github.com/python/cpython/pull/24761

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Change by Antoine Pitrou :


--
nosy: +neologix

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Change by Antoine Pitrou :


--
keywords: +patch
pull_requests: +23527
stage: needs patch -> patch review
pull_request: https://github.com/python/cpython/pull/24756

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

Antoine Pitrou  added the comment:

Here is a reproducer:
https://gist.github.com/pitrou/e5a566e644730516b51de71145c5ea06

If you execute it, it will fail after a few iterations:

sig 2
sig 2
sig 2
Traceback (most recent call last):
  File "/home/antoine/cpython/default/setinterrupt.py", line 30, in 
main()
  File "/home/antoine/cpython/default/setinterrupt.py", line 27, in main
cycle_handlers(signum)
  File "/home/antoine/cpython/default/setinterrupt.py", line 19, in 
cycle_handlers
signal.signal(signum, handler)
  File "/home/antoine/cpython/default/Lib/signal.py", line 48, in signal
return _int_to_enum(handler, Handlers)
  File "/home/antoine/cpython/default/Lib/signal.py", line 30, in _int_to_enum
return enum_klass(value)
  File "/home/antoine/cpython/default/Lib/enum.py", line 606, in __call__
return cls.__new__(cls, value)
  File "/home/antoine/cpython/default/Lib/enum.py", line 927, in __new__
ve_exc = ValueError("%r is not a valid %s" % (value, cls.__qualname__))
TypeError: 'int' object is not callable

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue43406] Possible race condition between signal catching and signal.signal

[Antoine Pitrou]

New submission from Antoine Pitrou :

We can receive signals (at the C level, in trip_signal() in signalmodule.c) 
while signal.signal is being called to modify the corresponding handler.  Later 
when PyErr_CheckSignals() is called to handle the given signal, the handler may 
be a non-callable object and will raise a cryptic asynchronous exception.

--
components: Interpreter Core, Library (Lib)
messages: 388131
nosy: pitrou
priority: normal
severity: normal
stage: needs patch
status: open
title: Possible race condition between signal catching and signal.signal
type: behavior
versions: Python 3.10, Python 3.8, Python 3.9

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com