gh-110012: Fix `RuntimeWarning` in `test_socket` #110013

sobolevn · 2023-09-28T06:48:38Z

After:

Ran 483 tests in 27.474s

OK (skipped=245)

== Tests result: SUCCESS ==

1 test OK.

Total duration: 27.6 sec
Total tests: run=483 skipped=245
Total test files: run=1/1
Result: SUCCESS

I haven't checked if it also generates this warning on other platforms (different from macos sonoma), but in case it is not, I will just change quiet to True

Issue: test_socket generates several warnings #110012

AlexWaygood

You mean "malformed" and "quiet", not "mailformed" and "quite" :-)

Lib/test/test_socket.py

sobolevn · 2023-09-28T06:54:29Z

@AlexWaygood I am really into tpyos, thanks! :)

AlexWaygood · 2023-09-28T06:57:04Z

Yuo're wellcom!

vstinner · 2023-09-28T11:14:25Z

Lib/test/test_socket.py

@@ -165,6 +166,16 @@ def socket_setdefaulttimeout(timeout):
        socket.setdefaulttimeout(old_timeout)


+@contextlib.contextmanager
+def catch_malformed_data_warning(quiet=True):


Does it catch them or ignore them? What is the behavior of quiet=False?

Yes

cpython/Lib/test/support/warnings_helper.py

Lines 152 to 188 in 8f324b7

def _filterwarnings(filters, quiet=False):

"""Catch the warnings, then check if all the expected

warnings have been raised and re-raise unexpected warnings.

If 'quiet' is True, only re-raise the unexpected warnings.

"""

# Clear the warning registry of the calling module

# in order to re-raise the warnings.

frame = sys._getframe(2)

registry = frame.f_globals.get('__warningregistry__')

if registry:

registry.clear()

with warnings.catch_warnings(record=True) as w:

# Set filter "always" to record all warnings. Because

# test_warnings swap the module, we need to look up in

# the sys.modules dictionary.

sys.modules['warnings'].simplefilter("always")

yield WarningsRecorder(w)

# Filter the recorded warnings

reraise = list(w)

missing = []

for msg, cat in filters:

seen = False

for w in reraise[:]:

warning = w.message

# Filter out the matching messages

if (re.match(msg, str(warning), re.I) and

issubclass(warning.__class__, cat)):

seen = True

reraise.remove(w)

if not seen and not quiet:

# This filter caught nothing

missing.append((msg, cat.__name__))

if reraise:

raise AssertionError("unhandled warning %s" % reraise[0])

if missing:

raise AssertionError("filter (%r, %s) did not catch any warning" %

missing[0])

It catches warnings with warnings.catch_warnings(record=True)

Then it filters out ones that matched ("received malformed or improperly-truncated ancillary data", RuntimeWarning)

If there are any left, the test fails

quiet=True handles the cases where no warning was raised - just by ignoring the last check, if quiet=False we are required to have at least one matched warning

serhiy-storchaka

Is there a correlation between a warning and the emptiness of ancdata?

vstinner · 2023-09-28T13:34:43Z

"received malformed or improperly-truncated ancillary data" message comes from Python if get_cmsg_data_len() returns non-zero. Can you please check in which case your are?

static int 
get_cmsg_data_len(struct msghdr *msg, struct cmsghdr *cmsgh, size_t *data_len)
{
    size_t space, cmsg_data_len;

    if (!cmsg_min_space(msg, cmsgh, CMSG_LEN(0)) ||
        cmsgh->cmsg_len < CMSG_LEN(0))
        return -1; 
    cmsg_data_len = cmsgh->cmsg_len - CMSG_LEN(0);
    if (!get_cmsg_data_space(msg, cmsgh, &space))
        return -1;
    if (space >= cmsg_data_len) {
        *data_len = cmsg_data_len;
        return 0;
    }
    *data_len = space;
    return 1;
}

sobolevn · 2023-09-28T13:52:45Z

Local debug shows that I face

printf("3");
*data_len = space;
return 1;

sobolevn · 2023-09-28T13:58:15Z

Hm, I agree that this does not look like a proper fix. There's something else going on inside, for some reason there's not enough space to write the data.

serhiy-storchaka · 2023-09-28T14:21:41Z

The code was added in 2011 (bpo-6560, #50809, 96fe56a). It does not emit a warning on Linux (at least I never see it). Were the warnings always on Windows, or did they appear in some version? Are they always expected in these tests, do they signal a problem? Shouldn't these tests fail if there was not a warning on Windows?

sobolevn · 2023-09-28T14:49:06Z

We should ask @ncoghlan for help, I guess :)

serhiy-storchaka · 2023-09-28T14:54:00Z

@sobolevn Could you test whether warnings were present in old Python versions (it was 3.2 or 3.3 I guess). Since most core developers do not use Windows, it might just slip past their notice.

sobolevn · 2023-09-28T15:09:40Z

I cannot build anything below 3.7 on mac sonora (m2) :(

3.7:

~/Desktop/cpython2  0f56adb8d7 ✗                                                          
» ./python.exe -m test test_socket                         
0:00:00 load avg: 5.09 Run tests sequentially
0:00:00 load avg: 5.09 [1/1] test_socket
/Users/sobolev/Desktop/cpython2/Lib/test/test_socket.py:2323: RuntimeWarning: received malformed or improperly-truncated ancillary data
  result = sock.recvmsg(bufsize, *args)
/Users/sobolev/Desktop/cpython2/Lib/test/test_socket.py:2414: RuntimeWarning: received malformed or improperly-truncated ancillary data
  result = sock.recvmsg_into([buf], *args)

== Tests result: SUCCESS ==

1 test OK.

Total duration: 24.0 sec
Tests result: SUCCESS

3.8:

~/Desktop/cpython2  3.8 ✔                                                                 
» ./python.exe -m test test_socket                         
0:00:00 load avg: 6.11 Run tests sequentially
0:00:00 load avg: 6.11 [1/1] test_socket
/Users/sobolev/Desktop/cpython2/Lib/test/test_socket.py:2438: RuntimeWarning: received malformed or improperly-truncated ancillary data
  result = sock.recvmsg(bufsize, *args)
/Users/sobolev/Desktop/cpython2/Lib/test/test_socket.py:2529: RuntimeWarning: received malformed or improperly-truncated ancillary data
  result = sock.recvmsg_into([buf], *args)

== Tests result: SUCCESS ==

1 test OK.

Total duration: 23.2 sec
Tests result: SUCCESS

So, this is not new at least.

vstinner · 2023-09-28T15:32:41Z

@serhiy-storchaka: @sobolevn is trying to make warnings quiet on macOS, not on Windows. I don't know if these warnings are emitted on other platforms than macOS.

AlexWaygood · 2023-09-28T15:42:38Z

python -We -m test test_socket passes for me on Windows on a reasonably fresh build of Python 3.13, FWIW

sobolevn · 2023-09-28T15:47:44Z

@vstinner they appear both on Win and Mac, see my first commit:

ubuntu (failing): https://github.com/python/cpython/actions/runs/6335548712/job/17206922402
windows passing: https://github.com/python/cpython/actions/runs/6335548712/job/17206922528

vstinner · 2023-09-28T16:51:46Z

@vstinner they appear both on Win and Mac, see my first commit

Oh ok.

hugovk · 2025-01-16T08:09:50Z

@AlexWaygood @serhiy-storchaka @vstinner Any further comments on this PR?

serhiy-storchaka · 2025-01-16T08:37:35Z

I do not think that simply suppressing warnings here is the right solution. What is the warning about? What is the end user supposed to do with it? Non-zero result of get_cmsg_data_len() should be either an error or ignored (note that there are two non-zero returned values). If such situation is common, then perhaps the code just does not handle all cases correctly.

graingert · 2025-01-16T09:48:53Z

What we can do is set the warning filter to default for these warnings, add a TODO comment and not close this issue

AlexWaygood · 2025-01-16T11:38:22Z

@AlexWaygood @serhiy-storchaka @vstinner Any further comments on this PR?

I only ever suggested some typo fixes on this one; it's well out of my area of expertise :-)

graingert · 2025-01-16T17:02:58Z

What we can do is set the warning filter to default for these warnings, add a TODO comment and not close this issue

I've now done this in the warnings as errors PR

pythongh-110012: Fix RuntimeWarning in test_socket

0d28ddd

sobolevn requested a review from vstinner September 28, 2023 06:48

bedevere-app bot added awaiting review tests Tests in the Lib/test dir labels Sep 28, 2023

bedevere-app bot mentioned this pull request Sep 28, 2023

test_socket generates several warnings #110012

Open

sobolevn added the skip news label Sep 28, 2023

AlexWaygood reviewed Sep 28, 2023

View reviewed changes

Lib/test/test_socket.py Outdated Show resolved Hide resolved

Lib/test/test_socket.py Outdated Show resolved Hide resolved

Lib/test/test_socket.py Outdated Show resolved Hide resolved

Lib/test/test_socket.py Outdated Show resolved Hide resolved

AlexWaygood reviewed Sep 28, 2023

View reviewed changes

Lib/test/test_socket.py Outdated Show resolved Hide resolved

Typos!

da358cc

Typos!

cf0cfd0

Use quiet=True by default

dfda254

vstinner reviewed Sep 28, 2023

View reviewed changes

serhiy-storchaka reviewed Sep 28, 2023

View reviewed changes

sobolevn mentioned this pull request Sep 28, 2023

Tests crash on macos Sonoma #109981

Closed

sobolevn mentioned this pull request Oct 4, 2023

gh-110332: Remove mentions of random.WichmannHill from test_zlib #110334

Merged

graingert added 2 commits January 16, 2025 07:00

Merge branch 'main' into issue110012

8e97f04

Update test_socket.py

f45b3b8

graingert mentioned this pull request Jan 16, 2025

gh-128770: raise warnings as errors in test suite - except for test_socket which still logs warnings #128726

Merged

graingert added needs backport to 3.12 only security fixes needs backport to 3.13 bugs and security fixes labels Jan 16, 2025

graingert requested a review from hugovk January 16, 2025 08:09

tomasr8 removed the needs backport to 3.12 only security fixes label Apr 10, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

gh-110012: Fix `RuntimeWarning` in `test_socket` #110013

gh-110012: Fix `RuntimeWarning` in `test_socket` #110013

sobolevn commented Sep 28, 2023 •

edited

Loading

AlexWaygood left a comment •

edited

Loading

sobolevn commented Sep 28, 2023

AlexWaygood commented Sep 28, 2023

vstinner Sep 28, 2023

sobolevn Sep 28, 2023

serhiy-storchaka left a comment

vstinner commented Sep 28, 2023

sobolevn commented Sep 28, 2023

sobolevn commented Sep 28, 2023

serhiy-storchaka commented Sep 28, 2023

sobolevn commented Sep 28, 2023

serhiy-storchaka commented Sep 28, 2023

sobolevn commented Sep 28, 2023

vstinner commented Sep 28, 2023

AlexWaygood commented Sep 28, 2023

sobolevn commented Sep 28, 2023

vstinner commented Sep 28, 2023

hugovk commented Jan 16, 2025

serhiy-storchaka commented Jan 16, 2025

graingert commented Jan 16, 2025 •

edited

Loading

AlexWaygood commented Jan 16, 2025

graingert commented Jan 16, 2025

	def _filterwarnings(filters, quiet=False):
	"""Catch the warnings, then check if all the expected
	warnings have been raised and re-raise unexpected warnings.
	If 'quiet' is True, only re-raise the unexpected warnings.
	"""
	# Clear the warning registry of the calling module
	# in order to re-raise the warnings.
	frame = sys._getframe(2)
	registry = frame.f_globals.get('__warningregistry__')
	if registry:
	registry.clear()
	with warnings.catch_warnings(record=True) as w:
	# Set filter "always" to record all warnings. Because
	# test_warnings swap the module, we need to look up in
	# the sys.modules dictionary.
	sys.modules['warnings'].simplefilter("always")
	yield WarningsRecorder(w)
	# Filter the recorded warnings
	reraise = list(w)
	missing = []
	for msg, cat in filters:
	seen = False
	for w in reraise[:]:
	warning = w.message
	# Filter out the matching messages
	if (re.match(msg, str(warning), re.I) and
	issubclass(warning.__class__, cat)):
	seen = True
	reraise.remove(w)
	if not seen and not quiet:
	# This filter caught nothing
	missing.append((msg, cat.__name__))
	if reraise:
	raise AssertionError("unhandled warning %s" % reraise[0])
	if missing:
	raise AssertionError("filter (%r, %s) did not catch any warning" %
	missing[0])

gh-110012: Fix RuntimeWarning in test_socket #110013

Are you sure you want to change the base?

gh-110012: Fix RuntimeWarning in test_socket #110013

Conversation

sobolevn commented Sep 28, 2023 • edited Loading

AlexWaygood left a comment • edited Loading

Choose a reason for hiding this comment

sobolevn commented Sep 28, 2023

AlexWaygood commented Sep 28, 2023

vstinner Sep 28, 2023

Choose a reason for hiding this comment

sobolevn Sep 28, 2023

Choose a reason for hiding this comment

serhiy-storchaka left a comment

Choose a reason for hiding this comment

vstinner commented Sep 28, 2023

sobolevn commented Sep 28, 2023

sobolevn commented Sep 28, 2023

serhiy-storchaka commented Sep 28, 2023

sobolevn commented Sep 28, 2023

serhiy-storchaka commented Sep 28, 2023

sobolevn commented Sep 28, 2023

vstinner commented Sep 28, 2023

AlexWaygood commented Sep 28, 2023

sobolevn commented Sep 28, 2023

vstinner commented Sep 28, 2023

hugovk commented Jan 16, 2025

serhiy-storchaka commented Jan 16, 2025

graingert commented Jan 16, 2025 • edited Loading

AlexWaygood commented Jan 16, 2025

graingert commented Jan 16, 2025

gh-110012: Fix `RuntimeWarning` in `test_socket` #110013

gh-110012: Fix `RuntimeWarning` in `test_socket` #110013

sobolevn commented Sep 28, 2023 •

edited

Loading

AlexWaygood left a comment •

edited

Loading

graingert commented Jan 16, 2025 •

edited

Loading