Mailman 3 March 2025 - Python-checkins

Update cryptographic CODEOWNERS entry (#131892)
by gpshead March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/2c3e3fe54ce9aaa57b1e870c450d4cad4d… commit: 2c3e3fe54ce9aaa57b1e870c450d4cad4db4b208 branch: main author: Bénédikt Tran <10796600+picnixz(a)users.noreply.github.com> committer: gpshead <greg(a)krypto.org> date: 2025-03-30T10:55:05-07:00 summary: Update cryptographic CODEOWNERS entry (#131892) update cryptographic CODEOWNERS entry files: M .github/CODEOWNERS M Misc/ACKS diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS index 2f684d4183e519..4bea55a495f776 100644 --- a/.github/CODEOWNERS +++ b/.github/CODEOWNERS @@ -107,13 +107,16 @@ Objects/exceptions.c @iritkatriel # Hashing & cryptographic primitives **/*hashlib* @gpshead @tiran @picnixz **/*hashopenssl* @gpshead @tiran @picnixz -**/*pyhash* @gpshead @tiran -**/sha* @gpshead @tiran @picnixz -Modules/md5* @gpshead @tiran @picnixz -**/*blake* @gpshead @tiran @picnixz -Modules/_hacl/** @gpshead +**/*pyhash* @gpshead @tiran @picnixz +Modules/*blake* @gpshead @tiran @picnixz +Modules/*md5* @gpshead @tiran @picnixz +Modules/*sha* @gpshead @tiran @picnixz +Modules/_hacl/** @gpshead @picnixz **/*hmac* @gpshead @picnixz +# libssl +**/*ssl* @gpshead @picnixz + # logging **/*logging* @vsajip diff --git a/Misc/ACKS b/Misc/ACKS index b0b4ea0db44b64..814a530f7b79d4 100644 --- a/Misc/ACKS +++ b/Misc/ACKS @@ -1910,6 +1910,7 @@ Matias Torchinsky Sandro Tosi Richard Townsend David Townshend +Bénédikt Tran Nathan Trapuzzano Laurence Tratt Alberto Trevino

1 0

gh-127794: Validate email header names according to RFC 5322 (#127820)
by picnixz March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/c432d0147bdf1a66604e7a3d6a71660ae7… commit: c432d0147bdf1a66604e7a3d6a71660ae79b5f45 branch: main author: Srinivas Reddy Thatiparthy (తాటిపర్తి శ్రీనివాస్ రెడ్డి) <thatiparthysreenivas(a)gmail.com> committer: picnixz <10796600+picnixz(a)users.noreply.github.com> date: 2025-03-30T12:29:29Z summary: gh-127794: Validate email header names according to RFC 5322 (#127820) `email.message.Message` objects now validate header names specified via `__setitem__` or `add_header` according to RFC 5322, §2.2 [1]. In particular, callers should expect a ValueError to be raised for invalid header names. [1]: https://datatracker.ietf.org/doc/html/rfc5322#section-2.2 --------- Co-authored-by: Bénédikt Tran <10796600+picnixz(a)users.noreply.github.com> Co-authored-by: R. David Murray <rdmurray(a)bitdance.com> files: A Misc/NEWS.d/next/Library/2024-12-11-17-44-36.gh-issue-127794.VwmRsp.rst M Lib/email/_policybase.py M Lib/email/policy.py M Lib/test/test_email/test_email.py M Lib/test/test_email/test_message.py diff --git a/Lib/email/_policybase.py b/Lib/email/_policybase.py index 4b63b97217a835..95e79b8938bb4c 100644 --- a/Lib/email/_policybase.py +++ b/Lib/email/_policybase.py @@ -4,6 +4,7 @@ """ import abc +import re from email import header from email import charset as _charset from email.utils import _has_surrogates @@ -14,6 +15,14 @@ 'compat32', ] +# validation regex from RFC 5322, equivalent to pattern re.compile("[!-9;-~]+$") +valid_header_name_re = re.compile("[\041-\071\073-\176]+$") + +def validate_header_name(name): + # Validate header name according to RFC 5322 + if not valid_header_name_re.match(name): + raise ValueError( + f"Header field name contains invalid characters: {name!r}") class _PolicyBase: @@ -314,6 +323,7 @@ def header_store_parse(self, name, value): """+ The name and value are returned unmodified. """ + validate_header_name(name) return (name, value) def header_fetch_parse(self, name, value): diff --git a/Lib/email/policy.py b/Lib/email/policy.py index 6e109b65011a44..4169150101a29d 100644 --- a/Lib/email/policy.py +++ b/Lib/email/policy.py @@ -4,7 +4,13 @@ import re import sys -from email._policybase import Policy, Compat32, compat32, _extend_docstrings +from email._policybase import ( + Compat32, + Policy, + _extend_docstrings, + compat32, + validate_header_name +) from email.utils import _has_surrogates from email.headerregistry import HeaderRegistry as HeaderRegistry from email.contentmanager import raw_data_manager @@ -138,6 +144,7 @@ def header_store_parse(self, name, value): CR or LF characters. """ + validate_header_name(name) if hasattr(value, 'name') and value.name.lower() == name.lower(): return (name, value) if isinstance(value, str) and len(value.splitlines())>1: diff --git a/Lib/test/test_email/test_email.py b/Lib/test/test_email/test_email.py index 2deb35721576b8..724af3b787d38b 100644 --- a/Lib/test/test_email/test_email.py +++ b/Lib/test/test_email/test_email.py @@ -728,6 +728,31 @@ def test_nonascii_add_header_with_tspecial(self): "attachment; filename*=utf-8''Fu%C3%9Fballer%20%5Bfilename%5D.ppt", msg['Content-Disposition']) + def test_invalid_header_names(self): + invalid_headers = [ + ('Invalid Header', 'contains space'), + ('Tab\tHeader', 'contains tab'), + ('Colon:Header', 'contains colon'), + ('', 'Empty name'), + (' LeadingSpace', 'starts with space'), + ('TrailingSpace ', 'ends with space'), + ('Header\x7F', 'Non-ASCII character'), + ('Header\x80', 'Extended ASCII'), + ] + for policy in (email.policy.default, email.policy.compat32): + for setter in (Message.__setitem__, Message.add_header): + for name, value in invalid_headers: + self.do_test_invalid_header_names( + policy, setter,name, value) + + def do_test_invalid_header_names(self, policy, setter, name, value): + with self.subTest(policy=policy, setter=setter, name=name, value=value): + message = Message(policy=policy) + pattern = r'(?i)(?=.*invalid)(?=.*header)(?=.*name)' + with self.assertRaisesRegex(ValueError, pattern) as cm: + setter(message, name, value) + self.assertIn(f"{name!r}", str(cm.exception)) + def test_binary_quopri_payload(self): for charset in ('latin-1', 'ascii'): msg = Message() diff --git a/Lib/test/test_email/test_message.py b/Lib/test/test_email/test_message.py index 96979db27f3a21..23c39775a8b2e5 100644 --- a/Lib/test/test_email/test_message.py +++ b/Lib/test/test_email/test_message.py @@ -1004,6 +1004,30 @@ def test_folding_with_long_nospace_http_policy_1(self): parsed_msg = message_from_bytes(m.as_bytes(), policy=policy.default) self.assertEqual(parsed_msg['Message-ID'], m['Message-ID']) + def test_invalid_header_names(self): + invalid_headers = [ + ('Invalid Header', 'contains space'), + ('Tab\tHeader', 'contains tab'), + ('Colon:Header', 'contains colon'), + ('', 'Empty name'), + (' LeadingSpace', 'starts with space'), + ('TrailingSpace ', 'ends with space'), + ('Header\x7F', 'Non-ASCII character'), + ('Header\x80', 'Extended ASCII'), + ] + for email_policy in (policy.default, policy.compat32): + for setter in (EmailMessage.__setitem__, EmailMessage.add_header): + for name, value in invalid_headers: + self.do_test_invalid_header_names(email_policy, setter, name, value) + + def do_test_invalid_header_names(self, policy, setter, name, value): + with self.subTest(policy=policy, setter=setter, name=name, value=value): + message = EmailMessage(policy=policy) + pattern = r'(?i)(?=.*invalid)(?=.*header)(?=.*name)' + with self.assertRaisesRegex(ValueError, pattern) as cm: + setter(message, name, value) + self.assertIn(f"{name!r}", str(cm.exception)) + def test_get_body_malformed(self): """test for bpo-42892""" msg = textwrap.dedent("""\ diff --git a/Misc/NEWS.d/next/Library/2024-12-11-17-44-36.gh-issue-127794.VwmRsp.rst b/Misc/NEWS.d/next/Library/2024-12-11-17-44-36.gh-issue-127794.VwmRsp.rst new file mode 100644 index 00000000000000..b6e39d66d04221 --- /dev/null +++ b/Misc/NEWS.d/next/Library/2024-12-11-17-44-36.gh-issue-127794.VwmRsp.rst @@ -0,0 +1,4 @@ +When headers are added to :class:`email.message.Message` objects, either through +:meth:`email.message.Message.__setitem__` or :meth:`email.message.Message.add_header`, +the field name is now validated according to :rfc:`RFC 5322, Section 2.2 <5322#section-2.2>` +and a :exc:`ValueError` is raised if the field name contains any invalid characters.

1 0

gh-131178: Update help message for `timeit` CLI (#131326)
by hugovk March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/55150a79cacbce44f50cea128c511782df… commit: 55150a79cacbce44f50cea128c511782df0ab277 branch: main author: Semyon Moroz <donbarbos(a)proton.me> committer: hugovk <1324225+hugovk(a)users.noreply.github.com> date: 2025-03-30T14:19:38+03:00 summary: gh-131178: Update help message for `timeit` CLI (#131326) files: M Lib/test/test_timeit.py M Lib/timeit.py diff --git a/Lib/test/test_timeit.py b/Lib/test/test_timeit.py index 72a104fc1a6790..f5ae0a84eb3506 100644 --- a/Lib/test/test_timeit.py +++ b/Lib/test/test_timeit.py @@ -297,9 +297,7 @@ def test_main_negative_reps(self): @unittest.skipIf(sys.flags.optimize >= 2, "need __doc__") def test_main_help(self): s = self.run_main(switches=['-h']) - # Note: It's not clear that the trailing space was intended as part of - # the help text, but since it's there, check for it. - self.assertEqual(s, timeit.__doc__ + ' ') + self.assertEqual(s, timeit.__doc__) def test_main_verbose(self): s = self.run_main(switches=['-v']) diff --git a/Lib/timeit.py b/Lib/timeit.py index c106e0f67356da..e767f0187826df 100644 --- a/Lib/timeit.py +++ b/Lib/timeit.py @@ -44,7 +44,6 @@ timeit(string, string) -> float repeat(string, string) -> list default_timer() -> float - """ import gc @@ -302,7 +301,7 @@ def main(args=None, *, _wrap_timer=None): precision += 1 verbose += 1 if o in ("-h", "--help"): - print(__doc__, end=' ') + print(__doc__, end="") return 0 setup = "\n".join(setup) or "pass"

1 0

gh-127945: mark `TestLocalization` as thread unsafe in ctypes (#131899)
by kumaraditya303 March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/eaffc34d3ce38710d633f7246d3e15aadd… commit: eaffc34d3ce38710d633f7246d3e15aaddee63d1 branch: main author: Kumar Aditya <kumaraditya(a)python.org> committer: kumaraditya303 <kumaraditya(a)python.org> date: 2025-03-30T11:08:42Z summary: gh-127945: mark `TestLocalization` as thread unsafe in ctypes (#131899) files: M Lib/test/test_ctypes/test_dlerror.py diff --git a/Lib/test/test_ctypes/test_dlerror.py b/Lib/test/test_ctypes/test_dlerror.py index 1c1b2aab3d589e..8af34e62b94faa 100644 --- a/Lib/test/test_ctypes/test_dlerror.py +++ b/Lib/test/test_ctypes/test_dlerror.py @@ -120,7 +120,7 @@ def test_null_dlsym(self): # Assert that the IFUNC was called self.assertEqual(os.read(pipe_r, 2), b'OK') - +(a)test.support.thread_unsafe('setlocale is not thread-safe') @unittest.skipUnless(os.name != 'nt', 'test requires dlerror() calls') class TestLocalization(unittest.TestCase):

1 0

gh-127945: acquire critical section around `PyCFuncPtr_call` (#131898)
by kumaraditya303 March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/46ada1e5e0b8942ea72e748880732a9468… commit: 46ada1e5e0b8942ea72e748880732a9468be1e92 branch: main author: Kumar Aditya <kumaraditya(a)python.org> committer: kumaraditya303 <kumaraditya(a)python.org> date: 2025-03-30T11:01:12Z summary: gh-127945: acquire critical section around `PyCFuncPtr_call` (#131898) files: M Modules/_ctypes/_ctypes.c diff --git a/Modules/_ctypes/_ctypes.c b/Modules/_ctypes/_ctypes.c index 7536d3fdc2b882..59ea579627e56f 100644 --- a/Modules/_ctypes/_ctypes.c +++ b/Modules/_ctypes/_ctypes.c @@ -4406,7 +4406,7 @@ _build_result(PyObject *result, PyObject *callargs, } static PyObject * -PyCFuncPtr_call(PyObject *op, PyObject *inargs, PyObject *kwds) +PyCFuncPtr_call_lock_held(PyObject *op, PyObject *inargs, PyObject *kwds) { PyObject *restype; PyObject *converters; @@ -4544,6 +4544,16 @@ PyCFuncPtr_call(PyObject *op, PyObject *inargs, PyObject *kwds) outmask, inoutmask, numretvals); } +static PyObject * +PyCFuncPtr_call(PyObject *op, PyObject *inargs, PyObject *kwds) +{ + PyObject *result; + Py_BEGIN_CRITICAL_SECTION(op); + result = PyCFuncPtr_call_lock_held(op, inargs, kwds); + Py_END_CRITICAL_SECTION(); + return result; +} + static int PyCFuncPtr_traverse(PyObject *op, visitproc visit, void *arg) {

1 0

gh-127945: make initialization of `error_object_name` thread safe in ctypes (#131896)
by kumaraditya303 March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/28e476f6a2e4c7621f6724c2c3f5764a30… commit: 28e476f6a2e4c7621f6724c2c3f5764a30623311 branch: main author: Kumar Aditya <kumaraditya(a)python.org> committer: kumaraditya303 <kumaraditya(a)python.org> date: 2025-03-30T10:50:35Z summary: gh-127945: make initialization of `error_object_name` thread safe in ctypes (#131896) files: M Modules/_ctypes/_ctypes.c M Modules/_ctypes/callproc.c diff --git a/Modules/_ctypes/_ctypes.c b/Modules/_ctypes/_ctypes.c index 803d41630dd59a..7536d3fdc2b882 100644 --- a/Modules/_ctypes/_ctypes.c +++ b/Modules/_ctypes/_ctypes.c @@ -6093,14 +6093,19 @@ _ctypes_mod_exec(PyObject *mod) } #ifdef WORDS_BIGENDIAN - st->swapped_suffix = PyUnicode_InternFromString("_le"); + st->swapped_suffix = PyUnicode_InternFromString("_le"); #else - st->swapped_suffix = PyUnicode_InternFromString("_be"); + st->swapped_suffix = PyUnicode_InternFromString("_be"); #endif if (st->swapped_suffix == NULL) { return -1; } + st->error_object_name = PyUnicode_InternFromString("ctypes.error_object"); + if (st->error_object_name == NULL) { + return -1; + } + if (_ctypes_add_types(mod) < 0) { return -1; } diff --git a/Modules/_ctypes/callproc.c b/Modules/_ctypes/callproc.c index 80b66afb3625d1..158422fb9b0471 100644 --- a/Modules/_ctypes/callproc.c +++ b/Modules/_ctypes/callproc.c @@ -164,12 +164,7 @@ _ctypes_get_errobj(ctypes_state *st, int **pspace) "cannot get thread state"); return NULL; } - if (st->error_object_name == NULL) { - st->error_object_name = PyUnicode_InternFromString("ctypes.error_object"); - if (st->error_object_name == NULL) { - return NULL; - } - } + assert(st->error_object_name != NULL); if (PyDict_GetItemRef(dict, st->error_object_name, &errobj) < 0) { return NULL; }

1 0

[3.13] gh-129994: update docs when using `dualstack_ipv6` in `socket.create_server` (GH-129996) (#131890)
by picnixz March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/19187991a8311e87735eb3abe55c0c4997… commit: 19187991a8311e87735eb3abe55c0c499739769d branch: 3.13 author: Miss Islington (bot) <31488909+miss-islington(a)users.noreply.github.com> committer: picnixz <10796600+picnixz(a)users.noreply.github.com> date: 2025-03-30T12:07:08+02:00 summary: [3.13] gh-129994: update docs when using `dualstack_ipv6` in `socket.create_server` (GH-129996) (#131890) gh-129994: update docs when using `dualstack_ipv6` in `socket.create_server` (GH-129996) (cherry picked from commit 044a1e13d5cfc91b0a8774e936877e91cbac3a16) Co-authored-by: Alexander Ershov <150328014+AEErshov(a)users.noreply.github.com> files: M Doc/library/socket.rst diff --git a/Doc/library/socket.rst b/Doc/library/socket.rst index 12f95b839a948e..e5baa3a002a01e 100644 --- a/Doc/library/socket.rst +++ b/Doc/library/socket.rst @@ -838,10 +838,10 @@ The following functions all create :ref:`socket objects <socket-objects>`. , a default reasonable value is chosen. *reuse_port* dictates whether to set the :data:`SO_REUSEPORT` socket option. - If *dualstack_ipv6* is true and the platform supports it the socket will - be able to accept both IPv4 and IPv6 connections, else it will raise - :exc:`ValueError`. Most POSIX platforms and Windows are supposed to support - this functionality. + If *dualstack_ipv6* is true, *family* is :data:`AF_INET6` and the platform + supports it the socket will be able to accept both IPv4 and IPv6 connections, + else it will raise :exc:`ValueError`. Most POSIX platforms and Windows are + supposed to support this functionality. When this functionality is enabled the address returned by :meth:`socket.getpeername` when an IPv4 connection occurs will be an IPv6 address represented as an IPv4-mapped IPv6 address.

1 0

gh-127945: fix thread safety of creating instances of ctypes structures (#131716)
by kumaraditya303 March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/bc5a028c132efd7598593b75ab86b69020… commit: bc5a028c132efd7598593b75ab86b690209a6e6b branch: main author: Kumar Aditya <kumaraditya(a)python.org> committer: kumaraditya303 <kumaraditya(a)python.org> date: 2025-03-30T15:22:30+05:30 summary: gh-127945: fix thread safety of creating instances of ctypes structures (#131716) files: M Modules/_ctypes/_ctypes.c M Modules/_ctypes/ctypes.h M Modules/_ctypes/stgdict.c diff --git a/Modules/_ctypes/_ctypes.c b/Modules/_ctypes/_ctypes.c index d3e717757b0645..803d41630dd59a 100644 --- a/Modules/_ctypes/_ctypes.c +++ b/Modules/_ctypes/_ctypes.c @@ -109,6 +109,7 @@ bytes(cdata) #include "pycore_call.h" // _PyObject_CallNoArgs() #include "pycore_ceval.h" // _Py_EnterRecursiveCall() #include "pycore_unicodeobject.h" // _PyUnicode_EqualToASCIIString() +#include "pycore_pyatomic_ft_wrappers.h" #ifdef MS_WIN32 # include "pycore_modsupport.h" // _PyArg_NoKeywords() #endif @@ -710,13 +711,15 @@ StructUnionType_init(PyObject *self, PyObject *args, PyObject *kwds, int isStruc if (baseinfo == NULL) { return 0; } - + int ret = 0; + STGINFO_LOCK(baseinfo); /* copy base info */ - if (PyCStgInfo_clone(info, baseinfo) < 0) { - return -1; + ret = PyCStgInfo_clone(info, baseinfo); + if (ret >= 0) { + stginfo_set_dict_final_lock_held(baseinfo); /* set the 'final' flag in the baseclass info */ } - info->flags &= ~DICTFLAG_FINAL; /* clear the 'final' flag in the subclass info */ - baseinfo->flags |= DICTFLAG_FINAL; /* set the 'final' flag in the baseclass info */ + STGINFO_UNLOCK(); + return ret; } return 0; } @@ -3122,6 +3125,7 @@ PyCData_MallocBuffer(CDataObject *obj, StgInfo *info) * access. */ assert (Py_REFCNT(obj) == 1); + assert(stginfo_get_dict_final(info) == 1); if ((size_t)info->size <= sizeof(obj->b_value)) { /* No need to call malloc, can use the default buffer */ @@ -3167,7 +3171,7 @@ PyCData_FromBaseObj(ctypes_state *st, return NULL; } - info->flags |= DICTFLAG_FINAL; + stginfo_set_dict_final(info); cmem = (CDataObject *)((PyTypeObject *)type)->tp_alloc((PyTypeObject *)type, 0); if (cmem == NULL) { return NULL; @@ -3216,7 +3220,7 @@ PyCData_AtAddress(ctypes_state *st, PyObject *type, void *buf) return NULL; } - info->flags |= DICTFLAG_FINAL; + stginfo_set_dict_final(info); pd = (CDataObject *)((PyTypeObject *)type)->tp_alloc((PyTypeObject *)type, 0); if (!pd) { @@ -3451,7 +3455,7 @@ generic_pycdata_new(ctypes_state *st, return NULL; } - info->flags |= DICTFLAG_FINAL; + stginfo_set_dict_final(info); obj = (CDataObject *)type->tp_alloc(type, 0); if (!obj) diff --git a/Modules/_ctypes/ctypes.h b/Modules/_ctypes/ctypes.h index 2b8192059a0dc2..7ffe83dc63d6db 100644 --- a/Modules/_ctypes/ctypes.h +++ b/Modules/_ctypes/ctypes.h @@ -6,6 +6,8 @@ #include "pycore_moduleobject.h" // _PyModule_GetState() #include "pycore_typeobject.h" // _PyType_GetModuleState() +#include "pycore_critical_section.h" +#include "pycore_pyatomic_ft_wrappers.h" // Do we support C99 complex types in ffi? // For Apple's libffi, this must be determined at runtime (see gh-128156). @@ -375,7 +377,7 @@ typedef struct CFieldObject { typedef struct { int initialized; Py_ssize_t size; /* number of bytes */ - Py_ssize_t align; /* alignment requirements */ + Py_ssize_t align; /* alignment reqwuirements */ Py_ssize_t length; /* number of fields */ ffi_type ffi_type_pointer; PyObject *proto; /* Only for Pointer/ArrayObject */ @@ -390,15 +392,64 @@ typedef struct { PyObject *checker; PyObject *module; int flags; /* calling convention and such */ +#ifdef Py_GIL_DISABLED + PyMutex mutex; /* critical section mutex */ +#endif + uint8_t dict_final; /* pep3118 fields, pointers need PyMem_Free */ char *format; int ndim; Py_ssize_t *shape; -/* Py_ssize_t *strides; */ /* unused in ctypes */ -/* Py_ssize_t *suboffsets; */ /* unused in ctypes */ + /* Py_ssize_t *strides; */ /* unused in ctypes */ + /* Py_ssize_t *suboffsets; */ /* unused in ctypes */ } StgInfo; + +/* + To ensure thread safety in the free threading build, the `STGINFO_LOCK` and + `STGINFO_UNLOCK` macros use critical sections to protect against concurrent + modifications to `StgInfo` and assignment of the `dict_final` field. Once + `dict_final` is set, `StgInfo` is treated as read-only, and no further + modifications are allowed. This approach allows most read operations to + proceed without acquiring the critical section lock. + + The `dict_final` field is written only after all other modifications to + `StgInfo` are complete. The reads and writes of `dict_final` use the + sequentially consistent memory ordering to ensure that all other fields are + visible to other threads before the `dict_final` bit is set. +*/ + +#define STGINFO_LOCK(stginfo) Py_BEGIN_CRITICAL_SECTION_MUT(&(stginfo)->mutex) +#define STGINFO_UNLOCK() Py_END_CRITICAL_SECTION() + +static inline uint8_t +stginfo_get_dict_final(StgInfo *info) +{ + return FT_ATOMIC_LOAD_UINT8(info->dict_final); +} + +static inline void +stginfo_set_dict_final_lock_held(StgInfo *info) +{ + _Py_CRITICAL_SECTION_ASSERT_MUTEX_LOCKED(&info->mutex); + FT_ATOMIC_STORE_UINT8(info->dict_final, 1); +} + + +// Set the `dict_final` bit in StgInfo. It checks if the bit is already set +// and in that avoids acquiring the critical section (general case). +static inline void +stginfo_set_dict_final(StgInfo *info) +{ + if (stginfo_get_dict_final(info) == 1) { + return; + } + STGINFO_LOCK(info); + stginfo_set_dict_final_lock_held(info); + STGINFO_UNLOCK(); +} + extern int PyCStgInfo_clone(StgInfo *dst_info, StgInfo *src_info); extern void ctype_clear_stginfo(StgInfo *info); @@ -427,8 +478,6 @@ PyObject *_ctypes_callproc(ctypes_state *st, #define TYPEFLAG_ISPOINTER 0x100 #define TYPEFLAG_HASPOINTER 0x200 -#define DICTFLAG_FINAL 0x1000 - struct tagPyCArgObject { PyObject_HEAD ffi_type *pffi_type; diff --git a/Modules/_ctypes/stgdict.c b/Modules/_ctypes/stgdict.c index 6e4f8eb21d9632..c4c4ec2cc5266a 100644 --- a/Modules/_ctypes/stgdict.c +++ b/Modules/_ctypes/stgdict.c @@ -34,6 +34,10 @@ PyCStgInfo_clone(StgInfo *dst_info, StgInfo *src_info) dst_info->ffi_type_pointer.elements = NULL; memcpy(dst_info, src_info, sizeof(StgInfo)); +#ifdef Py_GIL_DISABLED + dst_info->mutex = (PyMutex){0}; +#endif + dst_info->dict_final = 0; Py_XINCREF(dst_info->proto); Py_XINCREF(dst_info->argtypes); @@ -248,23 +252,23 @@ PyCStructUnionType_update_stginfo(PyObject *type, PyObject *fields, int isStruct ctypes_state *st = get_module_state_by_def(Py_TYPE(type)); StgInfo *stginfo; if (PyStgInfo_FromType(st, type, &stginfo) < 0) { - goto error; + return -1; } if (!stginfo) { PyErr_SetString(PyExc_TypeError, "ctypes state is not initialized"); - goto error; + return -1; } PyObject *base = (PyObject *)((PyTypeObject *)type)->tp_base; StgInfo *baseinfo; if (PyStgInfo_FromType(st, base, &baseinfo) < 0) { - goto error; + return -1; } + STGINFO_LOCK(stginfo); /* If this structure/union is already marked final we cannot assign _fields_ anymore. */ - - if (stginfo->flags & DICTFLAG_FINAL) {/* is final ? */ + if (stginfo_get_dict_final(stginfo) == 1) {/* is final ? */ PyErr_SetString(PyExc_AttributeError, "_fields_ is final"); goto error; @@ -422,12 +426,13 @@ PyCStructUnionType_update_stginfo(PyObject *type, PyObject *fields, int isStruct goto error; } assert(info); - + STGINFO_LOCK(info); stginfo->ffi_type_pointer.elements[ffi_ofs + i] = &info->ffi_type_pointer; if (info->flags & (TYPEFLAG_ISPOINTER | TYPEFLAG_HASPOINTER)) stginfo->flags |= TYPEFLAG_HASPOINTER; - info->flags |= DICTFLAG_FINAL; /* mark field type final */ + stginfo_set_dict_final_lock_held(info); /* mark field type final */ + STGINFO_UNLOCK(); if (-1 == PyObject_SetAttr(type, prop->name, prop_obj)) { goto error; } @@ -461,15 +466,15 @@ PyCStructUnionType_update_stginfo(PyObject *type, PyObject *fields, int isStruct /* We did check that this flag was NOT set above, it must not have been set until now. */ - if (stginfo->flags & DICTFLAG_FINAL) { + if (stginfo_get_dict_final(stginfo) == 1) { PyErr_SetString(PyExc_AttributeError, "Structure or union cannot contain itself"); goto error; } - stginfo->flags |= DICTFLAG_FINAL; + stginfo_set_dict_final_lock_held(stginfo); retval = MakeAnonFields(type); -error: +error:; Py_XDECREF(layout_func); Py_XDECREF(kwnames); Py_XDECREF(align_obj); @@ -478,6 +483,7 @@ PyCStructUnionType_update_stginfo(PyObject *type, PyObject *fields, int isStruct Py_XDECREF(layout_fields); Py_XDECREF(layout); Py_XDECREF(format_spec_obj); + STGINFO_UNLOCK(); return retval; }

1 0

[3.12] gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (#131894)
by picnixz March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/525eddf220318f1d60644fec57890f0607… commit: 525eddf220318f1d60644fec57890f0607de4168 branch: 3.12 author: Miss Islington (bot) <31488909+miss-islington(a)users.noreply.github.com> committer: picnixz <10796600+picnixz(a)users.noreply.github.com> date: 2025-03-30T09:29:49Z summary: [3.12] gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (#131894) gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (cherry picked from commit edfbd8c062285e83dc5f596288df2a762e68bcdc) Co-authored-by: Adam Dangoor <adamdangoor(a)gmail.com> files: M Doc/library/stdtypes.rst diff --git a/Doc/library/stdtypes.rst b/Doc/library/stdtypes.rst index c0b6b9ed6c0023..31b0f6f12544bc 100644 --- a/Doc/library/stdtypes.rst +++ b/Doc/library/stdtypes.rst @@ -4577,7 +4577,7 @@ can be used interchangeably to index the same dictionary entry. such as an empty list. To get distinct values, use a :ref:`dict comprehension <dict>` instead. - .. method:: get(key, default=None) + .. method:: get(key, default=None, /) Return the value for *key* if *key* is in the dictionary, else *default*. If *default* is not given, it defaults to ``None``, so that this method @@ -4619,7 +4619,7 @@ can be used interchangeably to index the same dictionary entry. .. versionadded:: 3.8 - .. method:: setdefault(key, default=None) + .. method:: setdefault(key, default=None, /) If *key* is in the dictionary, return its value. If not, insert *key* with a value of *default* and return *default*. *default* defaults to

1 0

[3.13] gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (#131893)
by picnixz March 30, 2025

March 30, 2025

https://github.com/python/cpython/commit/2570896ad7a1e7cd1ec963f44ff4b7d0a0… commit: 2570896ad7a1e7cd1ec963f44ff4b7d0a075da44 branch: 3.13 author: Miss Islington (bot) <31488909+miss-islington(a)users.noreply.github.com> committer: picnixz <10796600+picnixz(a)users.noreply.github.com> date: 2025-03-30T09:28:49Z summary: [3.13] gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (#131893) gh-131885: Document that `dict.setdefault` and `dict.get` take no keyword arguments (GH-128208) (cherry picked from commit edfbd8c062285e83dc5f596288df2a762e68bcdc) Co-authored-by: Adam Dangoor <adamdangoor(a)gmail.com> files: M Doc/library/stdtypes.rst diff --git a/Doc/library/stdtypes.rst b/Doc/library/stdtypes.rst index 2f4cd43cae4ad9..942e89b6235c6b 100644 --- a/Doc/library/stdtypes.rst +++ b/Doc/library/stdtypes.rst @@ -4586,7 +4586,7 @@ can be used interchangeably to index the same dictionary entry. such as an empty list. To get distinct values, use a :ref:`dict comprehension <dict>` instead. - .. method:: get(key, default=None) + .. method:: get(key, default=None, /) Return the value for *key* if *key* is in the dictionary, else *default*. If *default* is not given, it defaults to ``None``, so that this method @@ -4628,7 +4628,7 @@ can be used interchangeably to index the same dictionary entry. .. versionadded:: 3.8 - .. method:: setdefault(key, default=None) + .. method:: setdefault(key, default=None, /) If *key* is in the dictionary, return its value. If not, insert *key* with a value of *default* and return *default*. *default* defaults to

1 0