[3.13] gh-148529: Minor improvements of the struct module documentation (GH-148565) (GH-149072)
* Document that 's' and 'p' accept bytes and bytearray.
* Fix some footnotes.
* Clarify that "string" is a byte string.
* Fix the module docstring.
(cherry picked from commit 3e5a3cb2bd222f97f793b01bc1c0f7bb62aefc31)
[3.13] Document that multiprocessing treats local same-user processes as trusted (GH-149001) (#149034)
Document that multiprocessing treats local same-user processes as trusted (GH-149001)
Clarify in the Authentication keys section that the authkey handshake
covers Listener/Client (addressable endpoints) only, not the anonymous
pipes behind Pipe() and Queue, and that isolation between same-user
processes must be arranged at the OS level.
(cherry picked from commit f27e91e37212f148b8fe72a3656a69b242625622)
Co-authored-by: Gregory P. Smith <68491+gpshead@users.noreply.github.com>
Gregory P. Smith [Sun, 26 Apr 2026 05:55:20 +0000 (22:55 -0700)]
[3.13] gh-105936: Properly update closure cells for __setattr__ and __delattr__ in frozen dataclasses with slots (GH-144021) (GH-148476)
gh-105936: Properly update closure cells for `__setattr__` and `__delattr__` in frozen dataclasses with slots (GH-144021)
(cherry picked from commit 8a398bfbbc6769f6cabb3177702e7a506e203d61)
The cherry-pick required additional changes beyond the original commit
because 3.13 lacks the `__class__` closure cell fixup machinery that
was added in 3.14 by GH-124455 (gh-90562). Specifically:
- Backported `_update_func_cell_for__class__()` helper function and the
closure fixup loop in `_add_slots()` from GH-124455. Without these,
renaming the closure variable from `cls` to `__class__` has no effect
because nothing updates the cell when the class is recreated with slots.
- Changed `_add_slots()` to use `newcls` instead of reusing `cls` for the
recreated class, so both old and new class references are available for
the fixup loop.
- Replaced `assertNotHasAttr` with `assertFalse(hasattr(...))` in tests
(assertNotHasAttr was added in 3.14).
- Dropped `test_original_class_is_gced` additions (that test does not
exist on 3.13; it was added by GH-137047 for gh-135228 which was not
backported to 3.13).
gh-148947: dataclasses: fix error on empty __class__ cell (GH-148948)
[3.13] gh-141473: Speed up subprocess test_communicate_timeout_large_input long tail (GH-149003) (#149005)
gh-141473: Speed up subprocess test_communicate_timeout_large_input long tail (GH-149003)
gh-141473: Speed up test_communicate_timeout_large_input
Replace the slow reader's 30s sleep with a parent-driven wake over a
loopback socket so post-timeout communicate() doesn't block waiting
for the child to wake on its own. Worst-case runtime: ~30s -> <1s.
(cherry picked from commit e1384cfd25b4fba5e0f8f3e6b536930e2e6cf5cf)
Co-authored-by: Gregory P. Smith <68491+gpshead@users.noreply.github.com>
[3.13] gh-142965: Fix Concatenate documentation to reflect valid use cases (GH-143316) (#148900)
The documentation previously stated that Concatenate is only valid
when used as the first argument to Callable, but according to PEP 612,
it can also be used when instantiating user-defined generic classes
with ParamSpec parameters.
(cherry picked from commit 75ff1afcb6a1bb2b3d54899e9b222a61798fa491)
Co-authored-by: John Seong <39040639+sandole@users.noreply.github.com>
_PyRawMutex_UnlockSlow CAS-removes the waiter from the list and then
calls _PySemaphore_Wakeup, with no handshake. If _PySemaphore_Wait
returns Py_PARK_INTR, the waiter can destroy its stack-allocated
semaphore before the unlocker's Wakeup runs, causing a fatal error from
ReleaseSemaphore / sem_post.
Loop in _PyRawMutex_LockSlow until _PySemaphore_Wait returns Py_PARK_OK,
which is only signalled when a matching Wakeup has been observed.
Also include GetLastError() and the handle in the Windows fatal messages
in _PySemaphore_Init, _PySemaphore_Wait, and _PySemaphore_Wakeup to make
similar races easier to diagnose in the future.
[3.13] gh-148763: Fix paramter name in `multiprocessing.connection.send_bytes/recv_bytes_into` docs (GH-126603) (#148787)
gh-148763: Fix paramter name in `multiprocessing.connection.send_bytes/recv_bytes_into` docs (GH-126603)
(cherry picked from commit e50acef0b2c2057874a9eec98c37ca6cf8ee98e1)
Doc: Fix buf argument name in multiprocessing connection send_bytes
Co-authored-by: Matthew Davis <7035647+mdavis-xyz@users.noreply.github.com>
Forbid marshalling recursive code objects which
cannot be correctly unmarshalled.
Add multiple tests for recursive data structures.
(cherry picked from commit 2e37d836411e99cff7bb341ba14be5ea95fac08c)
Victor Stinner [Thu, 16 Apr 2026 17:50:27 +0000 (19:50 +0200)]
[3.13] gh-148535: Don't use gcc -fprofile-update=atomic flag on i686 (#148554) (#148656)
gh-148535: Don't use gcc -fprofile-update=atomic flag on i686 (#148554)
The -fprofile-update=atomic flag was added to fix a random GCC
internal error on PGO build (gh-145801) caused by corruption of
profile data (.gcda files). The problem is that it makes the PGO
build way slower (up to 47x slower) on i686. Since the GCC internal
error was not seen on i686 so far, don't use -fprofile-update=atomic
on i686.
Use negative lookahead in option regex to prevent backtracking, and to avoid changing logic outside the regexes (since people could use the regex directly).
(cherry picked from commit 7e0a0be4097f9d29d66fe23f5af86f18a34ed7dd)
[3.13] gh-148192: Fix Generator._make_boundary behavior with CRLF line endings. (GH-148193) (#148549)
The Generator._make_boundary regex did not match on boundary phrases correctly when using CRLF line endings due to re.MULTILINE not considering \r\n as a line ending.
(cherry picked from commit 4af46b4ab5af49d8df034320a9a70fcbb062f7cf)
Co-authored-by: Henry Jones <44321887+henryivesjones@users.noreply.github.com>
[3.13] gh-148508: Add resilience to SSL preauth tests on iOS (GH-148536) (#148540)
Adds handling for a test case seen in the iOS SSL tests where an SSL connection fails to
handshake correctly.
(cherry picked from commit c40e8b016a90820e4d799922903b90a505ffaf55)
Co-authored-by: Russell Keith-Magee <russell@keith-magee.com>
[3.13] gh-146139: Disable `socketpair` authentication on WASI (GH-146140) (#148527)
gh-146139: Disable `socketpair` authentication on WASI (GH-146140)
Calling `connect(2)` on a non-blocking socket on WASI may leave the socket in a
"connecting" but not yet "connected" state. In the former case, calling
`getpeername(2)` on it will fail, leading to an unhandled exception in Python.
(cherry picked from commit a5b76d53bb29afd864243f44ef22968f6385dfa0)
Co-authored-by: Joel Dice <joel.dice@akamai.com> Co-authored-by: Victor Stinner <vstinner@python.org> Co-authored-by: Brett Cannon <brett@python.org>
[3.13] tests: use errno.EBADF instead of hardcoded number in _close_file() (GH-148345) (#148411)
tests: use errno.EBADF instead of hardcoded number in _close_file() (GH-148345)
test_interpreters: use errno.EBADF instead of hardcoded number in _close_file()
Replace the hardcoded `9` check in `Lib/test/test_interpreters/utils.py` with `errno.EBADF`.
Using `errno.EBADF` makes the helper portable across platforms with different errno numbering while preserving the intended behavior.
(cherry picked from commit cef334fd4c4c24a542ce81ad940b1426b5a7cdbd)
Ensures that the testbed's Gradle configuration uses the cross-build environment
variable, and that variable is passed to Gradle by the cross-build script.
(cherry picked from commit b29afe62f7236f7161c2670dccc24368217a7fb1)
Co-authored-by: Russell Keith-Magee <russell@keith-magee.com> Co-authored-by: Malcolm Smith <smith@chaquo.com>
[3.13] gh-146313: Fix multiprocessing ResourceTracker deadlock after os.fork() (GH-146316) (#148426)
gh-146313: Fix multiprocessing ResourceTracker deadlock after os.fork() (GH-146316)
`ResourceTracker.__del__` (added in gh-88887 circa Python 3.12) calls
os.waitpid(pid, 0) which blocks indefinitely if a process created via os.fork()
still holds the tracker pipe's write end. The tracker never sees EOF, never
exits, and the parent hangs at interpreter shutdown.
Fix with two layers:
- **At-fork handler.** An os.register_at_fork(after_in_child=...)
handler closes the inherited pipe fd in the child unless a preserve
flag is set. popen_fork.Popen._launch() sets the flag before its
fork so mp.Process(fork) children keep the fd and reuse the parent's
tracker (preserving gh-80849). Raw os.fork() children close the fd,
letting the parent reap promptly.
- **Timeout safety-net.** _stop_locked() gains a wait_timeout
parameter. When called from `__del__`, it polls with WNOHANG using
exponential backoff for up to 1 second instead of blocking
indefinitely. The at-fork handler makes this unreachable in
well-behaved paths; it remains for abnormal shutdowns.
(cherry picked from commit 3a7df632c96eb6c5de12fac08d1da42df9e25334)
Co-authored-by: Gregory P. Smith <68491+gpshead@users.noreply.github.com> Co-authored-by: Itamar Oren <itamarost@gmail.com>
[3.13] gh-147965: Add shutdown() to multiprocessing.Queue excluded methods (GH-147970) (#148417)
gh-147965: Add shutdown() to multiprocessing.Queue excluded methods (GH-147970)
The multiprocessing.Queue documentation states it implements all
methods of queue.Queue except task_done() and join(). Since
queue.Queue.shutdown() was added in Python 3.13,
multiprocessing.Queue also does not implement it. Update the docs
to include shutdown() in the list of excluded methods.
(cherry picked from commit 22290ed011a8ac4060390e57f53053ab932fb3f3)
[3.13] gh-145105: Fix crash in csv.reader with re-entrant iterator (GH-145106) (#148405)
gh-145105: Fix crash in csv.reader with re-entrant iterator (GH-145106)
When a custom iterator calls next() on the same csv.reader from
within __next__, the inner iteration sets self->fields to NULL.
The outer iteration then crashes in parse_save_field() by passing
NULL to PyList_Append.
Add a guard after PyIter_Next() to detect that fields was set to
NULL by a re-entrant call, and raise csv.Error instead of crashing.
(cherry picked from commit 20994b1809f9c162e4cae01a5af08bd492ede9f9)
Co-authored-by: Ramin Farajpour Cami <ramin.blackhat@gmail.com>
decoded[:-1] only strips one character, leaving a stray \r when eol
is two characters. Fix: decoded[:-len(eol)].
(cherry picked from commit 1a0edb1fa899c067f19b09598b45cdb6e733c4ee)
Co-authored-by: Stefan Zetzsche <120379523+stefanzetzsche@users.noreply.github.com>
[3.13] gh-144503: Pass sys.argv to forkserver as real argv elements (GH-148194) (#148196)
Avoid embedding the parent's sys.argv into the forkserver -c command
string via repr(). When sys.argv is large (e.g. thousands of file
paths from a pre-commit hook), the resulting single argument could
exceed the OS per-argument length limit (MAX_ARG_STRLEN on Linux,
typically 128 KiB), causing posix_spawn to fail and the parent to
observe a BrokenPipeError.
Instead, append the argv entries as separate command-line arguments
after -c; the forkserver child reads them back as sys.argv[1:]. This
cannot exceed any limit the parent itself did not already satisfy.
Pre-create the Android emulator image so that the the configuration can be
modified to use 4GB of RAM.
(cherry picked from commit a95ee3a21d97afdbe6bd2ce4cd8343a36cd13b02)
Malcolm Smith [Sun, 5 Apr 2026 23:47:01 +0000 (00:47 +0100)]
[3.13] gh-146541: Allow building the Android testbed for 32-bit targets (GH-146542) (#148107)
Allows building the Android testbed for 32-bit targets, adding the target triplets
`arm-linux-androideabi` and `i686-linux-android`.
(cherry picked from commit 848bbe9ff21ae0a3ee412cc25843835ace4f75df)
Co-authored-by: Robert Kirkman <31490854+robertkirkman@users.noreply.github.com>
[3.13] gh-146450: Normalise feature set of Android build script with other platform build scripts (GH-146451) (#148066)
Allows for cleaning a subset of targets, customization of the download cache and
cross-build directories, and modifies the build command to allow 'all', 'build'
and 'hosts' targets.
(cherry picked from commit b8470deb5d52f524ae18c6f232fecfc99b133397)
Co-authored-by: Russell Keith-Magee <russell@keith-magee.com>
Victor Stinner [Tue, 31 Mar 2026 15:06:38 +0000 (17:06 +0200)]
[3.13] gh-143050: Correct PyLong_FromString() to use _PyLong_Negate() (#145901) (#147437)
The long_from_string_base() might return a small integer, when the
_pylong.py is used to do conversion. Hence, we must be careful here to
not smash it "small int" bit by using the _PyLong_FlipSign().
Malcolm Smith [Sat, 28 Mar 2026 07:25:35 +0000 (07:25 +0000)]
[3.13] gh-145616: Detect Android sysconfig ABI correctly on 32-bit ARM Android on 64-bit ARM kernel (GH-145617) (#146539)
When Python is running on 32-bit ARM Android on a 64-bit ARM kernel, `os.uname().machine` is `armv8l`. Such devices run the same userspace code as `armv7l` devices, so apply the same `armeabi_v7a` Android ABI to them, which works.
(cherry picked from commit 3a2b81e919103c0be3bc60a47aaa74d34fea6e9e)
Co-authored-by: Robert Kirkman <31490854+robertkirkman@users.noreply.github.com>
projects / thirdparty / Python / cpython.git / log
