A command created with createcommand() held a strong reference to the
interpreter, forming an uncollectable cycle (interpreter -> command ->
interpreter) that kept the interpreter and the callback alive until the
command was removed with deletecommand() or destroy(). The command now
borrows the reference; it cannot outlive the interpreter, which deletes its
commands when finalized.
(cherry picked from commit bbf7786a90)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
When a Tcl command running its own event loop (such as vwait or
wait_variable) was active and the user typed input on stdin, the event
loop kept spinning at 100% CPU. The stdin file handler is now removed as
soon as input becomes available.
Also fix gh-139816: an exception raised in a callback no longer stops the
event loop to wait for Enter on a Python built without readline; pending
callbacks keep running until input is actually available on stdin.
(cherry picked from commit 3ffda34f5c)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: mdehoon <mjldehoon@yahoo.com>
Co-authored-by: Christopher Chavez <chrischavez@gmx.us>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
gh-152156: Fix a crash in `interpeters.create` under limited memory conditions (GH-152163)
(cherry picked from commit 3ad66bf10d)
Co-authored-by: sobolevn <mail@sobolevn.me>
gh-151929: Get uptime on BSD/macOS in pythoninfo (GH-152189)
* Check sysctlbyname() function and sys/sysctl.h header in
configure.
* Add _testcapi.uptime_bsd() function.
(cherry picked from commit b6d89edc4a)
Co-authored-by: Victor Stinner <vstinner@python.org>
gh-151929: Get machine ID and uptime on Windows in pythoninfo (#152146)
* Replace "linux." prefix with "system." in pythoninfo.
* Add _winapi.GetTickCount64() function.
(cherry picked from commit f9910519af)
Add docstrings for the curses.window, curses.error, curses.panel.panel
and curses.panel.error classes. Document the panel class and its error
exception in curses.panel.rst, using the real lowercase panel name.
(cherry picked from commit 560ff8e202)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Tcl 8.x crashes when title-casing a non-BMP character during Tk
initialization, so such a className is now rejected with a ValueError.
(cherry picked from commit 124c7cd91b)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Tcl/Tk 9 may embed the Tk script library in the Tk DLL on Windows. This embedded library is not found by Tcl by default.
Mount the loaded Tk DLL as a zipfs archive before calling Tk_Init(), so Tk can find its embedded tk_library using its existing library discovery logic.
Preserve Tk_Init()'s normal path if the library is not embedded.
(cherry picked from commit c4eb3adbb4)
Co-authored-by: Jonathan J. Helmus <jjhelmus@gmail.com>
The module-global curses_screen_encoding stored a borrowed pointer to the
encoding owned by the window returned by the first initscr() call. That
window can be deallocated while unctrl() and ungetch(), which have no window
of their own, still use the pointer to encode non-ASCII characters.
Keep a private copy of the encoding instead.
(cherry picked from commit 551f8e16f8)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
gh-151229: Finalize JIT tracer in test eval-frame stub (gh-151609)
(cherry picked from commit 3fa92e7c55)
Co-authored-by: Donghee Na <donghee.na@python.org>
Fix errors and clarify the curses, curses.panel and curses.ascii docs
against X/Open Curses and ncurses, and sync the affected docstrings.
(cherry picked from commit 65afcdd8df)
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
gh-149044: Improve Py_tp_base[s] docs & error message for non-type bases (GH-151252)
The initial implementation of PEP 820 worsened the error message
when non-types are given as base types in Py_tp_bases & Py_tp_base.
Bring back the 'bases must be types' wording and add a 'got' note for
easier debugging.
Improve slot ID documentation, and soft-deprecate Py_tp_base
(as per the PEP).
(cherry picked from commit 16185e9fe2)
Co-authored-by: Petr Viktorin <encukou@gmail.com>
gh-151510: Fix __lazy_import__ without frame (GH-151511)
(cherry picked from commit eff805b7a7)
Co-authored-by: AN Long <aisk@users.noreply.github.com>
gh-151223: fix tsan data races in load global specializations (GH-151393)
(cherry picked from commit e9d5280f6c)
Co-authored-by: Kumar Aditya <kumaraditya@python.org>
gh-151424: Fix impossible stack traces in `RemoteUnwinder(..., cache_frames=True)` by copying chunks on cache miss (GH-151426)
(cherry picked from commit 6ce088e20a)
Co-authored-by: Maurycy Pawłowski-Wieroński <maurycy@maurycy.com>
The standard says that a call to `memcpy` must pass a valid source and
destination pointer even if the size is 0, so we must avoid calling
`memcpy` when our source pointer is NULL. If we don't, an optimizing
compiler can decide that the pointer must be non-NULL based on the
presence of UB, and optimize out checks for null pointers.
Specifically, note that the standard says:
Where an argument declared as size_t n specifies the length of the
array for a function, n can have the value zero on a call to that
function. Unless explicitly stated otherwise in the description of
a particular function in this subclause, pointer arguments on such
a call shall still have valid values, as described in 7.1.4.
And section 7.1.4 says:
If an argument to a function has an invalid value (such as a value
outside the domain of the function, or a pointer outside the address
space of the program, or a null pointer, or a pointer to
non-modifiable storage when the corresponding parameter is not
const-qualified) or a type (after default argument promotion) not
expected by a function with a variable number of arguments, the
behavior is undefined.
The specification for `memcpy` doesn't state that it's allowed to be
called with null pointers, and Linux's `/usr/include/string.h` declares
`memcpy` as `__nonnull ((1, 2))`.
(cherry picked from commit c37599200f)
Co-authored-by: Matt Wozniski <mwozniski@bloomberg.net>
gh-143988: Fix re-entrant mutation crashes in socket sendmsg/recvmsg_into (GH-143987)
Fix crashes in socket.sendmsg() and socket.recvmsg_into() that could
occur if buffer sequences are mutated re-entrantly during argument
parsing via __buffer__ protocol callbacks.
The bug occurs because:
1. PySequence_Fast() returns the original list object when the input
is already a list (not a copy).
2. During iteration, PyObject_GetBuffer() triggers __buffer__
callbacks which may clear the list.
3. Subsequent iterations access invalid memory (heap OOB read).
The fix replaces PySequence_Fast() with PySequence_Tuple() which
always creates a new tuple, ensuring the sequence cannot be mutated
during iteration.
(cherry picked from commit 896f7fdc7d)
Co-authored-by: tonghuaroot (童话) <tonghuaroot@gmail.com>
Co-authored-by: tonghuaroot <23011166+tonghuaroot@users.noreply.github.com>
ass_subscript_slice() returned early when the computed slice length
was zero, bypassing validation performed for non-empty slices.
(cherry picked from commit fc9c4db130)
Co-authored-by: Jiseok CHOI <jiseok.dev@gmail.com>
gh-150750: Fix a race condition in `deque.index` with free-threading (GH-150779)
(cherry picked from commit d83d50b5b7)
Co-authored-by: sobolevn <mail@sobolevn.me>
