Victor Stinner [Thu, 19 Mar 2020 16:40:12 +0000 (17:40 +0100)]
bpo-40010: Optimize signal handling in multithreaded applications (GH-19067)
If a thread different than the main thread gets a signal, the
bytecode evaluation loop is no longer interrupted at each bytecode
instruction to check for pending signals which cannot be handled.
Only the main thread of the main interpreter can handle signals.
Previously, the bytecode evaluation loop was interrupted at each
instruction until the main thread handles signals.
Changes:
* COMPUTE_EVAL_BREAKER() and SIGNAL_PENDING_SIGNALS() no longer set
eval_breaker to 1 if the current thread cannot handle signals.
* take_gil() now always recomputes eval_breaker.
Victor Stinner [Thu, 19 Mar 2020 01:41:21 +0000 (02:41 +0100)]
bpo-39984: Move pending calls to PyInterpreterState (GH-19066)
If Py_AddPendingCall() is called in a subinterpreter, the function is
now scheduled to be called from the subinterpreter, rather than being
called from the main interpreter.
Each subinterpreter now has its own list of scheduled calls.
* Move pending and eval_breaker fields from _PyRuntimeState.ceval
to PyInterpreterState.ceval.
* new_interpreter() now calls _PyEval_InitThreads() to create
pending calls lock.
* Fix Py_AddPendingCall() for subinterpreters. It now calls
_PyThreadState_GET() which works in a subinterpreter if the
caller holds the GIL, and only falls back on
PyGILState_GetThisThreadState() if _PyThreadState_GET()
returns NULL.
Pablo Galindo [Wed, 18 Mar 2020 23:02:09 +0000 (23:02 +0000)]
bpo-39220: Do not optimise annotation if 'from __future__ import annotations' is used (GH-17866)
Do not apply AST-based optimizations if 'from __future__ import annotations' is used in order to
prevent information lost in the final version of the annotations.
* trip_signal() and Py_AddPendingCall() now get the current Python
thread state using PyGILState_GetThisThreadState() rather than
_PyRuntimeState_GetThreadState() to be able to get it even if the
GIL is released.
* _PyEval_SignalReceived() now expects tstate rather than ceval.
* Remove ceval parameter of _PyEval_AddPendingCall(): ceval is now
get from tstate parameter.
* _PyThreadState_DeleteCurrent() now takes tstate rather than
runtime.
* Add ensure_tstate_not_null() helper to pystate.c.
* Add _PyEval_ReleaseLock() function.
* _PyThreadState_DeleteCurrent() now calls
_PyEval_ReleaseLock(tstate) and frees PyThreadState memory after
this call, not before.
* PyGILState_Release(): rename "tcur" variable to "tstate".
Victor Stinner [Tue, 17 Mar 2020 17:09:46 +0000 (18:09 +0100)]
bpo-39824: module_traverse() don't call m_traverse if md_state=NULL (GH-18738)
Extension modules: m_traverse, m_clear and m_free functions of
PyModuleDef are no longer called if the module state was requested
but is not allocated yet. This is the case immediately after the
module is created and before the module is executed (Py_mod_exec
function). More precisely, these functions are not called if m_size is
greater than 0 and the module state (as returned by
PyModule_GetState()) is NULL.
Extension modules without module state (m_size <= 0) are not affected.
Matthias Braun [Tue, 17 Mar 2020 16:51:44 +0000 (09:51 -0700)]
bpo-26067: Do not fail test_shutil / chown when gid/uid cannot be resolved (#19032)
* bpo-26067: Do not fail test_shutil.chown when gid/uid cannot be resolved
There is no guarantee that the users primary uid or gid can be resolved
in the unix group/account databases. Skip the last part of the chown
test if we cannot resolve the gid or uid to a name.
Batuhan Taşkaya [Sun, 15 Mar 2020 19:45:56 +0000 (22:45 +0300)]
bpo-39360: Ensure all workers exit when finalizing a multiprocessing Pool (GH-19009)
When the pull is not used via the context manager or terminate() is called, there is a system in multiprocessing.util that handles finalization of all pools via an atexit handler (the Finalize) class. This class registers the _terminate_pool handler in the registry of finalizers of the module, and that registry is called on interpreter exit via _exit_function. The problem is that the "happy" path with the context manager or manual call to finalize() does some extra steps that _terminate_pool does not. The step that is not executed when the atexit() handler calls _terminate_pool is pinging the _change_notifier queue to unblock the maintenance threads.
This commit moves the notification to the _terminate_pool function so is called from both code paths.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
Peter Eisentraut [Sat, 14 Mar 2020 11:47:09 +0000 (12:47 +0100)]
Make cpython/abstract.h compatible with C90 (#18481)
Some inline functions use mixed declarations and code. These end up
visible in third-party code that includes Python.h, which might not be
using a C99 compiler. Fix by moving the declarations first, like in
the old days.
Victor Stinner [Fri, 13 Mar 2020 17:15:33 +0000 (18:15 +0100)]
bpo-39947: Add PyInterpreterState_Get() function (GH-18979)
* Rename _PyInterpreterState_Get() to PyInterpreterState_Get() and
move it the limited C API.
* Add _PyInterpreterState_Get() alias to PyInterpreterState_Get() for
backward compatibility with Python 3.8.
Victor Stinner [Fri, 13 Mar 2020 17:03:56 +0000 (18:03 +0100)]
bpo-39947: Use _PyInterpreterState_GET_UNSAFE() (GH-18978)
Replace _PyInterpreterState_Get() function call with
_PyInterpreterState_GET_UNSAFE() macro which is more efficient but
don't check if tstate or interp is NULL.
_Py_GetConfigsAsDict() now uses _PyThreadState_GET().
Victor Stinner [Fri, 13 Mar 2020 15:51:52 +0000 (16:51 +0100)]
bpo-39947: Hide implementation detail of trashcan macros (GH-18971)
Py_TRASHCAN_BEGIN_CONDITION and Py_TRASHCAN_END macro no longer
access PyThreadState attributes, but call new private
_PyTrash_begin() and _PyTrash_end() functions which hide
implementation details.
Victor Stinner [Fri, 13 Mar 2020 15:39:12 +0000 (16:39 +0100)]
bpo-35370: Add _PyEval_SetTrace() function (GH-18975)
* sys.settrace(), sys.setprofile() and _lsprof.Profiler.enable() now
properly report PySys_Audit() error if "sys.setprofile" or
"sys.settrace" audit event is denied.
* Add _PyEval_SetProfile() and _PyEval_SetTrace() function: similar
to PyEval_SetProfile() and PyEval_SetTrace() but take a tstate
parameter and return -1 on error.
* Add _PyObject_FastCallTstate() function.
Victor Stinner [Fri, 13 Mar 2020 09:19:38 +0000 (10:19 +0100)]
bpo-39947: Move Py_EnterRecursiveCall() to internal C API (GH-18972)
Move the static inline function flavor of Py_EnterRecursiveCall() and
Py_LeaveRecursiveCall() to the internal C API: they access
PyThreadState attributes. The limited C API provides regular
functions which hide implementation details.
PyInterpreterState.eval_frame function now requires a tstate (Python
thread state) parameter.
Add private functions to the C API to get and set the frame
evaluation function:
* Add tstate parameter to _PyFrameEvalFunction function type.
* Add _PyInterpreterState_GetEvalFrameFunc() and
_PyInterpreterState_SetEvalFrameFunc() functions.
* Add tstate parameter to _PyEval_EvalFrameDefault().
AIX: Fix _aix_support module when the subprocess is not available,
when building Python from scratch. It now uses new private
_bootsubprocess module, rather than having two implementations
depending if subprocess is available or not. So
_aix_support.aix_platform() result is now the same if subprocess is
available or not.
The 32-bit (49-day) TickCount relied on in EnterNonRecursiveMutex can overflow
in the gap between the 'target' time and the 'now' time WaitForSingleObjectEx
returns, causing the loop to think it needs to wait another 49 days. This is
most likely to happen when the machine is hibernated during
WaitForSingleObjectEx.
This makes acquiring a lock/event/etc from the _thread or threading module
appear to never timeout.
Replace with GetTickCount64 - this is OK now Python no longer supports XP which
lacks it, and is in use for time.monotonic().
Co-authored-by: And Clover <and.clover@bromium.com>
Miro Hrončok [Tue, 10 Mar 2020 21:16:28 +0000 (22:16 +0100)]
bpo-38662: ensurepip invokes pip via runpy (GH-18901)
The ensurepip module now invokes pip via the runpy module.
Hence it is no longer tightly coupled with the internal API of the bundled
pip version, allowing easier updates to a newer pip version both
internally and for distributors.
This way, any changes to the internal pip API won't mean ensurepip needs to be
changed as well. Also, distributors can update their pip wheels independent on
CPython release schedule.
Adam Johnson [Tue, 10 Mar 2020 18:18:50 +0000 (18:18 +0000)]
tracemalloc: 'pretty top' example no longer changes the filename (GH-18903)
I've used this recipe a couple times and the filename editing has always
been less than useful and something I've removed. This is because many
modules end up losing which package they are located in, e.g. `util/date.py`.
Serhiy Storchaka [Tue, 10 Mar 2020 16:52:34 +0000 (18:52 +0200)]
bpo-34822: Simplify AST for subscription. (GH-9605)
* Remove the slice type.
* Make Slice a kind of the expr type instead of the slice type.
* Replace ExtSlice(slices) with Tuple(slices, Load()).
* Replace Index(value) with a value itself.
All non-terminal nodes in AST for expressions are now of the expr type.
Add --with-platlibdir option to the configure script: name of the
platform-specific library directory, stored in the new sys.platlitdir
attribute. It is used to build the path of platform-specific dynamic
libraries and the path of the standard library.
It is equal to "lib" on most platforms. On Fedora and SuSE, it is
equal to "lib64" on 64-bit systems.