This PR adds the ability to enable the GIL if it was disabled at
interpreter startup, and modifies the multi-phase module initialization
path to enable the GIL when loading a module, unless that module's spec
includes a slot indicating it can run safely without the GIL.
PEP 703 called the constant for the slot `Py_mod_gil_not_used`; I went
with `Py_MOD_GIL_NOT_USED` for consistency with gh-104148.
A warning will be issued up to once per interpreter for the first
GIL-using module that is loaded. If `-v` is given, a shorter message
will be printed to stderr every time a GIL-using module is loaded
(including the first one that issues a warning).
The GC keeps track of the number of allocations (less deallocations)
since the last GC. This buffers the count in thread-local state and uses
atomic operations to modify the per-interpreter count. The thread-local
buffering avoids contention on shared state.
A consequence is that the GC scheduling is not as precise, so
"test_sneaky_frame_object" is skipped because it requires that the GC be
run exactly after allocating a frame object.
We do not want to add locking in `tp_traverse` slot implementations.
Instead, stop the world when calling `gc.get_referents`. Note that the the
stop the world call is a no-op in the default build.
Co-authored-by: Pablo Galindo Salgado <Pablogsal@gmail.com>
This splits part of Modules/gcmodule.c of into Python/gc.c, which
now contains the core garbage collection implementation. The Python
module remain in the Modules/gcmodule.c file.
* gh-112529: Use atomic operations for `gcstate->collecting`
The `collecting` field in `GCState` is used to prevent overlapping garbage
collections within the same interpreter. This is updated to use atomic
operations in order to be thread-safe in `--disable-gil` builds.
The GC code is refactored a bit to support this. More of the logic is pushed
down to `gc_collect_main()` so that we can safely order the logic setting
`collecting`, the selection of the generation, and the invocation of callbacks
with respect to the atomic operations and the (future) stop-the-world pauses.
The change uses atomic operations for both `--disable-gil` and the default
build (with the GIL) to avoid extra `#ifdef` guards and ease the maintenance
burden.
In PyObject_GC_Del, in Py_DEBUG mode, when warning about GC objects that
were not properly untracked before starting destruction, take care to
untrack the object _before_ warning, to avoid triggering a GC run and
causing the problem the code tries to warn about. Also make sure to save and
restore any pending exceptions, which the warning would otherwise clobber or
trigger an assertion error on.
Replace most of calls of _PyErr_WriteUnraisableMsg() and some
calls of PyErr_WriteUnraisable(NULL) with PyErr_FormatUnraisable().
Co-authored-by: Victor Stinner <vstinner@python.org>
Statistics gathering is now off by default. Use the "-X pystats"
command line option or set the new PYTHONSTATS environment variable
to 1 to turn statistics gathering on at Python startup.
Statistics are no longer dumped at exit if statistics gathering was
off or statistics have been cleared.
Changes:
* Add PYTHONSTATS environment variable.
* sys._stats_dump() now returns False if statistics are not dumped
because they are all equal to zero.
* Add PyConfig._pystats member.
* Add tests on sys functions and on setting PyConfig._pystats to 1.
* Add Include/cpython/pystats.h and Include/internal/pycore_pystats.h
header files.
* Rename '_py_stats' variable to '_Py_stats'.
* Exclude Include/cpython/pystats.h from the Py_LIMITED_API.
* Move pystats.h include from object.h to Python.h.
* Add _Py_StatsOn() and _Py_StatsOff() functions. Remove
'_py_stats_struct' variable from the API: make it static in
specialize.c.
* Document API in Include/pystats.h and Include/cpython/pystats.h.
* Complete pystats documentation in Doc/using/configure.rst.
* Don't write "all zeros" stats: if _stats_off() and _stats_clear()
or _stats_dump() were called.
* _PyEval_Fini() now always call _Py_PrintSpecializationStats() which
does nothing if stats are all zeros.
Co-authored-by: Michael Droettboom <mdboom@gmail.com>
Move private _PyDict functions to the internal C API (pycore_dict.h):
* _PyDict_Contains_KnownHash()
* _PyDict_DebugMallocStats()
* _PyDict_DelItemIf()
* _PyDict_GetItemWithError()
* _PyDict_HasOnlyStringKeys()
* _PyDict_MaybeUntrack()
* _PyDict_MergeEx()
No longer export these functions.
Remove _PyWeakref_GetWeakrefCount() and _PyWeakref_ClearRef() from
the public C API: move them to the internal C API.
Refactor also _weakref_getweakrefs() code to make it more readable.
* gcmodule.c reuses _Py_AS_GC(op) for AS_GC()
* Move gcmodule.c FROM_GC() implementation to a new _Py_FROM_GC()
static inline function in pycore_gc.h.
* _PyObject_IS_GC(): only get the type once
* gc_is_finalized(à) and PyObject_GC_IsFinalized() use
_PyGC_FINALIZED(), instead of _PyGCHead_FINALIZED().
* Remove _Py_CAST() in pycore_gc.h: this header file is not built
with C++.
Here we are doing no more than adding the value for Py_mod_multiple_interpreters and using it for stdlib modules. We will start checking for it in gh-104206 (once PyInterpreterState.ceval.own_gil is added in gh-104204).
his involves moving tp_dict, tp_bases, and tp_mro to PyInterpreterState, in the same way we did for tp_subclasses. Those three fields are effectively const for builtin static types (unlike tp_subclasses). In theory we only need to make their values immortal, along with their contents. However, that isn't such a simple proposition. (See gh-103823.) In the meantime the simplest solution is to move the fields into the interpreter.
One alternative is to statically allocate the values, but that's its own can of worms.
This is the implementation of PEP683
Motivation:
The PR introduces the ability to immortalize instances in CPython which bypasses reference counting. Tagging objects as immortal allows up to skip certain operations when we know that the object will be around for the entire execution of the runtime.
Note that this by itself will bring a performance regression to the runtime due to the extra reference count checks. However, this brings the ability of having truly immutable objects that are useful in other contexts such as immutable data sharing between sub-interpreters.
* Make sure that the current exception is always normalized.
* Remove redundant type and traceback fields for the current exception.
* Add new API functions: PyErr_GetRaisedException, PyErr_SetRaisedException
* Add new API functions: PyException_GetArgs, PyException_SetArgs
* code_sizeof() now uses an unsigned type (size_t) to compute the result.
* Fix _PyObject_ComputedDictPointer(): cast _PyObject_VAR_SIZE() to
Py_ssize_t, rather than long: it's a different type on 64-bit Windows.
* Clarify that _PyObject_VAR_SIZE() uses an unsigned type (size_t).
This was added for bpo-40514 (gh-84694) to test out a per-interpreter GIL. However, it has since proven unnecessary to keep the experiment in the repo. (It can be done as a branch in a fork like normal.) So here we are removing:
* the configure option
* the macro
* the code enabled by the macro
Fix a crash in subinterpreters related to the garbage collector. When
a subinterpreter is deleted, untrack all objects tracked by its GC.
To prevent a crash in deallocator functions expecting objects to be
tracked by the GC, leak a strong reference to these objects on
purpose, so they are never deleted and their deallocator functions
are not called.
Convert the PyType_SUPPORTS_WEAKREFS() macro to a regular function.
It no longer access the PyTypeObject.tp_weaklistoffset member
directly.
Add _PyType_SUPPORTS_WEAKREFS() static inline functions, used
internally by Python for best performance.
Py_EndInterpreter() now explicitly untracks all objects currently
tracked by the GC. Previously, if an object was used later by another
interpreter, calling PyObject_GC_UnTrack() on the object crashed if
the previous or the next object of the PyGC_Head structure became a
dangling pointer.
If the DEBUG_STATS debug flag is set, gc_collect_main() now uses
_PyTime_GetPerfCounter() instead of _PyTime_GetMonotonicClock() to
measure the elapsed time.
On Windows, _PyTime_GetMonotonicClock() only has a resolution of 15.6
ms, whereas _PyTime_GetPerfCounter() is closer to a resolution of 100
ns.
Add new C-API functions to control the state of the garbage collector:
PyGC_Enable(), PyGC_Disable(), PyGC_IsEnabled(),
corresponding to the functions in the gc module.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
Co-authored-by: Victor Stinner <vstinner@python.org>
Pass the current interpreter (interp) rather than the current Python
thread state (tstate) to internal functions which only use the
interpreter.
Modified functions:
* _PyXXX_Fini() and _PyXXX_ClearFreeList() functions
* _PyEval_SignalAsyncExc(), make_pending_calls()
* _PySys_GetObject(), sys_set_object(), sys_set_object_id(), sys_set_object_str()
* should_audit(), set_flags_from_config(), make_flags()
* _PyAtExit_Call()
* init_stdio_encoding()
* etc.
No longer use deprecated aliases to functions:
* Replace PyObject_MALLOC() with PyObject_Malloc()
* Replace PyObject_REALLOC() with PyObject_Realloc()
* Replace PyObject_FREE() with PyObject_Free()
* Replace PyObject_Del() with PyObject_Free()
* Replace PyObject_DEL() with PyObject_Free()
The last GC collection is now done before clearing builtins and sys
dictionaries. Add also assertions to ensure that gc.collect() is no
longer called after _PyGC_Fini().
Pass also the tstate to PyInterpreterState_Clear() to pass the
correct tstate to _PyGC_CollectNoFail() and _PyGC_Fini().
