_PyFrame_Copy() copied interpreter frames into generator and
frame-object storage without initializing the visited byte. Incremental
GC later reads frame->visited in mark_stacks() on non-start passes, so
copied frames could expose an uninitialized value once they became live
on a thread stack again.
Reset visited when copying a frame so copied frames start with defined
GC bookkeeping state. Preserve lltrace in Py_DEBUG builds.
Fix huge page leak in datastack chunk allocator
The original fix rounded datastack chunk allocations in pystate.c so that
_PyObject_VirtualFree() would receive the full huge page mapping size.
Change direction and move that logic into _PyObject_VirtualAlloc() and
_PyObject_VirtualFree() instead. The key invariant is that munmap() must see
the full mapped size, so alloc and free now apply the same platform-specific
rounding in the allocator layer.
This keeps _PyStackChunk bookkeeping in requested-size units, avoids a
hardcoded 2 MB assumption, and also covers other small virtual-memory users
such as the JIT tracer state allocation in optimizer.c.
Add the padded parameter in functions related to Base32 and Base64 codecs
in the binascii and base64 modules. In the encoding functions it controls
whether the pad character can be added in the output, in the decoding
functions it controls whether padding is required in input.
Padding of input no longer required in base64.urlsafe_b64decode() by default.
The gc_stats struct contains ring buffers of gc_generation_stats
entries (11 young + 3×2 old on default builds). Embedding it inline
in _gc_runtime_state, which is itself inline in PyInterpreterState,
pushed fields like _gil.locked and threads.head to offsets beyond
what out-of-process profilers and debuggers can reasonably read in
a single buffer (e.g. offset 9384 for _gil.locked vs an 8 KiB read
buffer).
Heap-allocate generation_stats via PyMem_RawCalloc in _PyGC_Init and
free it in _PyGC_Fini. This shrinks PyInterpreterState by ~1.6 KiB
and keeps the GIL, thread-list, and other frequently-inspected fields
at stable, low offsets.
Revert "GH-126910: Make `_Py_get_machine_stack_pointer` return the stack pointer (#147945)"
This reverts commit 255026d9ee,
which broke a tier-1 buildbot.
When Python is built in debug mode:
* long_alloc() now initializes digits with a pattern to detect usage of
uninitialized digits.
* _PyLong_CompactValue() now makes sure that the digit is zero when the
sign is zero.
* PyLongWriter_Finish() now raises SystemError if it detects uninitialized
digits
Co-authored-by: Serhiy Storchaka <storchaka@gmail.com>
With this we can assume, that _PyLong_SetSignAndDigitCount() and
_PyLong_SetDigitCount() operate on non-immortal integers.
Co-authored-by: Victor Stinner <vstinner@python.org>
* Make _Py_get_machine_stack_pointer return the stack pointer (or close to it), not the frame pointer
* Make ``_Py_ReachedRecursionLimit`` inline again
* Remove ``_Py_MakeRecCheck`` relacing its use with ``_Py_ReachedRecursionLimit``
* Move stack swtiching check into ``_Py_CheckRecursiveCall``
- Add Py_TARGET_ABI3T macro.
- Add ".abi3t.so" to importlib EXTENSION_SUFFIXES.
- Remove ".abi3.so" from importlib EXTENSION_SUFFIXES on Free Threading.
- Adjust tests
This is part of the implementation for PEP-803.
Detailed documentation to come later.
Co-authored-by: Nathan Goldbaum <nathan.goldbaum@gmail.com>
Accepted range for the bytes_per_sep argument of bytes.hex(),
bytearray.hex(), memoryview.hex(), and binascii.b2a_hex()
is now increased, so passing sys.maxsize and -sys.maxsize is now
valid.
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().
Co-authored-by: Victor Stinner <vstinner@python.org>
Rename from _Py_INTERNAL_ABI_SLOT to _Py_ABI_SLOT
and define the macro using _PyABIInfo_DEFAULT.
Use the ABI slot in stdlib extension modules to enable running
a check of ABI version compatibility.
_tkinter, _tracemalloc and readline don't use the slots, hence they need
explicit handling.
Co-authored-by: Victor Stinner <vstinner@python.org>
Also fix a few related issues in the pyatomic headers:
* Fix _Py_atomic_store_uint_release in pyatomic_msc.h to use __stlr32
on ARM64 instead of a plain volatile store (which is only relaxed on
ARM64).
* Add missing _Py_atomic_store_uint_release to pyatomic_gcc.h.
* Fix pseudo-code comment for _Py_atomic_store_ptr_release in
pyatomic.h.
* Add the alphabet parameter in functions b2a_base64(), a2b_base64(),
b2a_base85(), and a2b_base85().
* And a number of "*_ALPHABET" constants.
* Remove b2a_z85() and a2b_z85().
Add special cases for classmethod and staticmethod descriptors in
_PyObject_GetMethodStackRef() to avoid calling tp_descr_get, which
avoids reference count contention on the bound method and underlying
callable. This improves scaling when calling classmethods and
staticmethods from multiple threads.
Also refactor method_vectorcall in classobject.c into a new _PyObject_VectorcallPrepend() helper so that it can be used by
PyObject_VectorcallMethod as well.
* Convert DEOPT_IFs to EXIT_IFs for guards. Keep DEOPT_IF for intentional drops to the interpreter.
* Modify BINARY_OP_SUBSCR_LIST_INT and STORE_SUBSCR_LIST_INT to handle negative indices, to keep EXIT_IFs and DEOPT_IFs in different uops
