Allow Py_LIMITED_API for (Py_GIL_DISABLED && _Py_OPAQUE_PYOBJECT)
API that's removed when _Py_OPAQUE_PYOBJECT is defined:
- PyObject_HEAD
- _PyObject_EXTRA_INIT
- PyObject_HEAD_INIT
- PyObject_VAR_HEAD
- struct _object (i.e. PyObject) (opaque)
- struct PyVarObject (opaque)
- Py_SIZE
- Py_SET_TYPE
- Py_SET_SIZE
- PyModuleDef_Base (opaque)
- PyModuleDef_HEAD_INIT
- PyModuleDef (opaque)
- _Py_IsImmortal
- _Py_IsStaticImmortal
Note that the `_Py_IsImmortal` removal (and a few other issues)
means _Py_OPAQUE_PYOBJECT only works with limited
API 3.14+ now.
Co-authored-by: Victor Stinner <vstinner@python.org>
This commit fixes the following problems:
* The x86_64 trampolines are not preserving frame pointers
* The hardcoded offsets to the code segment from the FDE only worked properly for x64_64
* The CIE data was not following conventions of aarch64
* The eh_frame for aarch64 was not fully correct
Fixes build errors encountered in python-greenlet/greenlet#450 when building greenlet on the free-threaded build.
---------
Co-authored-by: Peter Bierma <zintensitydev@gmail.com>
Co-authored-by: Victor Stinner <vstinner@python.org>
Move PYOS_LOG2_STACK_MARGIN, PYOS_STACK_MARGIN,
PYOS_STACK_MARGIN_BYTES and PYOS_STACK_MARGIN_SHIFT macros to
pycore_pythonrun.h internal header. Add underscore (_) prefix to the
names to make them private. Rename _PYOS to _PyOS.
After Python finalization gets to the point where no other thread
can attach thread state, attempting to acquire a Python lock must hang.
Raise PythonFinalizationError instead of hanging.
This adds a "macro" to the optimizer DSL called "REPLACE_OPCODE_IF_EVALUATES_PURE", which allows automatically constant evaluating a bytecode body if certain inputs have no side effects upon evaluations (such as ints, strings, and floats).
Co-authored-by: Tomas R. <tomas.roun8@gmail.com>
The free threading build uses QSBR to delay the freeing of dictionary
keys and list arrays when the objects are accessed by multiple threads
in order to allow concurrent reads to proceed with holding the object
lock. The requests are processed in batches to reduce execution
overhead, but for large memory blocks this can lead to excess memory
usage.
Take into account the size of the memory block when deciding when to
process QSBR requests.
Also track the amount of memory being held by QSBR for mimalloc pages. Advance the write sequence if this memory exceeds a limit. Advancing the sequence will allow it to be freed more quickly.
Process the held QSBR items from the "eval breaker", rather than from `_PyMem_FreeDelayed()`. This gives a higher chance that the global read sequence has advanced enough so that items can be freed.
Co-authored-by: Sam Gross <colesbury@gmail.com>
This PR adds a PyJitRef API to the JIT's optimizer that mimics the _PyStackRef API. This allows it to track references and their stack lifetimes properly. Thus opening up the doorway to refcount elimination in the JIT.
For several builtin functions, we now fall back to __main__.__dict__ for the globals
when there is no current frame and _PyInterpreterState_IsRunningMain() returns
true. This allows those functions to be run with Interpreter.call().
The affected builtins:
* exec()
* eval()
* globals()
* locals()
* vars()
* dir()
We take a similar approach with "stateless" functions, which don't use any
global variables.
In this refactor we:
* move some code around
* make a couple of typedefs opaque
* decouple errors from session state
* improve tracebacks for propagated exceptions
This change helps simplify several upcoming changes.
* Replace _Py_ALIGN_AS(V) by _Py_ALIGNED_DEF(N, T)
This is now a common façade for the various `_Alignas` alternatives,
which behave in interesting ways -- see the source comment.
The new macro (and MSVC's `__declspec(align)`) should not be used
on a variable/member declaration that includes a struct declaraton.
A workaround is to separate the struct definition.
Do that for `PyASCIIObject.state`.
* Specify minimum PyGC_Head and PyObject alignment
As documented in InternalDocs/garbage_collector.md, the garbage collector
stores flags in the least significant two bits of the _gc_prev pointer
in struct PyGC_Head. Consequently, this pointer is only capable of storing
a location that's aligned to a 4-byte boundary.
Encode this requirement using _Py_ALIGNED_DEF.
This patch fixes a segfault in m68k, which was previously investigated
by Adrian Glaubitz here:
https://lists.debian.org/debian-68k/2024/11/msg00020.htmlhttps://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1087600
Original patch (using the GCC-only Py_ALIGNED) by Finn Thain.
Co-authored-by: Finn Thain <fthain@linux-m68k.org>
Co-authored-by: Victor Stinner <vstinner@python.org>
Co-authored-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Fix Py_RETURN_NONE, Py_RETURN_TRUE and Py_RETURN_FALSE macros in the
limited C API 3.11 and older:
Don't treat Py_None, Py_True and Py_False as immortal.
We were incorrectly handling a few opcodes that leave their operands on the stack. Treat all of these conservatively; assume that they always leave operands on the stack.
We add the following attributes on `_mi_assert_fail` to help IDE introspection:
* `__attribute__((__noreturn__))`
* `__attribute__((cold))`
* `__THROW` (GCC only)
Replace most PyUnicodeWriter_WriteUTF8() calls with
PyUnicodeWriter_WriteASCII().
Unrelated change to please the linter: remove an unused
import in test_ctypes.
Co-authored-by: Peter Bierma <zintensitydev@gmail.com>
Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>
Add _Py_PACK_VERSION for CPython's own definitions
Py_PACK_VERSION was added to limited API in 3.14, so if
Py_LIMITED_API is lower, the macro can't be used.
Add a private version that can be used in CPython headers
for checks like `Py_LIMITED_API+0 >= _Py_PACK_VERSION(3, 14)`.
