[C API] Prepare PyTypeObject structure for a stable ABI: avoid accessing members in the public API #84351

vstinner · 2020-04-03T11:57:03Z

BPO	40170
Nosy	@rhettinger, @ronaldoussoren, @vstinner, @corona10, @miss-islington, @shihai1991, @erlend-aasland
PRs	bpo-40170: Remove PyIndex_Check() macro #19375 bpo-40170: Convert PyObject_CheckBuffer() macro to a function #19376 bpo-40170: PyObject_GET_WEAKREFS_LISTPTR() becomes a function #19377 bpo-40170: PyType_HasFeature() no longer acccess directly tp_flags #19378 bpo-40170: PyObject_NEW() becomes an alias to PyObject_New() #19379 bpo-40170: Add _PyIndex_Check() internal function #19426 bpo-40170: Remove PyIndex_Check() macro #19428 bpo-40170: Convert PyObject_IS_GC() macro to a function #19464 bpo-40170: Add _PyObject_CheckBuffer() internal function #19541 Revert "bpo-40170: PyType_HasFeature() now always calls PyType_GetFlags() (GH-19378)" #21390 [3.9] Revert "bpo-40170: PyType_HasFeature() now always calls PyType_GetFlags() (GH-19378)" (GH-21390) #21391 bpo-40170: Use inline _PyType_HasFeature() function #22375 bpo-40170: Add PyType_SetBase() function #23153 bpo-40170: Hide implementation detail of Py_TRASHCAN_BEGIN macro #23235 bpo-40170: Fix PyType_Ready() refleak on static type #23236 bpo-40170: Convert PyDescr_IsData macro to C function #24535 bpo-40170: Convert PyIter_Check macro to a function #24548 bpo-40170: Always define PyExceptionClass_Name as a function #24553 bpo-40170: Move 3 NEWS entries to the C API section #24555 bpo-40170: Update What's New in Python 3.9 #29470 [3.10] bpo-40170: Update What's New in Python 3.9 (GH-29470) #29471 [3.9] bpo-40170: Update What's New in Python 3.9 (GH-29470) #29472 bpo-40170: PyType_SUPPORTS_WEAKREFS() becomes a regular function #30938 bpo-40170: Remove _Py_GetAllocatedBlocks() function #30940 bpo-40170: Remove PyHeapType_GET_MEMBERS() macro #30942 bpo-40170: Move _Py_GetAllocatedBlocks() to pycore_pymem.h #30943

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2022-01-28.14:25:55.330>
created_at = <Date 2020-04-03.11:57:03.054>
labels = ['expert-C-API', '3.10']
title = '[C API] Prepare PyTypeObject structure for a stable ABI: avoid accessing members in the public API'
updated_at = <Date 2022-01-28.14:25:55.330>
user = 'https://github.com/vstinner'

bugs.python.org fields:

activity = <Date 2022-01-28.14:25:55.330>
actor = 'vstinner'
assignee = 'none'
closed = True
closed_date = <Date 2022-01-28.14:25:55.330>
closer = 'vstinner'
components = ['C API']
creation = <Date 2020-04-03.11:57:03.054>
creator = 'vstinner'
dependencies = []
files = []
hgrepos = []
issue_num = 40170
keywords = ['patch']
message_count = 59.0
messages = ['365689', '365795', '365848', '365850', '365852', '365862', '365863', '365873', '365956', '365960', '365961', '365962', '365964', '365966', '366152', '366155', '366417', '366474', '366494', '366521', '366522', '366529', '372050', '372054', '372056', '373293', '373295', '373298', '373338', '373346', '377372', '380757', '381775', '385176', '387057', '387059', '387092', '387113', '387115', '387129', '387131', '387133', '387140', '387142', '387146', '387177', '405956', '405964', '405966', '411816', '411817', '411818', '411819', '411822', '411824', '411919', '411939', '411998', '411999']
nosy_count = 7.0
nosy_names = ['rhettinger', 'ronaldoussoren', 'vstinner', 'corona10', 'miss-islington', 'shihai1991', 'erlendaasland']
pr_nums = ['19375', '19376', '19377', '19378', '19379', '19426', '19428', '19464', '19541', '21390', '21391', '22375', '23153', '23235', '23236', '24535', '24548', '24553', '24555', '29470', '29471', '29472', '30938', '30940', '30942', '30943']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue40170'
versions = ['Python 3.10']

vstinner · 2020-04-03T11:57:02Z

Leaking the PyTypeObject structure in the C API indirectly causes ABI issue (especially for statically allocated types), cause practical issues when old fields are removed and new fields are added (ex: tp_vectorcall addition and tp_print removal caused a lot of troubles with C code generated by Cython: see bpo-37250), prevents us to add feature and experiment optimization.

I don't expect that we will be able to make PyTypeObject opaque soon. The purpose of this issue is to track the work done towards this goal.

I propose to slowly prepare the Python code base, the C API and third party code (especially Cython) to make PyTypeObject structure opaque.

We have to identify most common code patterns which access directly PyTypeObject fields, provide helper functions, and ease the migration to solutions which don't access directly PyTypeObject.

See also bpo-39573: "Make PyObject an opaque structure in the limited C API" and bpo-39947 "Make the PyThreadState structure opaque (move it to the internal C API)".

Longer rationale about making structures of the C API opaque:

--

Multiple practical issues are preventing us to make PyTypeObject opaque right now.

(*) Many C extension modules are still using statically allocated types: there is an on-going effort in bpo-40077 to convert C extension modules one by one to PyType_FromSpec().

(*) Py_TYPE(obj)->tp_name is commonly accessed to format an error message. Example:

        PyErr_Format(PyExc_TypeError, "exec() globals must be a dict, not %.100s",
                     Py_TYPE(globals)->tp_name);

I worked on bpo-34595 and started a discussion on python-dev to propose to add a new %T formatter to PyUnicode_FromFormatV() and so indirectly to PyUnicode_FromFormat() and PyErr_Format():
https://mail.python.org/archives/list/python-dev@python.org/thread/HKYUMTVHNBVB5LJNRMZ7TPUQKGKAERCJ/#3UAMHYG6UF4MPLXBZORHO4JVKUBRUZ53

Sadly, we failed to reach a consensus and I gave up on this idea. We should reconsider this idea.

We need to agree on how types should be formatted:

just the name without any dot "type_name",
qualified name "something.type_name",
fully qualified name "module.something.type_name"

There is also the question of breaking applications which rely on the current exact error message. And the question of removing legacy "%.100s" which was used before Python was able to allocate a buffer large enough to arbitrary string length. When an error is formatted in pure Python, names are never truncated.

(*) Call the function of the parent type when a method is overriden in a subclass. Example with PyTypeObject.tp_free called in a deallocator:

static void
abc_data_dealloc(_abc_data *self)
{
    PyTypeObject *tp = Py_TYPE(self);
    ...
    tp->tp_free(self);
    Py_DECREF(tp);
}

() The PEP-384 provides the most generic PyType_GetSlot() but it's not convenient to use: need to handle error (NULL), need to cast the void into the expected type (error prone cast), etc.

We should slowly add more and more helper functions for most common use cases. We can try to convert a few C extension modules of the Python stdlib to see which use cases are the most common.

Hopefully, many use cases are already abstracted by widely used functions like PyNumber_Add(), PySequence_Size(), etc.

(*) Likely other issues that I forgot.

vstinner · 2020-04-04T21:45:49Z

Macros and static inline functions of the public C API which access directly PyTypeObject fields. There may be more.

#define _PyObject_SIZE(typeobj) ( (typeobj)->tp_basicsize )

static inline vectorcallfunc
PyVectorcall_Function(PyObject *callable)
{
...
tp = Py_TYPE(callable);
offset = tp->tp_vectorcall_offset;
...
}

#define PyObject_CheckBuffer(obj) \
    ((Py_TYPE(obj)->tp_as_buffer != NULL) &&  \
     (Py_TYPE(obj)->tp_as_buffer->bf_getbuffer != NULL))

#define PyIndex_Check(obj)                              \
    (Py_TYPE(obj)->tp_as_number != NULL &&            \
     Py_TYPE(obj)->tp_as_number->nb_index != NULL)

#define PyObject_GET_WEAKREFS_LISTPTR(o) \
    ((PyObject **) (((char *) (o)) + Py_TYPE(o)->tp_weaklistoffset))

static inline int
PyType_HasFeature(PyTypeObject *type, unsigned long feature) {
#ifdef Py_LIMITED_API
    return ((PyType_GetFlags(type) & feature) != 0);
#else
    return ((type->tp_flags & feature) != 0);
#endif
}

#define _PyObject_SIZE(typeobj) ( (typeobj)->tp_basicsize )

vstinner · 2020-04-06T12:07:11Z

New changeset 38aefc5 by Victor Stinner in branch 'master':
bpo-40170: PyObject_GET_WEAKREFS_LISTPTR() becomes a function (GH-19377)
38aefc5

aixtools · 2020-04-06T12:55:56Z

Just manually verified that PR19377, when compiled against xlc - crashes during make:

rm -f libpython3.9d.a
ar rcs libpython3.9d.a [Modules/getbuildinfo.o](https://github.com/python/cpython/blob/main/Modules/getbuildinfo.o)  [Parser/acceler.o](https://github.com/python/cpython/blob/main/Parser/acceler.o)  [Parser/grammar1.o](https://github.com/python/cpython/blob/main/Parser/grammar1.o)  [Parser/listnode.o](https://github.com/python/cpython/blob/main/Parser/listnode.o)  [Parser/node.o](https://github.com/python/cpython/blob/main/Parser/node.o)  [Parser/parser.o](https://github.com/python/cpython/blob/main/Parser/parser.o)  [Parser/token.o](https://github.com/python/cpython/blob/main/Parser/token.o)   [Parser/myreadline.o](https://github.com/python/cpython/blob/main/Parser/myreadline.o) [Parser/parsetok.o](https://github.com/python/cpython/blob/main/Parser/parsetok.o) [Parser/tokenizer.o](https://github.com/python/cpython/blob/main/Parser/tokenizer.o)  [Objects/abstract.o](https://github.com/python/cpython/blob/main/Objects/abstract.o)  [Objects/accu.o](https://github.com/python/cpython/blob/main/Objects/accu.o)  [Objects/boolobject.o](https://github.com/python/cpython/blob/main/Objects/boolobject.o)  [Objects/bytes_methods.o](https://github.com/python/cpython/blob/main/Objects/bytes_methods.o)  [Objects/bytearrayobject.o](https://github.com/python/cpython/blob/main/Objects/bytearrayobject.o)  [Objects/bytesobject.o](https://github.com/python/cpython/blob/main/Objects/bytesobject.o)  [Objects/call.o](https://github.com/python/cpython/blob/main/Objects/call.o)  [Objects/capsule.o](https://github.com/python/cpython/blob/main/Objects/capsule.o)  [Objects/cellobject.o](https://github.com/python/cpython/blob/main/Objects/cellobject.o)  [Objects/classobject.o](https://github.com/python/cpython/blob/main/Objects/classobject.o)  [Objects/codeobject.o](https://github.com/python/cpython/blob/main/Objects/codeobject.o)  [Objects/complexobject.o](https://github.com/python/cpython/blob/main/Objects/complexobject.o)  [Objects/descrobject.o](https://github.com/python/cpython/blob/main/Objects/descrobject.o)  [Objects/enumobject.o](https://github.com/python/cpython/blob/main/Objects/enumobject.o)  [Objects/exceptions.o](https://github.com/python/cpython/blob/main/Objects/exceptions.o)  [Objects/genobject.o](https://github.com/python/cpython/blob/main/Objects/genobject.o)  [Objects/fileobject.o](https://github.com/python/cpython/blob/main/Objects/fileobject.o)  [Objects/floatobject.o](https://github.com/python/cpython/blob/main/Objects/floatobject.o)  [Objects/frameobject.o](https://github.com/python/cpython/blob/main/Objects/frameobject.o)  [Objects/funcobject.o](https://github.com/python/cpython/blob/main/Objects/funcobject.o)  [Objects/interpreteridobject.o](https://github.com/python/cpython/blob/main/Objects/interpreteridobject.o)  [Objects/iterobject.o](https://github.com/python/cpython/blob/main/Objects/iterobject.o)  [Objects/listobject.o](https://github.com/python/cpython/blob/main/Objects/listobject.o)  [Objects/longobject.o](https://github.com/python/cpython/blob/main/Objects/longobject.o)  [Objects/dictobject.o](https://github.com/python/cpython/blob/main/Objects/dictobject.o)  [Objects/odictobject.o](https://github.com/python/cpython/blob/main/Objects/odictobject.o)  [Objects/memoryobject.o](https://github.com/python/cpython/blob/main/Objects/memoryobject.o)  [Objects/methodobject.o](https://github.com/python/cpython/blob/main/Objects/methodobject.o)  [Objects/moduleobject.o](https://github.com/python/cpython/blob/main/Objects/moduleobject.o)  [Objects/namespaceobject.o](https://github.com/python/cpython/blob/main/Objects/namespaceobject.o)  [Objects/object.o](https://github.com/python/cpython/blob/main/Objects/object.o)  [Objects/obmalloc.o](https://github.com/python/cpython/blob/main/Objects/obmalloc.o)  [Objects/picklebufobject.o](https://github.com/python/cpython/blob/main/Objects/picklebufobject.o)  [Objects/rangeobject.o](https://github.com/python/cpython/blob/main/Objects/rangeobject.o)  [Objects/setobject.o](https://github.com/python/cpython/blob/main/Objects/setobject.o)  [Objects/sliceobject.o](https://github.com/python/cpython/blob/main/Objects/sliceobject.o)  [Objects/structseq.o](https://github.com/python/cpython/blob/main/Objects/structseq.o)  [Objects/tupleobject.o](https://github.com/python/cpython/blob/main/Objects/tupleobject.o)  [Objects/typeobject.o](https://github.com/python/cpython/blob/main/Objects/typeobject.o)  [Objects/unicodeobject.o](https://github.com/python/cpython/blob/main/Objects/unicodeobject.o)  [Objects/unicodectype.o](https://github.com/python/cpython/blob/main/Objects/unicodectype.o)  [Objects/weakrefobject.o](https://github.com/python/cpython/blob/main/Objects/weakrefobject.o)  [Python/_warnings.o](https://github.com/python/cpython/blob/main/Python/_warnings.o)  [Python/Python-ast.o](https://github.com/python/cpython/blob/main/Python/Python-ast.o)  [Python/asdl.o](https://github.com/python/cpython/blob/main/Python/asdl.o)  [Python/ast.o](https://github.com/python/cpython/blob/main/Python/ast.o)  [Python/ast_opt.o](https://github.com/python/cpython/blob/main/Python/ast_opt.o)  [Python/ast_unparse.o](https://github.com/python/cpython/blob/main/Python/ast_unparse.o)  [Python/bltinmodule.o](https://github.com/python/cpython/blob/main/Python/bltinmodule.o)  [Python/ceval.o](https://github.com/python/cpython/blob/main/Python/ceval.o)  [Python/codecs.o](https://github.com/python/cpython/blob/main/Python/codecs.o)  [Python/compile.o](https://github.com/python/cpython/blob/main/Python/compile.o)  [Python/context.o](https://github.com/python/cpython/blob/main/Python/context.o)  [Python/dynamic_annotations.o](https://github.com/python/cpython/blob/main/Python/dynamic_annotations.o)  [Python/errors.o](https://github.com/python/cpython/blob/main/Python/errors.o)  [Python/frozenmain.o](https://github.com/python/cpython/blob/main/Python/frozenmain.o)  [Python/future.o](https://github.com/python/cpython/blob/main/Python/future.o)  [Python/getargs.o](https://github.com/python/cpython/blob/main/Python/getargs.o)  [Python/getcompiler.o](https://github.com/python/cpython/blob/main/Python/getcompiler.o)  [Python/getcopyright.o](https://github.com/python/cpython/blob/main/Python/getcopyright.o)  [Python/getplatform.o](https://github.com/python/cpython/blob/main/Python/getplatform.o)  [Python/getversion.o](https://github.com/python/cpython/blob/main/Python/getversion.o)  [Python/graminit.o](https://github.com/python/cpython/blob/main/Python/graminit.o)  [Python/hamt.o](https://github.com/python/cpython/blob/main/Python/hamt.o)  [Python/import.o](https://github.com/python/cpython/blob/main/Python/import.o)  [Python/importdl.o](https://github.com/python/cpython/blob/main/Python/importdl.o)  [Python/initconfig.o](https://github.com/python/cpython/blob/main/Python/initconfig.o)  [Python/marshal.o](https://github.com/python/cpython/blob/main/Python/marshal.o)  [Python/modsupport.o](https://github.com/python/cpython/blob/main/Python/modsupport.o)  [Python/mysnprintf.o](https://github.com/python/cpython/blob/main/Python/mysnprintf.o)  [Python/mystrtoul.o](https://github.com/python/cpython/blob/main/Python/mystrtoul.o)  [Python/pathconfig.o](https://github.com/python/cpython/blob/main/Python/pathconfig.o)  [Python/peephole.o](https://github.com/python/cpython/blob/main/Python/peephole.o)  [Python/preconfig.o](https://github.com/python/cpython/blob/main/Python/preconfig.o)  [Python/pyarena.o](https://github.com/python/cpython/blob/main/Python/pyarena.o)  [Python/pyctype.o](https://github.com/python/cpython/blob/main/Python/pyctype.o)  [Python/pyfpe.o](https://github.com/python/cpython/blob/main/Python/pyfpe.o)  [Python/pyhash.o](https://github.com/python/cpython/blob/main/Python/pyhash.o)  [Python/pylifecycle.o](https://github.com/python/cpython/blob/main/Python/pylifecycle.o)  [Python/pymath.o](https://github.com/python/cpython/blob/main/Python/pymath.o)  [Python/pystate.o](https://github.com/python/cpython/blob/main/Python/pystate.o)  [Python/pythonrun.o](https://github.com/python/cpython/blob/main/Python/pythonrun.o)  [Python/pytime.o](https://github.com/python/cpython/blob/main/Python/pytime.o)  [Python/bootstrap_hash.o](https://github.com/python/cpython/blob/main/Python/bootstrap_hash.o)  [Python/structmember.o](https://github.com/python/cpython/blob/main/Python/structmember.o)  [Python/symtable.o](https://github.com/python/cpython/blob/main/Python/symtable.o)  [Python/sysmodule.o](https://github.com/python/cpython/blob/main/Python/sysmodule.o)  [Python/thread.o](https://github.com/python/cpython/blob/main/Python/thread.o)  [Python/traceback.o](https://github.com/python/cpython/blob/main/Python/traceback.o)  [Python/getopt.o](https://github.com/python/cpython/blob/main/Python/getopt.o)  [Python/pystrcmp.o](https://github.com/python/cpython/blob/main/Python/pystrcmp.o)  [Python/pystrtod.o](https://github.com/python/cpython/blob/main/Python/pystrtod.o)  [Python/pystrhex.o](https://github.com/python/cpython/blob/main/Python/pystrhex.o)  [Python/dtoa.o](https://github.com/python/cpython/blob/main/Python/dtoa.o)  [Python/formatter_unicode.o](https://github.com/python/cpython/blob/main/Python/formatter_unicode.o)  [Python/fileutils.o](https://github.com/python/cpython/blob/main/Python/fileutils.o)  [Python/dynload_shlib.o](https://github.com/python/cpython/blob/main/Python/dynload_shlib.o)        [Modules/config.o](https://github.com/python/cpython/blob/main/Modules/config.o)  [Modules/getpath.o](https://github.com/python/cpython/blob/main/Modules/getpath.o)  [Modules/main.o](https://github.com/python/cpython/blob/main/Modules/main.o)  [Modules/gcmodule.o](https://github.com/python/cpython/blob/main/Modules/gcmodule.o)  [Modules/posixmodule.o](https://github.com/python/cpython/blob/main/Modules/posixmodule.o)  [Modules/errnomodule.o](https://github.com/python/cpython/blob/main/Modules/errnomodule.o)  [Modules/pwdmodule.o](https://github.com/python/cpython/blob/main/Modules/pwdmodule.o)  [Modules/_sre.o](https://github.com/python/cpython/blob/main/Modules/_sre.o)  [Modules/_codecsmodule.o](https://github.com/python/cpython/blob/main/Modules/_codecsmodule.o)  [Modules/_weakref.o](https://github.com/python/cpython/blob/main/Modules/_weakref.o)  [Modules/_functoolsmodule.o](https://github.com/python/cpython/blob/main/Modules/_functoolsmodule.o)  [Modules/_operator.o](https://github.com/python/cpython/blob/main/Modules/_operator.o)  [Modules/_collectionsmodule.o](https://github.com/python/cpython/blob/main/Modules/_collectionsmodule.o)  [Modules/_abc.o](https://github.com/python/cpython/blob/main/Modules/_abc.o)  [Modules/itertoolsmodule.o](https://github.com/python/cpython/blob/main/Modules/itertoolsmodule.o)  [Modules/atexitmodule.o](https://github.com/python/cpython/blob/main/Modules/atexitmodule.o)  [Modules/signalmodule.o](https://github.com/python/cpython/blob/main/Modules/signalmodule.o)  [Modules/_stat.o](https://github.com/python/cpython/blob/main/Modules/_stat.o)  [Modules/timemodule.o](https://github.com/python/cpython/blob/main/Modules/timemodule.o)  [Modules/_threadmodule.o](https://github.com/python/cpython/blob/main/Modules/_threadmodule.o)  [Modules/_localemodule.o](https://github.com/python/cpython/blob/main/Modules/_localemodule.o)  [Modules/_iomodule.o](https://github.com/python/cpython/blob/main/Modules/_iomodule.o) [Modules/iobase.o](https://github.com/python/cpython/blob/main/Modules/iobase.o) [Modules/fileio.o](https://github.com/python/cpython/blob/main/Modules/fileio.o) [Modules/bytesio.o](https://github.com/python/cpython/blob/main/Modules/bytesio.o) [Modules/bufferedio.o](https://github.com/python/cpython/blob/main/Modules/bufferedio.o) [Modules/textio.o](https://github.com/python/cpython/blob/main/Modules/textio.o) [Modules/stringio.o](https://github.com/python/cpython/blob/main/Modules/stringio.o)  [Modules/faulthandler.o](https://github.com/python/cpython/blob/main/Modules/faulthandler.o)  [Modules/_tracemalloc.o](https://github.com/python/cpython/blob/main/Modules/_tracemalloc.o) [Modules/hashtable.o](https://github.com/python/cpython/blob/main/Modules/hashtable.o)  [Modules/symtablemodule.o](https://github.com/python/cpython/blob/main/Modules/symtablemodule.o)  [Modules/xxsubtype.o](https://github.com/python/cpython/blob/main/Modules/xxsubtype.o)  [Python/frozen.o](https://github.com/python/cpython/blob/main/Python/frozen.o)
./Modules/makexp_aix [Modules/python.exp](https://github.com/python/cpython/blob/main/Modules/python.exp) . libpython3.9d.a;  xlc_r     -Wl,-bE:Modules/python.exp -lld -o python [Programs/python.o](https://github.com/python/cpython/blob/main/Programs/python.o) libpython3.9d.a -lintl -ldl  -lm   -lm 
 ./python -E -S -m sysconfig --generate-posix-vars ; if test $? -ne 0 ; then  echo "generate-posix-vars failed" ;  rm -f ./pybuilddir.txt ;  exit 1 ;  fi

Objects/genobject.c:127: _PyObject_GC_TRACK: Assertion "!(((PyGC_Head *)(op)-1)->_gc_next != 0)" failed: object already tracked by the garbage collector
Enable tracemalloc to get the memory block allocation traceback
object address : 30084150
object refcount : 0
object type : 20013aa8
object type name: generator
object repr : <refcnt 0 at 30084150>
Fatal Python error: _PyObject_AssertFailed: _PyObject_AssertFailed
Python runtime state: core initialized
Current thread 0x00000001 (most recent call first):
File "<frozen importlib._bootstrap_external>", line 1593 in _setup
File "<frozen importlib._bootstrap_external>", line 1634 in _install
File "<frozen importlib._bootstrap>", line 1189 in _install_external_importers
/bin/sh: 24117648 IOT/Abort trap(coredump)
make: 1254-004 The error code from the last command is 134.
Stop.

FYI: about two hours ago I verified that xlc and 08050e9 : bpo-40147: Fix a compiler warning on Windows in Python/compile.c (GH-19389)

all was green.

aixtools · 2020-04-06T13:22:48Z

Just checked - seems to be SPECIFIC to xlc-v16 as neither xlv-v11 nor xlc-v13 have any issues building.

vstinner · 2020-04-06T15:24:18Z

Py_TRASHCAN_BEGIN() access directly PyTypeObject.tp_dealloc:

#define Py_TRASHCAN_BEGIN(op, dealloc) \
    Py_TRASHCAN_BEGIN_CONDITION(op, \
        Py_TYPE(op)->tp_dealloc == (destructor)(dealloc))

It should use PyType_GetSlot() or a new getter function (to read PyTypeObject.tp_dealloc) should be added.

vstinner · 2020-04-06T15:25:56Z

It should use PyType_GetSlot()

Oh. It seems like currently, PyType_GetSlot() can only be used on a heap allocated types :-( The function starts with:

    if (!PyType_HasFeature(type, Py_TPFLAGS_HEAPTYPE) || slot < 0) {
        PyErr_BadInternalCall();
        return NULL;
    }

vstinner · 2020-04-06T16:52:08Z

Just checked - seems to be SPECIFIC to xlc-v16 as neither xlv-v11 nor xlc-v13 have any issues building.

That sounds like an AIX specific issue. Please open a separated issue.

vstinner · 2020-04-07T22:38:19Z

New changeset 9205520 by Victor Stinner in branch 'master':
bpo-40170: PyObject_NEW() becomes an alias to PyObject_New() (GH-19379)
9205520

vstinner · 2020-04-07T23:13:57Z

New changeset ef5c615 by Victor Stinner in branch 'master':
bpo-40170: Convert PyObject_CheckBuffer() macro to a function (GH-19376)
ef5c615

vstinner · 2020-04-07T23:42:31Z

New changeset 45ec5b9 by Victor Stinner in branch 'master':
bpo-40170: PyType_HasFeature() now always calls PyType_GetFlags() (GH-19378)
45ec5b9

vstinner · 2020-04-08T00:02:05Z

New changeset a15e260 by Victor Stinner in branch 'master':
bpo-40170: Add _PyIndex_Check() internal function (GH-19426)
a15e260

vstinner · 2020-04-08T00:04:00Z

Hum, once most changes will land, maybe it would be worth it to document them at:
https://docs.python.org/dev/whatsnew/3.9.html#build-and-c-api-changes

vstinner · 2020-04-08T00:26:48Z

New changeset 307b9d0 by Victor Stinner in branch 'master':
bpo-40170: Remove PyIndex_Check() macro (GH-19428)
307b9d0

vstinner · 2020-04-10T20:41:49Z

/* Test if an object has a GC head */
#define PyObject_IS_GC(o) \
    (PyType_IS_GC(Py_TYPE(o)) \
     && (Py_TYPE(o)->tp_is_gc == NULL || Py_TYPE(o)->tp_is_gc(o)))

This macro should be converted to an opaque function.

shihai1991 · 2020-04-10T21:09:20Z

This macro should be converted to an opaque function
Can I try it?

vstinner · 2020-04-14T18:11:24Z

New changeset 675d9a3 by Hai Shi in branch 'master':
bpo-40170: Convert PyObject_IS_GC() macro to a function (GH-19464)
675d9a3

vstinner · 2020-04-15T01:38:50Z

On python-dev, Ronald Oussoren asked to add support for the buffer protocol in PyType_FromSpec() and PyTypeSpec API:

Ronald:

BTW. This will require growing the PyTypeSpec ABI a little, there are features you cannot implement using that API for example the buffer protocol.

https://mail.python.org/archives/list/python-dev@python.org/message/PGKRW7S2IUOWVRX6F7RT6VAWD3ZPUDYS/

See also PyType_FromSpec() issue with opaque PyObject:
https://bugs.python.org/issue39573#msg366473

ronaldoussoren · 2020-04-15T08:36:15Z

Something else that probably needs attention with the TypeSpec API is subclassing type in an extension when that subclass adds fields to the type object.

I use this in PyObjC to (dynamically) create types that have some additional data. I could probably work around this issue by adding a level of indirection (basically storing the extra data in a WeakKeyDictionary), but haven't looked into this yet.

shihai1991 · 2020-04-15T14:30:16Z

Py_TRASHCAN_BEGIN() access directly PyTypeObject.tp_dealloc

Looks like this macro not recorded in docs. Do we need using function to replace this macro?

vstinner · 2020-04-15T14:34:39Z

Looks like this macro not recorded in docs.

It never prevented anyone to use a function of the C API :-)

shihai1991 · 2020-04-15T16:23:04Z

It never prevented anyone to use a function of the C API :-)
Got it. If possible someone uses it, I will try to add a function to repalce it(MAYBE udpate this docs too;) )

vstinner · 2020-06-22T08:39:13Z

Py_TRASHCAN_BEGIN() access directly PyTypeObject.tp_dealloc (...) currently, PyType_GetSlot() can only be used on a heap allocated types

I created bpo-41073: [C API] PyType_GetSlot() should accept static types.

vstinner · 2021-02-15T20:42:55Z

Should we strive to fix the cases in Include/internal as well?

No. The internal C API access directly to structure members on purpose, for best performances.

vstinner · 2021-02-16T07:50:11Z

New changeset 871eb42 by Erlend Egeberg Aasland in branch 'master':
bpo-40170: Convert PyDescr_IsData() to static inline function (GH-24535)
871eb42

vstinner · 2021-02-16T15:06:02Z

New changeset cc54001 by Erlend Egeberg Aasland in branch 'master':
bpo-40170: Always define PyIter_Check() as a function (GH-24548)
cc54001

vstinner · 2021-02-16T15:06:46Z

New changeset cc54001 by Erlend Egeberg Aasland in branch 'master':
bpo-40170: Always define PyIter_Check() as a function (GH-24548)

For macOS which doesn't use LTO compiler optimization, we added private static inline functions of some "Check" functions. But I don't think that it's worth it here (I don't think that the function is commonly called in "hot code").

erlend-aasland · 2021-02-16T21:03:33Z

For PyExceptionClass_Name: Is it ok to just remove the macro version (like with #68736)?

vstinner · 2021-02-16T21:28:43Z

For PyExceptionClass_Name: Is it ok to just remove the macro version (like with #68736)?

Yes, I think so.

erlend-aasland · 2021-02-16T22:33:22Z

Thanks, Victor.

For PySequence_ITEM, I guess adding a private C version (for example _PySequence_Item) and redirecting the macro to the C version would be acceptable.

Ditto for PyHeapType_GET_MEMBERS and PyType_SUPPORTS_WEAKREFS.

vstinner · 2021-02-17T10:51:33Z

New changeset cd80f43 by Erlend Egeberg Aasland in branch 'master':
bpo-40170: Always define PyExceptionClass_Name() as a function (GH-24553)
cd80f43

vstinner · 2021-02-17T10:55:54Z

For PySequence_ITEM, I guess adding a private C version (for example _PySequence_Item) and redirecting the macro to the C version would be acceptable.

This can introduce a performance slowdown and so should wait until the PEP-620 is accepted.

erlend-aasland · 2021-02-17T11:32:43Z

This can introduce a performance slowdown and so should wait until the PEP-620 is accepted.

Noted.

vstinner · 2021-02-17T18:15:47Z

New changeset 630264a by Erlend Egeberg Aasland in branch 'master':
bpo-40170: Move 3 NEWS entries to the C API section (GH-24555)
630264a

vstinner · 2021-11-08T16:42:08Z

New changeset 99c7e98 by Victor Stinner in branch 'main':
bpo-40170: Update What's New in Python 3.9 (GH-29470)
99c7e98

miss-islington · 2021-11-08T17:05:22Z

New changeset 69b3de6 by Miss Islington (bot) in branch '3.10':
bpo-40170: Update What's New in Python 3.9 (GH-29470)
69b3de6

vstinner · 2021-11-08T17:10:50Z

New changeset 80580f5 by Miss Islington (bot) in branch '3.9':
bpo-40170: Update What's New in Python 3.9 (GH-29470) (GH-29472)
80580f5

vstinner · 2022-01-27T00:46:40Z

TODO:

Macros still accessing directly PyTypeObject members:
- PyHeapType_GET_MEMBERS()
- PySequence_ITEM()
- _PyObject_SIZE()
- _PyObject_VAR_SIZE()
- PyType_SUPPORTS_WEAKREFS()
Try again to apply "bpo-40170: PyType_HasFeature() now always calls PyType_GetFlags() (bpo-40170: PyType_HasFeature() no longer acccess directly tp_flags #19378)" which has been reverted?
Py_TYPE(obj)->tp_name is still commonly used to format error messages: see rejected bpo-34595

Other TODO tasks which can be addressed in follow-up issues:

Add buffer protocol to PyType_FromSpec()
Stdlib C extensions still define static types: see bpo-40077
3rd party C extensions still define static types: PEP-630 and others propose heap types

vstinner · 2022-01-27T01:07:41Z

I searched for "_PyGC_FINALIZED" in top 5000 PyPI projects. It seems like only Cython is impacted.

ddtrace and guppy3 use directly the internal C API.

== Cython 0.29.26 ==

Cython/Compiler/ModuleNode.py: finalised_check = '!_PyGC_FINALIZED(o)'
Cython/Compiler/ModuleNode.py: '(!PyType_IS_GC(Py_TYPE(o)) || !_PyGC_FINALIZED(o))')

== ddtrace 0.57.3 ==

In ddtrace/profiling/collector/stack.pyx:

    IF PY_MINOR_VERSION >= 9:
        # Needed for accessing _PyGC_FINALIZED when we build with -DPy_BUILD_CORE
        cdef extern from "<internal/pycore_gc.h>":
            pass

== guppy3-3.1.2 ==

In src/heapy/hv.c:

#if PY_MAJOR_VERSION >= 3 && PY_MINOR_VERSION >= 9
# define Py_BUILD_CORE
/* PyGC_Head */
#  undef _PyGC_FINALIZED
#  include <internal/pycore_gc.h>
# undef Py_BUILD_CORE
#endif

vstinner · 2022-01-27T01:10:55Z

I searched for "_PyObject_DebugMallocStats" in top 5000 PyPI projects. There is a single project using it: guppy3.

Extract of guppy3 src/heapy/xmemstats.c:

...
    dlptr__PyObject_DebugMallocStats = addr_of_symbol("_PyObject_DebugMallocStats");
...
static PyObject *
hp_xmemstats(PyObject *self, PyObject *args)
{
    if (dlptr__PyObject_DebugMallocStats) {
        fprintf(stderr, "======================================================================\n");
        fprintf(stderr, "Output from _PyObject_DebugMallocStats()\n\n");
        dlptr__PyObject_DebugMallocStats(stderr);
    }
    ...
}

addr_of_symbol() is implemented with dlsym() or GetModuleHandle(NULL)+GetProcAddress(): it searchs for the symbol in the current process.

vstinner · 2022-01-27T01:16:56Z

In the top 5000 PyPI projects, the _PyObject_SIZE() and _PyObject_VAR_SIZE() functions are used by 7 projects:

Cython-0.29.26
frozendict-2.2.0: implement "sizeof" function, found in copies of Objects/dictobject.c file
JPype1-1.3.0
numpy-1.22.1: gentype_alloc() in numpy/core/src/multiarray/scalartypes.c.src, used as type tp_alloc functions
pickle5-0.0.12
pyobjc-core-8.2
recordclass-0.17.1

vstinner · 2022-01-27T02:01:06Z

New changeset af32b3e by Victor Stinner in branch 'main':
bpo-40170: PyType_SUPPORTS_WEAKREFS() becomes a regular function (GH-30938)
af32b3e

vstinner · 2022-01-27T02:35:55Z

New changeset 6b491b9 by Victor Stinner in branch 'main':
bpo-40170: Remove _Py_GetAllocatedBlocks() function (GH-30940)
6b491b9

vstinner · 2022-01-27T20:23:38Z

New changeset 0575551 by Victor Stinner in branch 'main':
bpo-40170: Move _Py_GetAllocatedBlocks() to pycore_pymem.h (GH-30943)
0575551

vstinner · 2022-01-27T23:40:09Z

New changeset 18ea973 by Victor Stinner in branch 'main':
bpo-40170: Remove PyHeapType_GET_MEMBERS() macro (GH-30942)
18ea973

vstinner · 2022-01-28T14:19:20Z

Changes already done:

Macros converted to regular functions:
- PyObject_GET_WEAKREFS_LISTPTR(); add internal inline _PyObject_GET_WEAKREFS_LISTPTR()
- PyType_SUPPORTS_WEAKREFS(); add internal inline _PyType_SUPPORTS_WEAKREFS()
- PyObject_CheckBuffer(); no fast internal API
- PyObject_IS_GC(); no fast internal API
- PyDescr_IsData(); no fast internal API
Always implemented as a function, remove macro optimization in the non-limited API:
- PyIter_Check(); no fast internal API
- PyIndex_Check(); add internal inline _PyIndex_Check()
- PyExceptionClass_Name(); no fast internal API
Py_TRASHCAN_BEGIN() macro now calls _PyTrash_cond() function: no longer read directly tp_dealloc member
PyObject_NEW() macro becomes an alias to PyObject_New()
Remove PyHeapType_GET_MEMBERS() (move it to the internal C API)

PyType_HasFeature() is left unchanged since the change caused performance issue on macOS, whereas Python is not built with LTO.

vstinner · 2022-01-28T14:25:55Z

I close the issue. While this issue is not fully fixed, it's a milestone of my "stable ABI" goal. I prefer to address remaining issues in new separted issues.

This issues is not fully fixed, there are are still a 4 macros which access directly PyTypeObject members:

PySequence_ITEM()
_PyObject_SIZE()
_PyObject_VAR_SIZE()
PyType_HasFeature() (if Py_LIMITED_API is not defined)

PySequence_ITEM() and PyType_HasFeature() are important for performance, I prefer to have a PEP before changing these two functions.

_PyObject_SIZE() and _PyObject_VAR_SIZE() should be made public with a better name like PyObject_SizeOf() and PyVarObject_SizeOf(). But I prefer to do that in a separated issue.

IMO it would be interesting to merge the PyHeapTypeObject structure into the PyTypeObject structure:
https://bugs.python.org/issue46433#msg411167

And make the PyTypeObject opaque: deprecate static types and promote the usage of heap types in C extensions. That's a big project which may be splitted into multiple issues and the final change may need its own PEP.

vstinner added 3.9 only security fixes topic-C-API labels Apr 3, 2020

vstinner added 3.10 only security fixes and removed 3.9 only security fixes labels Jun 22, 2020

vstinner changed the title ~~[C API] Make PyTypeObject structure an opaque structure in the public C API~~ [C API] Prepare PyTypeObject structure for a stable ABI: avoid accessing members in the public API Jan 28, 2022

vstinner closed this as completed Jan 28, 2022

ezio-melotti transferred this issue from another repository Apr 10, 2022

vstinner mentioned this issue Jun 21, 2023

C API: Investigate how the PyTypeObject members can be removed from the public C API #105970

Closed

[C API] Prepare PyTypeObject structure for a stable ABI: avoid accessing members in the public API #84351

[C API] Prepare PyTypeObject structure for a stable ABI: avoid accessing members in the public API #84351

Comments

vstinner commented Apr 3, 2020

vstinner commented Apr 3, 2020

vstinner commented Apr 4, 2020

vstinner commented Apr 6, 2020

aixtools commented Apr 6, 2020

aixtools commented Apr 6, 2020

vstinner commented Apr 6, 2020

vstinner commented Apr 6, 2020

vstinner commented Apr 6, 2020

vstinner commented Apr 7, 2020

vstinner commented Apr 7, 2020

vstinner commented Apr 7, 2020

vstinner commented Apr 8, 2020

vstinner commented Apr 8, 2020

vstinner commented Apr 8, 2020

vstinner commented Apr 10, 2020

shihai1991 commented Apr 10, 2020

vstinner commented Apr 14, 2020

vstinner commented Apr 15, 2020

ronaldoussoren commented Apr 15, 2020

shihai1991 commented Apr 15, 2020

vstinner commented Apr 15, 2020

shihai1991 commented Apr 15, 2020

vstinner commented Jun 22, 2020

vstinner commented Feb 15, 2021

vstinner commented Feb 16, 2021

vstinner commented Feb 16, 2021

vstinner commented Feb 16, 2021

erlend-aasland commented Feb 16, 2021

vstinner commented Feb 16, 2021

erlend-aasland commented Feb 16, 2021

vstinner commented Feb 17, 2021

vstinner commented Feb 17, 2021

erlend-aasland commented Feb 17, 2021

vstinner commented Feb 17, 2021

vstinner commented Nov 8, 2021

miss-islington commented Nov 8, 2021

vstinner commented Nov 8, 2021

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 27, 2022

vstinner commented Jan 28, 2022

vstinner commented Jan 28, 2022