PYLONG_BITS_IN_DIGIT differs between MinGW and MSVC

@pfmoore

BPO	35037
Nosy	@pfmoore, @mdickinson, @scoder, @tjguk, @zware, @serhiy-storchaka, @zooba, @BenjaminLiuPenrose
PRs	bpo-35037 Fix issue in cython project: Cython files don't compile on Mingw-w64 64-bit #19326

^{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 = None
created_at = <Date 2018-10-21.08:22:52.766>
labels = ['3.9', '3.10', '3.11', 'extension-modules', 'build', 'OS-windows']
title = 'PYLONG_BITS_IN_DIGIT differs between MinGW and MSVC'
updated_at = <Date 2021-12-22.13:16:07.818>
user = 'https://github.com/scoder'

bugs.python.org fields:

activity = <Date 2021-12-22.13:16:07.818>
actor = 'mark.dickinson'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Extension Modules', 'Windows']
creation = <Date 2018-10-21.08:22:52.766>
creator = 'scoder'
dependencies = []
files = []
hgrepos = []
issue_num = 35037
keywords = ['patch']
message_count = 9.0
messages = ['328197', '328198', '328202', '328203', '328204', '328206', '328341', '404733', '409034']
nosy_count = 8.0
nosy_names = ['paul.moore', 'mark.dickinson', 'scoder', 'tim.golden', 'zach.ware', 'serhiy.storchaka', 'steve.dower', 'Beier Liu']
pr_nums = ['19326']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = 'compile error'
url = 'https://bugs.python.org/issue35037'
versions = ['Python 3.9', 'Python 3.10', 'Python 3.11']

Linked PRs

gh-79218: Define MS_WIN64 macro for Mingw-w64 64bit on Windows #100137

I see reports from Cython users on Windows-64 that extension modules that use "longintrepr.h" get miscompiled by MinGW. A failing setup looks as follows:

Stock 64 bit CPython on Windows, e.g.
Python 3.6.3 (v3.6.3:2c5fed8, Oct 3 2017, 18:11:49) [MSC v.1900 64 bit (AMD64)]

MinGW uses this compile time setup, i.e. 15 bit digits:
PyLong_BASE 0x8000
PyLong_MASK 7FFF
PyLong_SHIFT 15
sizeof(digit) 2
sizeof(sdigit) 2

Whereas CPython reports the following, indicating that it was in fact built with 30 bit digits:
sys.getsize(1, 2**14, 2**15, 2**29, 2**30, 2**63, 2**64) (28, 28, 28, 28, 32, 36, 36)

I'm not sure if this also applies to Py2.7, but I don't think the PyLong implementations differ in this regard.

The compile time PyLong digit size is not remembered by the CPython installation and instead determined on the fly by the extension compiler. It seems that the MSVC build of the stock CPython packages differs from what MinGW decides here. This renders MinGW unusable for code that uses "longintrepr.h", which Cython does by default in order to accelerate its unpacking of PyLong values.

Is there a reason why CPython cannot store its compile time value for the PYLONG_BITS_IN_DIGIT setting somewhere?

Some more information. CPython uses this code snippet to decide on the PyLong digit size:

#ifndef PYLONG_BITS_IN_DIGIT
#if SIZEOF_VOID_P >= 8
#define PYLONG_BITS_IN_DIGIT 30
#else
#define PYLONG_BITS_IN_DIGIT 15
#endif
#endif

In MinGW, "SIZEOF_VOID_P" is 4, whereas "sizeof(void*)" resolves to 8.

Is there a reason why CPython cannot store its compile time value for the PYLONG_BITS_IN_DIGIT setting somewhere?

There's PyLong_GetInfo [1], and at Python level, the corresponding sys.long_info (Python 2) / sys.int_info (Python 3). Does that help?

[1]

cpython/Objects/longobject.c

Lines 5578 to 5595 in b2e2025

    
           PyObject * 
        
           PyLong_GetInfo(void) 
        
           { 
        
               PyObject* int_info; 
        
               int field = 0; 
        
               int_info = PyStructSequence_New(&Int_InfoType); 
        
               if (int_info == NULL) 
        
                   return NULL; 
        
               PyStructSequence_SET_ITEM(int_info, field++, 
        
                                         PyLong_FromLong(PyLong_SHIFT)); 
        
               PyStructSequence_SET_ITEM(int_info, field++, 
        
                                         PyLong_FromLong(sizeof(digit))); 
        
               if (PyErr_Occurred()) { 
        
                   Py_CLEAR(int_info); 
        
                   return NULL; 
        
               } 
        
               return int_info; 
        
           }

At runtime this information is available as sys.int_info.bits_per_digit.

In MinGW, "SIZEOF_VOID_P" is 4, whereas "sizeof(void*)" resolves to 8.

Looks like a misconfiguration. I suppose this will cause more issues than just wrong PYLONG_BITS_IN_DIGIT.

There's PyLong_GetInfo ...

Thanks, I understand that this information can be found at runtime. However, in order to correctly compile C code against a given CPython runtime, there needs to be a guarantee that extension module builds use the same configuration as the CPython build.

Misconfiguration

I searched some more, and it looks like this was already reported almost ten years ago in bpo-4709.

The work-around there is to pass -DMS_WIN64, which apparently is not defined by MinGW but required by the CPython headers.

That variable should be inferred somewhere, either in PC/pyconfig.h or pyport.h. If there's a way to also infer it on 64-bit MinGW then we would consider a patch for that.

The relevant macro seems to be "__MINGW64__". I have neither a Windows environment nor MinGW-64 for testing, but the logic should be the same as for the "_WIN64" macro, i.e.

#if defined(_WIN64) || defined(__MINGW64__)
#define MS_WIN64
#endif

And then there's probably also something to do to record the compiler version in the long CPython version string, etc. I also can't say if MinGW-64 sets the "MS_WIN32" macro. Having "MS_WIN64" without "MS_WIN32" might be unexpected for some code.

This should probably be a separate issue, but I wonder whether the 15-bit digit option has value any more. Should we just drop that option and always use 30-bit digits?

30-bit digits were introduced at a time when we couldn't rely on a 64-bit integer type always being available, so we still needed to keep the 15-bit fallback option. But I think that's no longer the case.

This should probably be a separate issue,

Specifically, bpo-45569.

is this still relevant now that we always default to 30-bit?

@gpshead

@gpshead I guess the PYLONG_BITS_IN_DIGIT issue still exists in theory, but it would be harder to hit it by accident at this point. The wider issue of SIZEOF_VOID_P being wrong seems like it would still be relevant.

gh-79218: Define `MS_WIN64` macro for Mingw-w64 64bit on Windows (GH-…

Merge branch 'main' into typedocs

Navigation Menu

Repository navigation

PYLONG_BITS_IN_DIGIT differs between MinGW and MSVC #79218

Linked PRs

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

	PyObject *
	PyLong_GetInfo(void)
	{
	PyObject* int_info;
	int field = 0;
	int_info = PyStructSequence_New(&Int_InfoType);
	if (int_info == NULL)
	return NULL;
	PyStructSequence_SET_ITEM(int_info, field++,
	PyLong_FromLong(PyLong_SHIFT));
	PyStructSequence_SET_ITEM(int_info, field++,
	PyLong_FromLong(sizeof(digit)));
	if (PyErr_Occurred()) {
	Py_CLEAR(int_info);
	return NULL;
	}
	return int_info;
	}

PYLONG_BITS_IN_DIGIT differs between MinGW and MSVC #79218

Description

Linked PRs

Activity

scoder commented on Oct 21, 2018

scoder commented on Oct 21, 2018

mdickinson commented on Oct 21, 2018

serhiy-storchaka commented on Oct 21, 2018

scoder commented on Oct 21, 2018

zooba commented on Oct 21, 2018

scoder commented on Oct 23, 2018

mdickinson commented on Oct 22, 2021

mdickinson commented on Dec 22, 2021

gpshead commented on Aug 5, 2022

mdickinson commented on Aug 6, 2022

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions