ctypes not converting None to Null in 64-bit system

[wenkats]

BPO	7505
Nosy	@theller, @pitrou, @briancurtin, @skrah
Files	test.py

^{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 = 'https://github.com/theller'
closed_at = <Date 2010-08-04.03:40:25.468>
created_at = <Date 2009-12-14.17:10:40.511>
labels = ['ctypes', 'type-crash']
title = 'ctypes not converting None to Null in 64-bit system'
updated_at = <Date 2010-08-04.03:40:25.468>
user = 'https://bugs.python.org/wenkats'

bugs.python.org fields:

activity = <Date 2010-08-04.03:40:25.468>
actor = 'terry.reedy'
assignee = 'theller'
closed = True
closed_date = <Date 2010-08-04.03:40:25.468>
closer = 'terry.reedy'
components = ['ctypes']
creation = <Date 2009-12-14.17:10:40.511>
creator = 'wenkat_s'
dependencies = []
files = ['15557']
hgrepos = []
issue_num = 7505
keywords = []
message_count = 5.0
messages = ['96385', '99127', '99134', '99162', '99286']
nosy_count = 5.0
nosy_names = ['theller', 'pitrou', 'brian.curtin', 'skrah', 'wenkat_s']
pr_nums = []
priority = 'normal'
resolution = 'out of date'
stage = 'test needed'
status = 'closed'
superseder = None
type = 'crash'
url = 'https://bugs.python.org/issue7505'
versions = ['Python 2.5']

[wenkats]

The attached code is failing to convert None to Null. It is successful
if the int* is shifted towards the beginning of argument list. The bug
is actually in 32/64 bit pointers as the code works on a 32 bit system.

The bug disappears if I reduce the number of arguments to less than 10.

I created the so using the command - "gcc -shared -fPIC -o libtest.so
libtest.c", (gcc (Debian 4.3.2-1.1) 4.3.2)

The c file is as follows:

libtest.c
#include <stdint.h>
#include <stdio.h>


int32_t where ( int32_t a, int32_t c, int32_t d, int32_t e, int32_t f, 
int32_t g, int32_t h, int32_t *b, int32_t i, int32_t j, int32_t k, 
int32_t l)
{
    printf("b = %p\n", (void *)b);
    printf("a = %d, c = %d, d = %d, e = %d, f = %d, g = %d, h = %d, i = 
%d, j = %d, k = %d,  l = %d", a,c,d,e,f,g,h,i,j,k,l);

return 100;

} /* where() */

Python Code:
import ctypes

libtest = ctypes.CDLL('./libtest.so')
where = libtest.where
where.restype = ctypes.c_int
where.argtypes = [ 
                    ctypes.c_int32, ctypes.c_int32, ctypes.c_int32,          
ctypes.c_int32, ctypes.c_int32, ctypes.c_int32, ctypes.c_int32,
                    ctypes.POINTER(ctypes.c_int32),
                    ctypes.c_int32, ctypes.c_int32, ctypes.c_int32, 
ctypes.c_int32,
                 ]


a = 7
b = None

status = where(a,a,a,a,a,a,a,b,a,a,a,a)

[briancurtin]

What is the output of your example? I'm not on Linux at the moment, but I'm trying to see if this is actually a crash.

[wenkats]

The output was a valid pointer which was not null. I did not investigate further on this issue, because I had a work around i.e I used ctypes.POINTER(ctypes.c_cint)() instead of None.

[pitrou]

It works in 64-bit mode under Mandriva Linux (gcc 4.4.1), with Python 2.6, 2.7 and 3.2.

$ python issue7505.py 
b = (nil)
a = 7, c = 7, d = 7, e = 7, f = 7, g = 7, h = 7, i = 7, j = 7, k = 7,  l = 7

I also works with a hand-compiled Python (2.6, 2.7, 3.2) under a 64-bit Ubuntu box. However, it fails with the system Python 2.5 shipped with Ubuntu. So perhaps this has to do with how Debian/Ubuntu compile their Pythons, or their libffi.

$ python issue7505.py 
b = 0x7f5300000000
a = 7, c = 7, d = 7, e = 7, f = 7, g = 7, h = 7, i = 7, j = 7, k = 7,  l = 7

[skrah]

Confirm that 2.5 is the only troublesome version, but also when compiled from source (Ubuntu 64-bit, 2.5.5).