_posixsubprocess module broken on x32

[vapier]

BPO	15000
Nosy	@gpshead, @amauryfa, @vapier
Files	python-3.2.3-x32.patch

^{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/gpshead'
closed_at = <Date 2012-06-05.20:32:09.345>
created_at = <Date 2012-06-04.18:07:37.206>
labels = ['extension-modules']
title = '_posixsubprocess module broken on x32'
updated_at = <Date 2012-06-05.20:37:32.646>
user = 'https://github.com/vapier'

bugs.python.org fields:

activity = <Date 2012-06-05.20:37:32.646>
actor = 'Arfrever'
assignee = 'gregory.p.smith'
closed = True
closed_date = <Date 2012-06-05.20:32:09.345>
closer = 'gregory.p.smith'
components = ['Extension Modules']
creation = <Date 2012-06-04.18:07:37.206>
creator = 'vapier'
dependencies = []
files = ['25819']
hgrepos = []
issue_num = 15000
keywords = ['patch']
message_count = 8.0
messages = ['162281', '162310', '162312', '162313', '162317', '162366', '162369', '162372']
nosy_count = 5.0
nosy_names = ['gregory.p.smith', 'amaury.forgeotdarc', 'vapier', 'Arfrever', 'python-dev']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = None
url = 'https://bugs.python.org/issue15000'
versions = ['Python 3.2', 'Python 3.3']

[vapier]

the direct call to the getdents syscall is broken on x32. there, the first two args are not unsigned long, but unsigned long long. patch attached to fix the issue.

[gpshead]

What is x32?

[gpshead]

Answering my own question I believe x32 is the new ABI being worked on as described in http://lwn.net/Articles/456731/.

In that light, your patch makes sense assuming the __ILP32__ define is something gcc's cpp will always be defining when targeting x32. I will apply it.

[gpshead]

for reference - https://sites.google.com/site/x32abi/

[vapier]

$ echo | gcc -m32 -E -P -dD - | grep LP
<nothing>
$ echo | gcc -m64 -E -P -dD - | grep LP
#define _LP64 1
#define __LP64__ 1
$ echo | gcc -mx32 -E -P -dD - | grep LP
#define _ILP32 1
#define __ILP32__ 1

[amauryfa]

Would a uint64_t be recommended in this case?

[vapier]

a uint64_t would fix it for x86_64, but break it most 32bit systems as sizeof(unsigned long) == 32bit for them

[python-dev]

New changeset aa9cfeea07ad by Gregory P. Smith in branch '3.2':
Fixes issue bpo-15000: support the odd x32 abi on posixsubprocess's system call.
http://hg.python.org/cpython/rev/aa9cfeea07ad

New changeset 5a4d5e714d08 by Gregory P. Smith in branch 'default':
Fixes issue bpo-15000: support the odd x32 abi on posixsubprocess's system call.
http://hg.python.org/cpython/rev/5a4d5e714d08