This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

classification
Title: accept() on AF_UNIX sockets broken on arm as of 2.5.3
Type: Stage:
Components: Extension Modules Versions: Python 2.5
process
Status: closed Resolution: wont fix
Dependencies: Superseder:
Assigned To: Nosy List: hmoffatt, loewis
Priority: normal Keywords:

Created on 2009-01-14 03:48 by hmoffatt, last changed 2022-04-11 14:56 by admin. This issue is now closed.

Files
File name Uploaded Description Edit
accept_test.py hmoffatt, 2009-01-14 03:48
Messages (5)
msg79825 - (view) Author: Hamish Moffatt (hmoffatt) Date: 2009-01-14 03:48
As of 2.5.3, calling accept() on an AF_UNIX socket fails on arm. The
following exception occurs:

SystemError: Negative size passed to PyString_FromStringAndSize

I have attached a test program that creates a socket and connects to it.
It works on x86-64 but fails on armeb (gcc 4.1.2).

This failure has been noted a few times by the arm buildbot too:
http://archives.free.net.ph/message/20081012.121034.89069ea2.en.html

It's already fixed in 2.6.x in rev 60214. I applied the same change to
2.5.4 and the problem was resolved.
msg79836 - (view) Author: Martin v. Löwis (loewis) * (Python committer) Date: 2009-01-14 06:33
No further bug fixes are accepted for Python 2.5. Closing this as "won't
fix".
msg79839 - (view) Author: Hamish Moffatt (hmoffatt) Date: 2009-01-14 06:52
Thanks for your feedback, but I'm a bit confused by the note on the web
site:

"This is the last bugfix release of Python 2.5. Python 2.5 is now in
bugfix-only mode; no new features are being added."

This implies further bug fixes are possible (and further fixes were just
made to 2.4 despite the above comment about 2.4.5). 

Also it's a regression from 2.5.2, which must be of some concern?
msg79840 - (view) Author: Martin v. Löwis (loewis) * (Python committer) Date: 2009-01-14 07:25
> "This is the last bugfix release of Python 2.5. Python 2.5 is now in
> bugfix-only mode; no new features are being added."

Thanks for pointing that out; I have fixed the page now.

> Also it's a regression from 2.5.2, which must be of some concern?

It is; and it is unfortunate that this wasn't fixed before the release.
However, support for AF_UNIX sockets on ARM is not important enough
to warrant another bug fix release.
msg79842 - (view) Author: Hamish Moffatt (hmoffatt) Date: 2009-01-14 07:55
Understood.

Would it be appropriate to list this issue on the known bugs page?
http://www.python.org/download/releases/2.5.4/bugs/

I can't reproduce it on x86, but it's probably not ARM specific. It
might even be a compiler bug. I tried the same version on x86 and x86-64
and didn't have a problem.

Thanks.
History
Date User Action Args
2022-04-11 14:56:44adminsetgithub: 49192
2009-01-14 07:55:10hmoffattsetmessages: + msg79842
2009-01-14 07:25:15loewissetmessages: + msg79840
2009-01-14 06:52:12hmoffattsetmessages: + msg79839
2009-01-14 06:33:56loewissetstatus: open -> closed
resolution: wont fix
messages: + msg79836
nosy: + loewis
2009-01-14 03:48:25hmoffattcreate