Wrong computation of max_fd on OpenBSD

[cedk]

BPO	23852
Nosy	@gpshead, @cedk, @skrah, @worr, @applio
Files	fd_dir.patch max_fd.patch max_fd.patch max_fd.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 = None
closed_at = <Date 2015-04-26.06:47:00.108>
created_at = <Date 2015-04-02.15:13:52.380>
labels = ['extension-modules', 'type-bug']
title = 'Wrong computation of max_fd on OpenBSD'
updated_at = <Date 2015-04-26.09:19:49.100>
user = 'https://github.com/cedk'

bugs.python.org fields:

activity = <Date 2015-04-26.09:19:49.100>
actor = 'berker.peksag'
assignee = 'none'
closed = True
closed_date = <Date 2015-04-26.06:47:00.108>
closer = 'gregory.p.smith'
components = ['Extension Modules']
creation = <Date 2015-04-02.15:13:52.380>
creator = 'ced'
dependencies = []
files = ['38807', '38824', '38831', '39200']
hgrepos = []
issue_num = 23852
keywords = ['patch']
message_count = 17.0
messages = ['239920', '240002', '240005', '240013', '240017', '240020', '240023', '240068', '240075', '240076', '240619', '240772', '240929', '241035', '241394', '241982', '242047']
nosy_count = 6.0
nosy_names = ['gregory.p.smith', 'ced', 'skrah', 'python-dev', 'worr', 'davin']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue23852'
versions = ['Python 3.4', 'Python 3.5']

[cedk]

The test_subprocess fails since bpo-21618 on OpenBSD because the FD_DIR is wrong (/dev/fd instead of /proc/self/fd).

[skrah]

There's a comment in _posixsubprocess:

"NetBSD and OpenBSD have a /proc fs available (though not necessarily
mounted) and do not have fdescfs for /dev/fd."

Is this still valid?

[cedk]

At least on OpenBSD procfs have been removed: http://www.openbsd.org/faq/current.html#20140908

[gpshead]

I'm not going to bother setting up a VM with an esoteric OS in it myself, if someone knows the past and current state of various OpenBSD versions and what to do there, feel free to commit OpenBSD conditional compilation bits as you see fit to make this do the right thing. :)

[cedk]

Indeed, I think my patch is not right as /dev/fd on OpenBSD is static.
I will continue to investigate.

[cedk]

The problem comes from safe_get_max_fd which return a too low value because of a bug in sysconf on OpenBSD [1]:

The value for _SC_STREAM_MAX is a minimum maximum, and required to be the same as ANSI C's FOPEN_MAX, so the returned value is a ridiculously small and misleading number.

[1] http://www.openbsd.org/cgi-bin/man.cgi/OpenBSD-current/man3/sysconf.3

[cedk]

Here is a patch that uses getrlimit (that's works on OpenBSD) before using sysconf.

[skrah]

Unfortunately I don't have an OpenBSD install either. From the
sysconf.c source ...

http://cvsweb.openbsd.org/cgi-bin/cvsweb/~checkout~/src/lib/libc/gen/sysconf.c?rev=1.22&content-type=text/plain

... it seems that sysconf(_SC_OPEN_MAX) also calls getrlimit().

[cedk]

But sysconf(_SC_OPEN_MAX) uses rlim_cur which is too low instead of rlim_max. My proposal is indeed describe in msg219477, it is not prefect but at least better than the current one for OpenBSD.

[cedk]

Correctly cast to long instead of int.