On 21/02/15 21:34, Chris Johns wrote:
On 20/02/2015 11:00 pm, Sebastian Huber wrote:
On 20/02/15 12:18, Chris Johns wrote:
On 20/02/2015 8:29 pm, Sebastian Huber wrote:
On 20/02/15 10:06, Nick Withers wrote:
On Fri, 2015-02-20 at 08:59 +0100, Sebastian Huber wrote:
>Hello Nick,
>
>thanks for the patch. I checked in a slightly different version
which
>ignores the return status completely.
Hi Sebastian,
Your commit effectively reverts Joel's changes in
https://git.rtems.org/rtems/commit/?id=3211e8e9c0565fae8b13ba0115e3a23ad3ae8ade
Would chucking a (void) in front of the mkdir() shut Coverity up?
Ok, I should have checked the commit history since I wondered why I
added this error check in the first place. Lets try a (void) cast.
What about doing a stat, create the directory if not present and then
error on a mkdir failure ? There are other failures that could happen
making a directory that we should correctly report.
The stat approach is not thread safe.
Its up to the users of the create files to do the error handling.
Maybe just check the error code for EEXIST and ignore it, plus the
call is protected by a posix init once variable so I am not sure I
follow this thread safe bit.
Chris
The purpose of this function is to provide default passwd and group
file. At this point it doesn't matter if this fails. Its up to the users
of these files to decide what to do if they don't exist or have the
wrong format. You can for example delete these files right after the
pgdgrp_init() or unplug the network cable if you use NFS.
The stat() function is pretty worthless since after the check everything
might have changed.
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.hu...@embedded-brains.de
PGP : Public key available on request.
Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
