On Fri, Feb 20, 2015 at 01:22:15PM -0500, Russell Bryant wrote:
> strncpy() has pretty annoying semantics. One bit that is odd is that
> it does not guarantee that the destination string will be null
> terminated. If no terminator is found in the first N bytes of the
> source, it just fills the destination buffer and doesn't terminate it.
>
> The changes here all just subtract one from the number of characters
> allowed to be copied, always leaving one last byte in the destination
> to ensure that the result is null terminated. In all cases, the
> source has been initialized with memset() prior to calling strncpy().
>
> BSD variants provide strlcpy() which behaves more sanely in this
> regard, but this change seems to be the smallest change to address the
> problem.
>
> Signed-off-by: Russell Bryant <rbry...@redhat.com>
Thanks for the report, and the patch.
All of these, and in the next patch too, can use ovs_strlcpy()
instead. It's good to add a makefile rule to keep new uses from
creeping over time.
I sent out a pair of patches to do both of those:
http://openvswitch.org/pipermail/dev/2015-February/051458.html
http://openvswitch.org/pipermail/dev/2015-February/051457.html
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
