Date: Fri, 21 Sep 2012 01:00:44 -0700 From: Anuranjan Shukla <anshukla@juniper.net> To: Julian Elischer <julian@freebsd.org> Cc: "freebsd-net@freebsd.org" <freebsd-net@freebsd.org> Subject: Re: Proposal for changes to network device drivers and network stack (RFC) Message-ID: <CC816A36.1A297%anshukla@juniper.net> In-Reply-To: <504A1755.6090902@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 9/7/12 8:48 AM, "Julian Elischer" <julian@freebsd.org> wrote:
>>>> struct socket {
>>>>
>>>> int so_fibnum; /* routing domain for this socket */
>>>> uint32_t so_user_cookie;
>>>> + u_int so_oqueue; /* manage send prioritizing based on
>>>> application
>>>> needs */
>>>> + u_short so_lrid; /* logical routing */
>>>> };
>>>>
>>>
>We have the second one with the SETFIB socket option, which allows
>a socket to select the routing instance (fib) to use.
>
>it's in the diff, 3 lines up.
Thanks Julian. I had wondered if our usage of so_lrid was different enough
to justify another field, primarily because I wasn't sure about potential
clash with setfib() for the process (since Junos doesn't use it). Having
looked a bit more into setfib() after your comment, it looks ok for Junos
to use so_fibnum. I'll remove so_lrid from our diff.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CC816A36.1A297%anshukla>