cvs commit: src/sys/netinet ip_carp.c

[Yar Tikhiy]

On Thu, Oct 27, 2005 at 07:20:00PM -0400, Tom Rhodes wrote:
> On Wed, 26 Oct 2005 14:15:47 +0200
> Max Laier <max at love2party.net> wrote:
> 
> > On Wednesday 26 October 2005 11:58, Robert Watson wrote:
> > > On Wed, 26 Oct 2005, Gleb Smirnoff wrote:
> > > > On Wed, Oct 26, 2005 at 10:15:09AM +0100, Robert Watson wrote:
> > > > R> I think we may actually be in need of either a new flag,
> > > > R> IFF_OKSODONTTREATTHISQUITELIKEANINTERFACE, or maybe a more reliable
> > > > way R> for protocols to ask if an interface is a loopback interface or
> > > > not.
> > > >
> > > > I'd prefer to rewrite those subsystems that use interface layer but
> > > > aren't actually interfaces. I have plans to do this for CARP.
> > >
> > > At least in the case of if_disc, this won't help.  I'm not quite sure why
> > > if_disc is IFF_LOOPBACK.
> > 
> > Sad answer seems to be: copy and paste.  IFF_LOOPBACK is part of 1.1 which 
> > also contains the following comment:
> > 
> > /*
> >  * Discard interface driver for protocol testing and timing.
> >  * (Based on the loopback.)
> >  */
> > 
> > So it might be a good idea to get rid of it and work from there.
> 
> IIRC, someone told me or I read somewhere that if_disc is
> somewhat based on lo(4).  There may be more sections where
> code is similar - rotted - over time.

My browsing through if_disc.c revealed none of them except for
already noticed IFF_LOOPBACK, but I'll take another look when
going to commit the result of erasing IFF_LOOPBACK.

-- 
Yar