From owner-freebsd-net  Fri Jun  2 11:50:31 2000
Delivered-To: freebsd-net@freebsd.org
Received: from yagosys.com (mail.yagosys.com [207.135.89.130])
	by hub.freebsd.org (Postfix) with ESMTP id 6A6ED37B709
	for <freebsd-net@FreeBSD.ORG>; Fri,  2 Jun 2000 11:50:28 -0700 (PDT)
	(envelope-from shiva@yagosys.com)
Received: from yagosys.com by yagosys.com (8.8.8+Sun/SMI-SVR4-Yago)
	id LAA29721; Fri, 2 Jun 2000 11:48:28 -0700 (PDT)
Message-ID: <393801D4.84C60B59@yagosys.com>
Date: Fri, 02 Jun 2000 11:49:56 -0700
From: Shiva Shenoy <shiva@yagosys.com>
Reply-To: shiva@yagosys.com
Organization: Cabletron Systems, Inc.
X-Mailer: Mozilla 4.7 [en] (X11; I; SunOS 5.5.1 sun4u)
X-Accept-Language: en
MIME-Version: 1.0
To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
Cc: freebsd-net@FreeBSD.ORG
Subject: Re: [Q] Clarification regarding RTF_PRCLONING and RTF_CLONING.
References: <3937F28A.5168FABC@yagosys.com> <200006021822.OAA42179@khavrinen.lcs.mit.edu>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-net@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

Garrett Wollman wrote:
> 
> <<On Fri, 02 Jun 2000 10:44:42 -0700, Shiva Shenoy <shiva@yagosys.com> said:
> 
> > 1. What is the difference between RTF_PRCLONING and RTF_CLONING
> >    I see that these 2 attributes are mutually exclusive.
> 
> One is enabled by the protocol and the other is enabled by the
> interface or by a routing process.  The former was a hack perpetrated
> by yours truly to make the routing table behave like a per-host
> cache.  I know better now, but I haven't had the time or inclination
> over the past four years to sit down and fix it.
> 
> > 2. Do ARP entries, cloned off of interface routes, get deleted
> >    when an interface route is brought down or deleted?
> 
> I don't recall.  I think there was some code to support that, but I
> don't know whether it was actually effective or not.
>



(All code snippets below are from net/route.c)

The code that may have been there is under RTM_DELETE. This call
here after the route has been deleted was intended to do that, I think.
rnh->rnh_walktree_from(rnh, dst, netmask,
                        rt_fixdelete, rt);

When an interface route is blown away, arp entries are not removed.
Reason: rt_fixdelete() - the first line precludes arp entries, cloned
off 
of the interface route from passing the test. 
The parent route is NULL for arp entries!
    if (rt->rt_parent == rt0 && !(rt->rt_flags & RTF_PINNED)) {

This can be fixed by changing the line in rtrequest: 

 if ((*ret_nrt)->rt_flags & RTF_PRCLONING) {
    rt->rt_parent = (*ret_nrt);

to: 

 if ((*ret_nrt)->rt_flags & (RTF_CLONING | RTF_PRCLONING)) {
    rt->rt_parent = (*ret_nrt);

Do you see any serious problem in this logic?


> > 3. How come ARP rtentries do not have rt_parent set to the
> >    interface routes that they are cloned off of? In fact they
> >    are set to NULL.
> 
> Because they are created from RTF_CLONING routes and not RTF_PRCLONING
> routes.  This may actually be a bug.

Thanks again for your feedback.

-Shiva Shenoy


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message