From owner-freebsd-net  Mon Mar  5 12: 4:47 2001
Delivered-To: freebsd-net@freebsd.org
Received: from zibbi.icomtek.csir.co.za (zibbi.icomtek.csir.co.za [146.64.24.58])
	by hub.freebsd.org (Postfix) with ESMTP id 6B91A37B718
	for <freebsd-net@FreeBSD.ORG>; Mon,  5 Mar 2001 12:04:40 -0800 (PST)
	(envelope-from jhay@zibbi.icomtek.csir.co.za)
Received: (from jhay@localhost)
	by zibbi.icomtek.csir.co.za (8.11.1/8.11.1) id f25K3vd32980;
	Mon, 5 Mar 2001 22:03:57 +0200 (SAT)
	(envelope-from jhay)
From: John Hay <jhay@icomtek.csir.co.za>
Message-Id: <200103052003.f25K3vd32980@zibbi.icomtek.csir.co.za>
Subject: Re: kernel: nd6_storelladdr failed, mbuf leak
In-Reply-To: <005901c0a5a1$c54476d0$becbca18@jehovah> from Bosko Milekic at "Mar 5, 2001 01:26:17 pm"
To: bmilekic@technokratis.com (Bosko Milekic)
Date: Mon, 5 Mar 2001 22:03:57 +0200 (SAT)
Cc: freebsd-net@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL54 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-net@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

> 
> > I have configured a 4-stable machine to be a router, routing ipv4,
> ipv6
> > and ipx. To be able to do Ethernet_II framing on one interface and
> 802.3
> > on another I have used if_ef.ko.
> >
> > I then noticed that "... kernel: nd6_storelladdr failed" gets logged
> > often and after a while all mbufs are used. It turned out that in
> > sys/net/if_ethersubr.c in ether_output() when nd6_storelladdr()
> fails,
> > it does a return(0) and does not free the mbuf. I checked -current
> > and it is still like that.
> 
>     It should not be freeing the mbuf, because that mbuf is being
> passed as an argument to ether_output(). It is typically the caller
> that ought to be responsible for freeing the mbuf in this case.

The purpose of ether_output() is normally to get the packet (mbuf)
onto the correct interface queue so that it will be sent out. As such
whoever calls ether_output() is not going to do anything with that
mbuf again. If ether_output don't put it on a queue, it should free
it.

Well maybe I was a bit vague about the exact place. I claim that the
return(0) at line 188 of if_ethersubr.c in -stable and line 189 in
-current should be "goto bad" or something that will either use the
mbuf or do a m_freem() and not just return and leave the mbuf in the
air.

The code looks something like this:

        case AF_INET6:
                if (!nd6_storelladdr(&ac->ac_if, rt, m, dst, (u_char *)edst)) {
                        /* this must be impossible, so we bark */
                        printf("nd6_storelladdr failed\n");
===>                    return(0);
                }

> 
> > Now the reason it fails is that the ef(4) device use an ifp->if_type
> > (IFT_XETHER) that nd6_storelladdr() does not expect.
> >
> > Oh as a workaround I have configured route6d to ignore fxp1f0.
> >

John
-- 
John Hay -- John.Hay@icomtek.csir.co.za

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