From owner-freebsd-net  Fri Apr 30  7:21: 1 1999
Delivered-To: freebsd-net@freebsd.org
Received: from labinfo.iet.unipi.it (labinfo.iet.unipi.it [131.114.9.5])
	by hub.freebsd.org (Postfix) with SMTP id AD57214F45
	for <net@freebsd.org>; Fri, 30 Apr 1999 07:20:42 -0700 (PDT)
	(envelope-from luigi@labinfo.iet.unipi.it)
Received: from localhost (luigi@localhost) by labinfo.iet.unipi.it (8.6.5/8.6.5) id LAA14634; Fri, 30 Apr 1999 11:25:52 +0200
From: Luigi Rizzo <luigi@labinfo.iet.unipi.it>
Message-Id: <199904300925.LAA14634@labinfo.iet.unipi.it>
Subject: Re: possible bug in udp_usrreq ?
To: luigi@labinfo.iet.unipi.it (Luigi Rizzo)
Date: Fri, 30 Apr 1999 11:25:52 +0200 (MET DST)
Cc: net@freebsd.org
In-Reply-To: <199904300918.LAA14603@labinfo.iet.unipi.it> from "Luigi Rizzo" at Apr 30, 99 11:18:35 am
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Content-Length: 2385      
Sender: owner-freebsd-net@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

Whoops...
i correct myself: the bug appears to be also in the code in the
Stevens book, the reference figure for the multicast case is 23.26
(I am Bcc-ing Richard just in case he knows already...)

and maybe the fix is slightly different than the one i propose,
something like

    m->m_len -= iphdrlen + sizeof (struct udphdr) ;
    m->m_pkthdr.len -= iphdrlen + sizeof (struct udphdr) ;
    m->m_data += iphdrlen + sizeof (struct udphdr) ;

because we can't (probably) touch iphdrlen since it is already
incremented in the unicast section (unless they are mutually
exclusive).

	cheers
	luigi

> Hi,
> 
> i just noticed a possible bug in udp_usrreq.c:udp_input()
> 
> When demuxing datagrams to udp socket, near line 199 of the file,
> there is the following section of code to skip the ip and udp
> headers:
> 
> 	/*
> 	 * Construct sockaddr format source address.
> 	 */
> 	udp_in.sin_port = uh->uh_sport;
> 	udp_in.sin_addr = ip->ip_src;
>  -->	m->m_len -= sizeof (struct udpiphdr);
>  -->	m->m_data += sizeof (struct udpiphdr);
> 
> note, there is no update to m->m_pkthdr.len such as
> 
> 	m->m_pkthdr.len -= sizeof (struct udpiphdr);
> 
> which in my opinion should be there, as it is instead done in the
> Stevens TCPIP/Ill.vol.2 pg 775 (fig.23.25) and also in the section
> of code related to unicast datagrams near line 313:
> 
>         iphlen += sizeof(struct udphdr);
>         m->m_len -= iphlen;
>         m->m_pkthdr.len -= iphlen;
>         m->m_data += iphlen;
> 
> Actually, looking at the differences, the multicast section of the
> code looks really broken and unable to handle ip options. It should
> be exactly the same as in the unicast case.
> 
> Just for curiosity, i cheched in the CVS tree and all revisions of
> udp_input() seems to have the same problem. Even my old 1.1.5
> machine also shows the same bug, so i wonder when the problem
> came out, maybe someone with access to older Berkeley sources can
> see when that happened ?
> 
> 	cheers
> 	luigi
> -----------------------------------+-------------------------------------
>   Luigi RIZZO                      .
>   EMAIL: luigi@iet.unipi.it        . Dip. di Ing. dell'Informazione
>   HTTP://www.iet.unipi.it/~luigi/  . Universita` di Pisa
>   TEL/FAX: +39-050-568.533/522     . via Diotisalvi 2, 56126 PISA (Italy)
> -----------------------------------+-------------------------------------
> 



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