NetBSD Problem Report #10495

Received: (qmail 20244 invoked from network); 2 Jul 2000 17:56:46 -0000
Message-Id: <200007021756.KAA19772@digital.clock.org>
Date: Sun, 2 Jul 2000 10:56:40 -0700 (PDT)
From: "Erik E. Fair" <fair@digital.clock.org>
Reply-To: fair@netbsd.org
To: gnats-bugs@gnats.netbsd.org
Subject: arplookup: unable to enter address for 169.254.173.186 (host is not on local network)
X-Send-Pr-Version: 3.95

>Number:         10495
>Category:       kern
>Synopsis:       arplookup: unable to enter address for 169.254.173.186 (host is not on local network)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    is
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jul 02 17:57:00 +0000 2000
>Closed-Date:    Tue Feb 24 10:01:14 +0000 2009
>Last-Modified:  Tue Feb 24 10:01:14 +0000 2009
>Originator:     Erik E. Fair
>Release:        1.4.2
>Organization:
International Organization of Internet Clock Watchers
>Environment:

System: NetBSD digital.clock.org 1.4.2A NetBSD 1.4.2A (DIGITAL) #2: Tue May 23 17:26:00 PDT 2000 root@digital.clock.org:/usr/obj/sys/arch/alpha/compile/DIGITAL alpha


>Description:
	This diagnostic is incomplete. The Ethernet MAC address
	must be reported along with the bogus IP address. Otherwise,
	finding the misconfigured host requires iteration over
	every host attached to the network.

>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:

From: Andrew Brown <atatat@atatdot.net>
To: "Erik E. Fair" <fair@digital.clock.org>
Cc: gnats-bugs@gnats.netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: kern/10495: arplookup: unable to enter address for 169.254.173.186 (host is not on local network)
Date: Tue, 4 Jul 2000 13:47:18 -0400

 >>Description:
 >	This diagnostic is incomplete. The Ethernet MAC address
 >	must be reported along with the bogus IP address. Otherwise,
 >	finding the misconfigured host requires iteration over
 >	every host attached to the network.

 it might also be nice to log the receiving interface, and maybe even
 the destination hardware address (unicast vs. multicast
 vs. broadcast).

 -- 
 |-----< "CODE WARRIOR" >-----|
 codewarrior@daemon.org             * "ah!  i see you have the internet
 twofsonet@graffiti.com (Andrew Brown)                that goes *ping*!"
 andrew@crossbar.com       * "information is power -- share the wealth."
State-Changed-From-To: open->analyzed 
State-Changed-By: is 
State-Changed-When: Mon Nov 20 04:20:57 PST 2000 
State-Changed-Why:  
Adding the arp-internal addresses, and the interface info, to the log is 
easy. 
Adding the Ethernet/ARCnet/whatever header to the log is not easy, because 
the ARP handling entities don't get it passed currently. Workaround is to 
tcpdump (when you suspect evil senders). 
Two possible changes to implement the latter solutions: make the individual 
drivers (that is, if_ethersubr, if_tokensubr, if_fddisubr, if_arc) complain 
when the inner information doesn't match the outer, OR 
change the interface between driver and ARP module to pass both (or rather, 
pointers to the source and destination addresses) so that the ARP module 
can do the sanity checks and logging. 


Responsible-Changed-From-To: kern-bug-people->is 
Responsible-Changed-By: is 
Responsible-Changed-When: Mon Nov 20 04:20:57 PST 2000 
Responsible-Changed-Why:  
I'll look at this. 
State-Changed-From-To: analyzed->feedback 
State-Changed-By: is 
State-Changed-When: Sun Mar 9 05:20:47 PST 2003 
State-Changed-Why:  
We have been logging the "inner" (claimed by the sender) link level address 
and the receiving interface along with the network level address which could 
not be entered for some time now. 

This would address your PR as long as you're checking for misconfigured 
hosts, rather than spoofing attacks. 

However, as a determined spoofer will also change the outer MAC address in 
the packet, I don't think that we should mix layers to provide that 
information. As a workaround, it can be analyzed with a simple tcpdump  
expression. 

Does this satisfy your concerns? 
-is 
State-Changed-From-To: feedback->closed 
State-Changed-By: perry 
State-Changed-When: Fri Apr 4 14:08:25 PST 2003 
State-Changed-Why:  
It appears that i.s. has added sufficient logging to largely alleviate 
the problem, and nearly a month has elapsed since the last request 
for feedback. 

From: "Erik E. Fair" <fair@clock.org>
To: perry@netbsd.org, gnats-bugs@netbsd.org
Cc: is@netbsd.org
Subject: Re: kern/10495 
Date: Fri, 04 Apr 2003 14:18:02 -0800

 The standard I use is three months, actually.

 I think the argument about ethernet MAC addresses being spoofable
 is specious; my request for that information to be logged was not
 to attempt to identify an attacker, but to identify misconfigured
 hosts which are unlikely to be using changed MAC addresses.

 Please reopen the PR. I don't feel satisfied yet.

 	Erik <fair@clock.org>

From: "Perry E. Metzger" <perry@piermont.com>
To: "Erik E. Fair" <fair@clock.org>
Cc: gnats-bugs@netbsd.org, is@netbsd.org
Subject: Re: kern/10495
Date: 04 Apr 2003 17:23:53 -0500

 "Erik E. Fair" <fair@clock.org> writes:
 > I think the argument about ethernet MAC addresses being spoofable
 > is specious; my request for that information to be logged was not
 > to attempt to identify an attacker, but to identify misconfigured
 > hosts which are unlikely to be using changed MAC addresses.

 And if you noticed, that information is indeed now available.

 Quoting is:

   We have been logging the "inner" (claimed by the sender) link level address 
   and the receiving interface along with the network level address which could 
   not be entered for some time now. 

   This would address your PR as long as you're checking for misconfigured 
   hosts, rather than spoofing attacks. 

 Since you are looking for misconfigured hosts, the fact that the
 inner information is logged is sufficient, is it not?

 Perry

From: "Perry E. Metzger" <perry@piermont.com>
To: "Erik E. Fair" <fair@clock.org>
Cc: gnats-bugs@netbsd.org, is@netbsd.org
Subject: Re: kern/10495
Date: 04 Apr 2003 17:28:27 -0500

 "Erik E. Fair" <fair@clock.org> writes:
 > Hm. I misunderstood, then. I thought Ignatious was saying that
 > getting the MAC address logged involved a layer violation and was
 > "too hard". What's the cvs log entry?

 No idea -- apparently the change is quite old.

From: "Erik E. Fair" <fair@clock.org>
To: "Perry E. Metzger" <perry@piermont.com>
Cc: gnats-bugs@netbsd.org, is@netbsd.org
Subject: Re: kern/10495 
Date: Fri, 04 Apr 2003 14:26:39 -0800

 Hm. I misunderstood, then. I thought Ignatious was saying that
 getting the MAC address logged involved a layer violation and was
 "too hard". What's the cvs log entry?

 	Erik <fair@clock.org>
State-Changed-From-To: closed->open 
State-Changed-By: perry 
State-Changed-When: Sat Apr 5 12:58:19 PST 2003 
State-Changed-Why:  
Ignatios is going to make some further improvements to this. See the 
audit trail. 

From: Ignatios Souvatzis <is@netbsd.org>
To: Cc: "Perry E. Metzger" <perry@piermont.com>, gnats-bugs@netbsd.org
Subject: Re: kern/10495
Date: Sat, 5 Apr 2003 22:55:46 +0200

 On Fri, Apr 04, 2003 at 05:28:27PM -0500, Perry E. Metzger wrote:
 > 
 > "Erik E. Fair" <fair@clock.org> writes:
 > > Hm. I misunderstood, then. I thought Ignatious was saying that
 > > getting the MAC address logged involved a layer violation and was
 > > "too hard". What's the cvs log entry?
 > 
 > No idea -- apparently the change is quite old.

 Perry, I think you misunderstood the code ... or I misunderstood what 
 Eric wants.

 Eric wants to log the outer MAC address, the MAC address of the real
 sender of the ARP $OP.

 We are logging the hardware address in the layer 3 of the machine on behalf
 of which the ARP $OP was sent.

 This is never the same for a proxy arp reply, and I can see why it would be
 useful to log this information in case of problems.

 Back then, I didn't add code to log this because the the ARP code doesn't
 get the Ethernet/ARCnet/Token Ring/FDDI header - it has no way to know
 the length of the outer header, so ethersubr and arcsubr etc. pass it an mbuf
 shortened to only contain the inner packet, with all the necessary information
 for normal operation.

 My original proposal was to write a tcpdump expression to capture outer
 source address != inner sha events, in case of suspected problems.

 Unfortunately, this will catch any proxy arp reply, unless you limit it,
 in addition, to the kernel-logged problem event addresses and hope that
 they happen again. 

 I can see that this is not optimal in case of problems in production
 systems... 

 Now, I browsed the code, and found that we have nowadays a way to add 
 auxilliary, tagged-by-type, data to the mbuf with the pkthdr.

 My proposal is to write some code so that, for some option set, the 
 ethersubr, arcsubr, etc. code will pass an mbuf with the packet source
 link level address to the upper layers as one of the aux data structurs.

 The ARP code would then look if this is available in the case of of problems
 and log it, too (else log "N/A").

 I propose to do it generic so that other parts of the networking code
 can do the same sort of logging if needed. I can imagine that the IPv6 
 neigbour discovery code would want to log this if available, same with
 appletalk arp).

 Do I read Erik's desire correctly? If yes, I'll outline my proposal to 
 tech-net for comments and search for an implementor.

 Regards,
 	Ignatios
 -- 
 seal your e-mail: http://www.gnupg.org/
State-Changed-From-To: open->closed
State-Changed-By: is@NetBSD.org
State-Changed-When: Tue, 24 Feb 2009 10:01:14 +0000
State-Changed-Why:
Should the desired functionality be needed, it can be implemented as 
a BPF expression - no need to implement this inside the kernel as 
outlined 69 months ago has been mentioned since then.


>Unformatted:

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.