[NEWS] IGMP Denial of Service Vulnerability
From: support@securiteam.comDate: 06/15/02
- Previous message: support@securiteam.com: "[NT] Microsoft SQL Server 2000 pwdencrypt() Buffer Overflow"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
From: support@securiteam.com To: list@securiteam.com Date: Sat, 15 Jun 2002 21:28:20 +0200 (CEST)
The following security advisory is sent to the securiteam mailing list, and can be found at the SecuriTeam web site: http://www.securiteam.com
- - promotion
When was the last time you checked your server's security?
How about a monthly report?
http://www.AutomatedScanning.com - Know that you're safe.
- - - - - - - - -
IGMP Denial of Service Vulnerability
------------------------------------------------------------------------
SUMMARY
IGMP, in a bid to optimize its functionality uses a report suppression
mechanism to prevent redundant IGMP member reports from being sent to the
quicker router. We refer the reader to RFC 2236 for a better understanding
of the IGMP protocol. The router on seeing the first membership report for
a group from a host in the subnet its serving works towards extending the
distribution tree for a Multicast group G all the way to itself. This
report suppression mechanism can be exploited as explained below.
DETAILS
Attack:
We consider different scenarios in which such an attack can be launched.
In these scenarios, we tested each of the following operating systems for
vulnerability to the attacks : Redhat Linux 7.3, Microsoft Windows 98, and
Microsoft Windows XP. We found all of the above listed operating systems
to be vulnerable.
Scenario 1
Host H1 and H2 are connected to a router R using a hub. Host H1 is a
member of the multicast group 230.0.0.1 and is receiving traffic from the
designated router R. R sends out periodic IGMP membership query messages
soliciting for membership reports from the hosts in the network it is
serving. Host H1 in response to the membership query message multicasts a
membership report to the same group. Now H2 starts the attack. H2 is
sniffing the network for membership query messages from R. On receiving
the message, H2 unicasts a membership report to H1. H1 on receiving the
message infers that some other host on the network is also interested in
receiving traffic and suppresses its reports. Little does H1 know that
there is something fishy going on. It would know this if it checked the
destination ethernet address on the report it received. The address would
have been its own and not a multicast ethernet address.
Effectively, now R doesn't receive any membership reports for the group
230.0.0.1 and hence blocks all traffic related to that group from flowing
into the subnet. What has just been described is a denial of service
attack against H1. This attack is partially unsuccessful if some other
hosts on the network also subscribe to the same multicast group. However,
H2 on seeing membership reports from those hosts can unicast the reports
to those hosts separately. This would cause all those hosts to suppress
their reports.
Scenario 2
Host H1, Host H2 and other hosts are connected to the router in a switched
environment. In such a network, host H2, the attacker receives membership
queries and membership reports periodically sent out by other hosts to the
group 230.0.0.1. The attacker can then send membership reports directly
addressed to the ethernet addresses of each host in the network, again
launching a subnet wide denial of service attack.
Scenario 3
Host H1, Host H2 and other hosts are connected to the router in a switched
environment with IGMP snooping enabled. In such an environment the
attacker will not hear membership reports sent out by other hosts on the
network. To launch a denial of service attack, the attacker will have to
methodically send membership reports to each and every host on the
network. This would result in the same denial of service attack as
detailed in the two scenarios given above.
Solution:
To counter such an attack, a host on receiving an IGMP packet, should
check the MAC address. If it is not a multicast ethernet address i.e. with
the prefix 01:00:5E, the host must drop the packet. This fix in the
handling of IGMP packets is an effective solution to the denial of service
attack.
Linux Patch
For Linux 2.4.18, we patched the kernel source against the denial of
service attack. Here is the patched igmp.c file that needs to be compiled
into the boot image. The changes made to the source are given here. The
igmp.c file is located in /usr/src/linux*/net/ipv4 directory.
Note please that the changes only check if the ethernet address is a valid
Multicast/Broadcast address or not (basically checks if the broadcast bit
in the MAC address is set). The code does not check for the correctness of
the ethernet address i.e. checks if the ethernet address corresponds to
the Multicast group IP). Having said that, this change is a sufficient
change to evade such attacks. However, we believe that tighter checks
should be made on operating systems running both at the host and the
router.
The changes should work as is for other kernel versions as well. However,
we haven't tested the same against other versions.
ADDITIONAL INFORMATION
The information has been provided by <mailto:krishna@cs.ucsb.edu> Krishna
Ramachandran, Arun Qamra, Mohit Sang.
========================================
This bulletin is sent to members of the SecuriTeam mailing list.
To unsubscribe from the list, send mail with an empty subject line and body to: list-unsubscribe@securiteam.com
In order to subscribe to the mailing list, simply forward this email to: list-subscribe@securiteam.com
====================
====================
DISCLAIMER:
The information in this bulletin is provided "AS IS" without warranty of any kind.
In no event shall we be liable for any damages whatsoever including direct, indirect, incidental, consequential, loss of business profits or special damages.
- Previous message: support@securiteam.com: "[NT] Microsoft SQL Server 2000 pwdencrypt() Buffer Overflow"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
- RE: FW: Legal? Road Runner proactive scanning.[Scanned]
... you lock the door and windows. ... If your host is on the internet I
consider it public and knocking on the ... Port scanning is not an attack it is probe.
... someone else's network without permission. ... (Security-Basics) - False positives & the nwfusion article
... That is a really good point Tom. ... But they didn't test RS Sentry, but the
older RS Network Sensor on Nokia. ... > "We considered an attack to be any compromise
of any ... > host, the IDS should report on it. ... (Focus-IDS) - RE: any recommendable anti-ddos solution?
... With DDOS you cannot simply block a host, DDOS is originating from lots of ...
different subnets on different geographic locations, so blocking a host ... attack,
for example if I know you have an IPS system that denies traffic ... and the switch that goes
to everything else inside the network. ... (Security-Basics) - Re: ESX Vmware Physically connected to different segments
... Most of them had at least one "full compromise of host by attack from guest"
in them. ... While there are fun ways to attack network vlans to ... (Pen-Test) - Re: Target based IDS review and discussion in Information Security
... > 1) A URL attack is seen by the sensor affecting Windows IIS. ... > each
and every step we took to investigate the attack (from IDS ... > impacted host
to manually verify if the attack was successful or not. ... Automated forensics are useful
and a nice step forward but if the ... (Focus-IDS)