Products & Services

Contents

Summary
Affected Products
Details
Vulnerability Scoring Details
Impact
Software Versions and Fixes
Workarounds
Obtaining Fixed Software
Exploitation and Public Announcements
Status of this Notice: FINAL
Distribution
Revision History
Cisco Security Procedures

Products that cache DNS responses and process DNS messages with the recursion desired (RD) flag set may be vulnerable to a DNS cache poisoning attack depending on implementation of the DNS protocol. Products that process DNS messages with the RD flag set will attempt to answer the question asked on behalf of the client. A product is only affected if using a vulnerable implementation of the DNS protocol, the DNS server functionality for the product is enabled, and the DNS feature for the product is configured to process recursive DNS query messages.

Vulnerable Products

Products Confirmed Not Vulnerable

The Domain Name System is an integral part of networks that are based on TCP/IP such as the Internet. Simply stated, the Domain Name System is a hierarchical database that contains mappings of hostnames and IP addresses. The DNS protocol is part of the TCP/IP protocol suite and allows DNS clients to query the DNS database to resolve hostnames to IP addresses.

A DNS server is an application that implements the DNS protocol and that has the ability to respond to queries made by DNS clients. When handling a query from a DNS client, a DNS server can look into its portion of the global DNS database (if the query is for a portion of the DNS database for which the DNS server is authoritative), or it can relay the query to other DNS servers (if it is configured to do so and if the query is for a portion of the DNS database for which the DNS server is not authoritative.)

Because of the processing time and bandwidth that is associated with handling a DNS query, most DNS servers locally store responses that are received from other DNS servers. The area where these responses are stored locally is called a "cache." Once a response is stored in a cache, the DNS server can use the locally stored response for a certain time (called the "time to live") before having to query DNS servers again to refresh the local (cached) copy of the response.

A DNS cache poisoning attack is an attack in which an entry in the DNS cache of a DNS server is changed so the IP address associated with a hostname in the cache does not point to the correct place. For example, if www.example.com is mapped to the IP address 192.168.0.1 and this mapping is present in the cache of a DNS server, an attacker who succeeds in poisoning the DNS cache of this server may be able to map www.example.com to 10.0.0.1 instead. If this happens, a user who is trying to visit www.example.com may end up contacting the wrong web server.

Although DNS cache poisoning attacks are not new, a security researcher recently presented a technique that allows an attacker to mount successful DNS cache poisoning attacks with low complexity tools and low traffic requirements. This technique exploits a weakness in most implementations of the DNS protocol. The fundamental implementation weakness is that the DNS transaction ID and source port number used to validate DNS responses are not sufficiently randomized and can easily be predicted, which allows an attacker to create forged responses to DNS queries that will match the expected values. The DNS server will consider such responses to be valid.

The following Cisco products that offer DNS server functionality have been found to be susceptible to DNS cache poisoning attacks:

• Cisco IOS Software: The vulnerability documented in Cisco bug ID CSCso81854 ( registered customers only) .
  
• Cisco Network Registrar: The vulnerability documented in Cisco bug ID CSCsq01298 ( registered customers only) .
  
• Cisco Application and Content Networking System (ACNS): The vulnerability documented in Cisco bug ID CSCsq21930 ( registered customers only) .
  

This vulnerability has been assigned Common Vulnerabilities and Exposures (CVE) ID CVE-2008-1447.

Port Address Translation Considerations

Port Address Translation (PAT) is a form of Network Address Translation (NAT) that allows multiple hosts in a private network to access a public network using a single, public IP address. This is accomplished by rewriting layer 4 information, specifically TCP and UDP source port numbers and checksums, as packets from the private network traverse a network device that is performing PAT. PAT is configured by network administrators and performed by network devices such as firewalls and routers in situations where public IP addresses are limited.

After the initial multi-vendor DNS advisory was published on July 8th, 2008 it was discovered that in some cases the fixes to DNS implementations to use random source ports when sending DNS queries could be negated when such queries traverse PAT devices. The reason for this is that in these cases the network device performing PAT uses a predictable source port allocation policy, such as incremental allocation, when performing the layer 4 rewrite operation that is necessary for PAT. Under this scenario, the fixes made by DNS vendors can be greatly diminished because, while DNS queries seen on the inside network have random source port numbers, the same queries have potentially predictable source port numbers when they leave the private network, depending on the type of traffic that transits through the device.

Several Cisco products are affected by this issue, and if DNS servers are deployed behind one of these affected products operating in PAT mode then the DNS infrastructure may still be at risk even if source port randomization updates have been applied to the DNS servers.

The affected Cisco products, and the respective Cisco bugs that have been created to track the issue, are the following:

Fixed software information for these bugs will not be added to this document. Instead, customers should use their regular support channels or the bug tracking features of the Bug Toolkit application on cisco.com to obtain fixed software information.

The above products use an incremental source port allocation policy when performing the source port rewrite operation that is needed for PAT. In the case of Cisco IOS, the original source port will be tried first, but if that port is already allocated and in use for an existing PAT translation then a new port will be incrementally assigned.

Note that traditional NAT, i.e. allocating one public IP address for each private IP address, is not affected by this problem because, unlike PAT, NAT only rewrites layer 3 information and does not modify layer 4 header information of packets traversing the NAT device.

Cisco has provided scores for the vulnerabilities in this advisory based on the Common Vulnerability Scoring System (CVSS). The CVSS scoring in this Security Advisory is done in accordance with CVSS version 2.0.

CVSS is a standards-based scoring method that conveys vulnerability severity and helps determine urgency and priority of response.

Cisco has provided a base and temporal score. Customers can then compute environmental scores to assist in determining the impact of the vulnerability in individual networks.

Cisco has provided an FAQ to answer additional questions regarding CVSS at

http://www.cisco.com/web/about/security/intelligence/cvss-qandas.html.

Cisco has also provided a CVSS calculator to help compute the environmental impact for individual networks at

http://intellishield.cisco.com/security/alertmanager/cvss.

Successful exploitation of the vulnerability described in this document may result in invalid hostname-to-IP address mappings in the cache of an affected DNS server. This may lead users of this DNS server to contact the wrong provider of network services. The ultimate impact varies greatly, ranging from a simple denial of service (for example, making www.example.com resolve to 127.0.0.1) to phishing and financial fraud.

When considering software upgrades, also consult http://www.cisco.com/go/psirt and any subsequent advisories to determine exposure and a complete upgrade solution.

In all cases, customers should exercise caution to be certain the devices to be upgraded contain sufficient memory and that current hardware and software configurations will continue to be supported properly by the new release. If the information is not clear, contact the Cisco Technical Assistance Center (TAC) or your contracted maintenance provider for assistance.

Cisco IOS Software

Each row of the Cisco IOS Software table (below) names a Cisco IOS Software release train. If a given release train is vulnerable, then the earliest possible releases that contain the fix (along with the anticipated date of availability for each, if applicable) are listed in the "First Fixed Release" column of the table. The "Recommended Release" column indicates the releases which have fixes for all the published vulnerabilities at the time of this Advisory. A device running a release in the given train that is earlier than the release in a specific column (less than the First Fixed Release) is known to be vulnerable. Cisco recommends upgrading to a release equal to or later than the release in the "Recommended Releases" column of the table.