Randy
Verified User
Published: 2007-07-24,
Last Updated: 2007-07-24 12:14:41 UTC
by Johannes Ullrich (Version: 1)
Amit Klein wrote about a paper he just released with details about a BIND 9 cache poisoning issue. This is one of the problems addressed by the latest version of BIND 9.
The very brief summary: BIND prior to version 9.4.1-P1 did not use a strong algorithm to create DNS transaction IDs. As a result, one can derive the next transaction ID BIND will use by knowning the last few transaction IDs. In this case, up to 15 queries are used.
Once the attacker knows the "state" of the targets BIND install, it is possible to forge a response. DNS uses UDP by default. Each query sent by the DNS server includes a random transaction ID. The server responding to the query will include this transaction ID so the querying DNS server knows what query is answered by this particular response. BIND always uses the same source port for its queries.
The attack appears to be quite feasible. Probably the main difficulty will be to get the spoofed packet routed. But unless the attackers network implements strict egress filtering, this is very much a feasible attack. Best to patch your BIND server soon.
CVE: CVE-2007-2926
Versions affected: BIND 9.0 (all versions)
BIND 9.1 (all versions)
BIND 9.2.0, 9.2.1, 9.2.2, 9.2.3, 9.2.4, 9.2.5, 9.2.6, 9.2.7, 9.2.8
BIND 9.3.0, 9.3.1, 9.3.2, 9.3.3, 9.3.4
BIND 9.4.0, 9.4.1
BIND 9.5.0a1, 9.5.0a2, 9.5.0a3, 9.5.0a4, 9.5.0a5
Not vulnerable: BIND 9.2.8-P1, BIND 9.3.4-P1, BIND 9.4.1-P1 or BIND 9.5.0a6
For details, see www.trusteer.com/docs/bind9dns.html
Last Updated: 2007-07-24 12:14:41 UTC
by Johannes Ullrich (Version: 1)
Amit Klein wrote about a paper he just released with details about a BIND 9 cache poisoning issue. This is one of the problems addressed by the latest version of BIND 9.
The very brief summary: BIND prior to version 9.4.1-P1 did not use a strong algorithm to create DNS transaction IDs. As a result, one can derive the next transaction ID BIND will use by knowning the last few transaction IDs. In this case, up to 15 queries are used.
Once the attacker knows the "state" of the targets BIND install, it is possible to forge a response. DNS uses UDP by default. Each query sent by the DNS server includes a random transaction ID. The server responding to the query will include this transaction ID so the querying DNS server knows what query is answered by this particular response. BIND always uses the same source port for its queries.
The attack appears to be quite feasible. Probably the main difficulty will be to get the spoofed packet routed. But unless the attackers network implements strict egress filtering, this is very much a feasible attack. Best to patch your BIND server soon.
CVE: CVE-2007-2926
Versions affected: BIND 9.0 (all versions)
BIND 9.1 (all versions)
BIND 9.2.0, 9.2.1, 9.2.2, 9.2.3, 9.2.4, 9.2.5, 9.2.6, 9.2.7, 9.2.8
BIND 9.3.0, 9.3.1, 9.3.2, 9.3.3, 9.3.4
BIND 9.4.0, 9.4.1
BIND 9.5.0a1, 9.5.0a2, 9.5.0a3, 9.5.0a4, 9.5.0a5
Not vulnerable: BIND 9.2.8-P1, BIND 9.3.4-P1, BIND 9.4.1-P1 or BIND 9.5.0a6
For details, see www.trusteer.com/docs/bind9dns.html
. As soon as there's an update, it gets installed 
.