The birthday attack actually dates back to at least 1997, it was at least discussed when we originally published this advisory in the days when IDs were still sequential: http://www.openbsd.org/advisories/res_random.txt (Based on prior work by Gene Spafford and Christopher Schuba) Oliver -----Original Message----- From: Roger A. Grimes [mailto:roger@xxxxxxxxxxxxxx] Sent: Tuesday, April 17, 2007 3:47 PM To: 3APA3A Cc: Makoto Shiotsuki; bugtraq@xxxxxxxxxxxxxxxxx Subject: RE: Re[2]: Windows DNS Cache Poisoning by Forwarder DNS Spoofing Thanks for responding. If this is the case, why is this report a report of a Windows DNS vulnerability, since it appears to be a DNS (or at least BIND and Windows) vulnerability? My guess is the original poster didn't include BIND in his test scope or something like that. Roger ***************************************************************** *Roger A. Grimes, InfoWorld, Security Columnist *CPA, CISSP, MCSE: Security (2000/2003/MVP), CEH, yada...yada... *email: roger_grimes@xxxxxxxxxxxxx or roger@xxxxxxxxxxxxxx *Author of Professional Windows Desktop and Server Hardening (Wrox) *http://www.amazon.com/gp/product/0764599909 ***************************************************************** -----Original Message----- From: 3APA3A [mailto:3APA3A@xxxxxxxxxxxxxxxx] Sent: Tuesday, April 17, 2007 4:50 PM To: Roger A. Grimes Cc: Makoto Shiotsuki; bugtraq@xxxxxxxxxxxxxxxxx Subject: Re[2]: Windows DNS Cache Poisoning by Forwarder DNS Spoofing Dear Roger A. Grimes, DNS spoofing attack in general can not be 'patched', because this is a weakness of DNS protocol itself. As for birthday attack applicability, this problem was discussed in 2002. In 2003 problem still exist in both bind 8 and 9. According to CERT (US-CERT) as on 10/18/2004 bind was still vulnerable. As far as I remember, there never was a patch for bind to prevent this specific attack, yet it can be a part of some later bind release. A possible mitigation against birthday attacks (not against spoofing in general) on the server software level are any of: 1. Do no reuse source port for DNS requests. Have every request to be issued from different source ports (resource consumption attack is possible). 2. Keep a table of issued requests and do not issue request for the same name before response for previous one is received (can not be implemented in scalable 'multiple processes' DNS server architecture) 3. Monitor if multiple replies are received for a single request. I don't know if bind actually use any. Hope, this helps. --Tuesday, April 17, 2007, 8:48:04 PM, you wrote to shio@xxxxxxxxxxxx: RAG> How does BIND stop this sort of attack? RAG> Can a BIND expert respond? RAG> Roger RAG> ***************************************************************** RAG> *Roger A. Grimes, InfoWorld, Security Columnist *CPA, CISSP, MCSE: RAG> Security (2000/2003/MVP), CEH, yada...yada... RAG> *email: roger_grimes@xxxxxxxxxxxxx or roger@xxxxxxxxxxxxxx *Author RAG> of Professional Windows Desktop and Server Hardening (Wrox) RAG> *http://www.amazon.com/gp/product/0764599909 RAG> ***************************************************************** RAG> -----Original Message----- RAG> From: Makoto Shiotsuki [mailto:shio@xxxxxxxxxxxx] RAG> Sent: Tuesday, April 17, 2007 12:31 PM RAG> To: Roger A. Grimes RAG> Cc: bugtraq@xxxxxxxxxxxxxxxxx RAG> Subject: Re: Windows DNS Cache Poisoning by Forwarder DNS Spoofing >>One question. Is BIND any better at preventing this type of attack? RAG> As far as I know, this vulnerability is specific to the Windows DNS. RAG> Makoto Shiotsuki -- ~/ZARAZA http://securityvulns.com/
