That's correct. Back in 2000-2001 I reported to Microsoft that they were using SERVER_NAME variable in some of their sample application which made some site even more vulnerable. Any server variable should be considered untrusted and validated like any other user input. This is the reason why our SecureObject product as been detecting server variable usage and protecting them automatically. For more information visit http://www.spidynamics.com/products/devinspectso2003/index.html Sacha Faust Manager - SPILabs S.P.I. Dynamics, Inc. sfaust@xxxxxxxxxxxxxxx www.spidynamics.com Secure. Protect. Inspect. -----Original Message----- From: 3APA3A [mailto:3APA3A@xxxxxxxxxxxxxxxx] Sent: August 23, 2005 6:19 AM To: inge_eivind.henriksen@xxxxxxxxx Cc: bugtraq@xxxxxxxxxxxxxxxxx Subject: Re: Remote IIS 5.x and IIS 6.0 Server Name Spoof Dear inge_eivind.henriksen@xxxxxxxxx, The bug here is not in ability to spoof SERVER_NAME, because SERVER_NAME is untrusted data from Host: request header or from proxy-style HTTP request (like in case of your example). SERVER_NAME is ALWAYS untrusted data. The bug here is in the way SERVER_NAME is used in error page genaration. So, you article should be called something like "Microsoft IIS error page access validation weakness". If any script use SERVER_NAME in this way, this is vulnerability of the script itself. --Monday, August 22, 2005, 7:23:08 PM, you wrote to bugtraq@xxxxxxxxxxxxxxxxx: ihcn> 6. Try and access it from a remote server with telnet again. This time use the following HTTP request: ihcn> GET http://localhost/test.asp HTTP/1.0 -- ~/ZARAZA Но Гарри... я безусловно отдаю предпочтение ему, за высокую питательность и какое-то особенно нежное мясо. (Твен)
