|
This vulnerability only affects proxy servers configured in transparent mode. This mode is not usually used. Reminder on proxy servers Proxy servers are servers which relay requests between the clients of a local network and servers outside the network. They can relay several protocols but essentially relay the exchanges based on the HTTP/HTTPS protocol (HTTP proxy) to allow users to connect to the Internet with the following features:
A proxy server configured in transparent mode relays connections without user interaction or browsers configuration. A proxy server can also be used to enable Internet users to connect to internal servers. This is referred to as reverse-proxy server. The RFC 2616 defines the "host" field of HTTP headers in such a way to allow the requests to specify the HTTP server and port number of the requested resource. This field is compulsory and its absence causes the rejection of the request with error code 400 (Bad Request). Note: If the port number is not defined then the default number of the requested service is used (e.g. 80 for HTTP). This month, a vulnerability affecting various proxy servers, was reported by US-CERT. It is detailed in the Cert-IST FA-2009.0041 flaw under investigation. It only impacts transparent mode configured proxy servers, which make connection decisions related to the HTTP requests, based on the HTTP header "host" field instead of the source and destination IP addresses of the requests. An attacker, who is able to craft HTTP requests, can indeed connect to all websites to which a vulnerable proxy server has access, by using a HTTP request with a crafted "host" field. Multiple proxy servers, including Squid and BlueCoat ProxySG, which are monitored by the Cert-IST, are vulnerable. Up to now, patch has still not been released. The Cert-IST will release a security advisory when official patches are released for proxy servers it monitors. In the meantime, it is possible for network administrators to reduce the level of risk associated with this vulnerability by:
Users can however limit the use of active web content (JavaScript, Flash ...) to trusted sites only.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||