Testing for Credentials Transported over an Encrypted Channel (OTG-AUTHN-001)
This article is part of the OWASP Testing Guide v3. The entire OWASP Testing Guide v3 can be downloaded here.
OWASP at the moment is working at the OWASP Testing Guide v4: you can browse the Guide here
This is a draft of a section of the new Testing Guide v3
The problem we are going to discuss is to verify that authentication data that the users are sending, are actually transferred via en encrypted channel to avoid being intercepted by some malicious users. The analysis focus simply on trying to understand if our data travel unencrypted from our web browser to the server, or if the web application takes the appropriate security measures using a protocol like HTTPS. This protocol, like others that use encryption, is built on TLS/SSL to encrypt the data that we want to transmit and to ensure that we are sending them towards the desired site. Clearly, the fact that our traffic is encrypted does not necessarily means that it's completely safe. The security also depends from the encryption algorithm used and from the robustness of the keys that we are using. But this particular topic will not be addressed in this section, for a more detailed discussion on testing the safety of our TLS/SSL channel you can refer to chapter Testing for SSL-TLS. We will just try to understand if the data that we put into the web form, in order to log into a web site, are transmitted using secure protocols that protect them from an attacker or not. To do this we will consider various examples.
Description of the Issue
Nowadays, the most common example of this issue is the login page of a web application. This page may use several methods to authenticate the user, but, apart which method it uses, we want to be sure that in the transaction that is established between clients (the web browsers) and server (the application itself), sensitive data (in this case username and password) are transmitted via en encrypted channel. In order to log into a web site, usually we have to fill a simple form that passes the inserted data with the POST method. What is less obvious is that these data can be passed using the HTTP protocol, and then in a non-secure way, or using HTTPS which encrypts the data. To further complicate things, there is the possibility that the site has the login page accessible via HTTP (making us believe to an insecure transmission), but then it actually sends data via HTTPS.We do this because we want to be sure that an attacker can not be able to retrieve sensitive information simply sniffing the net with a sniffer tool. Using HTTPS prevents packet sniffing and Man In The Middle attacks.
Black Box testing and example
In the following examples we will use WebScarab in order to capture packet headers and to inspect them. Anyway you can use any web proxy that you prefer.
Case study: Sending data with POST method through HTTP
We suppose that our login page presents a form with fields User, Pass and the Submit button to authenticate and have access to the application. If we look at the header of our request with WebScarab, we get something like this:
POST http://www.example.com/AuthenticationServlet HTTP/1.1 Host: www.example.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:22.214.171.124) Gecko/20080404 Accept: text/xml,application/xml,application/xhtml+xml Accept-Language: it-it,it;q=0.8,en-us;q=0.5,en;q=0.3 Accept-Encoding: gzip,deflate Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 Keep-Alive: 300 Connection: keep-alive Referer: http://www.example.com/index.jsp Cookie: JSESSIONID=LVrRRQQXgwyWpW7QMnS49vtW1yBdqn98CGlkP4jTvVCGdyPkmn3S! Content-Type: application/x-www-form-urlencoded Content-length: 64 delegated_service=218&User=test&Pass=test&Submit=SUBMIT
From this example we can understand that the POST sends the data to the page www.example.com/AuthenticationServlet simply using HTTP. So, in this case, data are transmitted without encryption and a malicious user could read our username and password simply sniffing the net with a tool like Wireshark.
Case study: Sending data with POST method through HTTPS
Suppose that our web application uses the HTTPS protocol to encrypt data we are sending (or at least for those relating to the authentication). In this case, trying to access the login page and to authenticate, the header of our POST request would be similar to the following:
POST https://www.exmple.com:443/cgi-bin/login.cgi HTTP/1.1 Host: www.exmple.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:126.96.36.199) Gecko/20080404 Accept: text/xml,application/xml,application/xhtml+xml,text/html Accept-Language: it-it,it;q=0.8,en-us;q=0.5,en;q=0.3 Accept-Encoding: gzip,deflate Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 Keep-Alive: 300 Connection: keep-alive Referer: https://www.exmple.com/cgi-bin/login.cgi Cookie: language=English; Content-Type: application/x-www-form-urlencoded Content-length: 50 Command=Login&User=test&Pass=test
We can see that the request is addressed to www.exmple.com:443/cgi-bin/login.cgi using the HTTPS protocol. This ensures that our data are sent through en encrypted channel and that they are not readable by other people. We could also understand this, from the fact that in the window of our web browser is visible the lock that indicates the use of a secure protocol to access the page and that the url of that page starts with HTTPS.
Case study: sending data with POST method via HTTPS on a page reachable via HTTP
Now we suppose to have a web page reacheable via HTTP and that then only data sent from the authentication form are shipped via HTTPS. This means that our data are transmitted in a secure way through encryption. This situation occurs for example when we are on a portal of a big company that offers various information and services publicly available, without identification, but which has also a private section acessible from the home page through a login. So when we try to log, the header of our request will look like the following example:
POST https://www.example.com:443/login.do HTTP/1.1 Host: www.example.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:188.8.131.52) Gecko/20080404 Accept: text/xml,application/xml,application/xhtml+xml,text/html Accept-Language: it-it,it;q=0.8,en-us;q=0.5,en;q=0.3 Accept-Encoding: gzip,deflate Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 Keep-Alive: 300 Connection: keep-alive Referer: http://www.example.com/homepage.do Cookie: SERVTIMSESSIONID=s2JyLkvDJ9ZhX3yr5BJ3DFLkdphH0QNSJ3VQB6pLhjkW6F Content-Type: application/x-www-form-urlencoded Content-length: 45 User=test&Pass=test&portal=ExamplePortal
We can see that our request is addressed to www.example.com:443/login.do using HTTPS. But if we have a look at the referer field in the header (the page from which we came) it is www.example.com/homepage.do and is accessible via simple HTTP. So in this case we have no lock inside our browser window that tells us that we are using a secure connection, but in reality we are sending data via HTTPS. This ensures us that no other people can read the data that we are sending.
Case study: Sending data with GET method through HTTPS
In this last example, suppose that the application transfers data using the GET method. This method should never be used in a form that transmits sensitive data such as username and password, because they are displayed in clear in the URL and this entails a whole set of security issues. So this example is purely demonstrative, but in reality it is strongly suggested to use the POST method instead.
GET https://www.example.com/success.html?user=test&pass=test HTTP/1.1 Host: www.example.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:184.108.40.206) Gecko/20080404 Accept: text/xml,application/xml,application/xhtml+xml,text/html Accept-Language: it-it,it;q=0.8,en-us;q=0.5,en;q=0.3 Accept-Encoding: gzip,deflate Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7 Keep-Alive: 300 Connection: keep-alive Referer: https://www.example.com/form.html If-Modified-Since: Mon, 30 Jun 2008 07:55:11 GMT If-None-Match: "43a01-5b-4868915f"
You can see that the data is transferred in clear text in the URL and not in the body of the message as before. But we must consider that TLS/SSL is a level 5 protocol, a lower level than the HTTP one, then the whole HTTP package is still encrypted and the URL is unreadable to an attacker. But is not a good practice to use the GET method in these cases, because it is more easily exploitable.
Gray Box testing and example
Talk with the developers of the application and try to understand if they are aware of differences between HTTP and HTTPS protocols and why they should use the last one for sensitive information transmissions.
Then check with them if HTTPS is used in every sensitive transmission, like those in login pages, to avoid to an unauthorized user to read the data.
- HTTP/1.1: Security Considerations - http://www.w3.org/Protocols/rfc2616/rfc2616-sec15.html