Server Side Request Forgery (SSRF)
Last updated
Last updated
SSRF is a vulnerability in web applications whereby an attacker can make further HTTP requests through the server. An attacker can make use of this vulnerability to communicate with any internal services on the server's network which are generally protected by firewalls.
The process would usually be something like this: an attacker finds an SSRF vulnerability on a website. The firewall allows all requests to the website. The attacker then exploits the SSRF vulnerability by forcing the webserver to request data from the database, which it then returns to the attacker. Because the request is coming from the webserver, rather than directly from the attacker, the firewall allows this to pass.
The main cause of the vulnerability is (as it often is) blindly trusting input from a user. In the case of an SSRF vulnerability, a user would be asked to input a URL (or maybe an IP address). The web application would use that to make a request. SSRF comes about when the input hasn't been properly checked or filtered.
Assume there is an application that takes the URL for an image, which the web page then displays for you. The vulnerable SSRF code would look like this:
This is simple PHP code which checks if there is information sent in a 'url' parameter then, without performing any kind of check on it, the code simply makes a request to the user-submitted URL. Attackers essentially have full control of the URL and can make arbitrary GET requests to any website on the Internet through the server -- as well as accessing resources on the server itself.
The above example shows a very small flask application which does the same thing:
It takes the value of the "url" parameter.
Then it makes a request to the given URL and shows the content of that URL to the user.
Again we see that there is no sanitisation or any kind of check performed on the user input. This is why you should always try as many different payloads as you can when testing an application.
This payload might give you the hint that there is an SSRF vulnerability, and give you a hint as to which payloads which you should try next.
Initially, start by searching for the localhost IP (127.0.0.1) with any port to see if the port is running a service. Say you wanted to check if the server has a hidden database, you might search for http://127.0.0.1:3306
, 3306 is the port for MySQL DB so if there is a database running, you will likely get a positive response.
In a similar manner, we could also have used "localhost" or "0.0.0.0" in place of 127.0.0.1
Now it's very possible that some sort of sanitization will be being applied to the input, so the system might detect strings like "localhost" or "127.0.0.1" and stop the request. That said, it's possible to try and bypass those kinds of restrictions.
The very first way is to try the IPv6 version of the localhost i.e [::]. So the payload from before would look like this http://[::]:3306
Flask/Django might interpret these payloads differently. If you fail with that payload, try removing the brackets (i.e try http://:::3306
)
It is possible that the IPv6 payload may also be detected. In that case what we usually do is to encode our IP: either into a decimal format or a hexadecimal format.
The IP "127.0.0.1
" can be replaced with its Decimal and Hexadecimal counterparts to bypass the restrictions. The decimal version of the localhost IP would be "2130706433" and the Hexadecimal version would be "0x7f000001".
There is a script to do this conversion process, you can find it here
If we start the URL with file://
it would then try to read the files from the server itself.
For example, a simple SSRF file reading payload would be file:///etc/passwd
, to read the /etc/passwd file on a Linux machine.