HTTP Parameter Pollution Vulnerabilities in Web Applications (Black Hat EU 2011)
3 likes11,650 views
The document discusses HTTP Parameter Pollution (HPP) vulnerabilities in web applications, highlighting its increased significance in web security due to the complexity of modern web applications. It introduces Papas, an automated detection system for HPP flaws, and presents findings from testing over 5,000 websites, revealing that approximately 30% were vulnerable, with about 14% of those being exploitable. The authors emphasize the need for better input validation and developer awareness to mitigate HPP risks.
![V-Scan: Testing for HPP vulnerabilities
• For every page, an innocuous URL-encoded parameter
(nonce) is injected
– E.g., “%26foo%3Dbar”
– Then check if the “&foo=bar” string is included inside the
URLs of links or forms in the answer page
• V-Scan starts by extracting the list PURL=[PU1,PU2,…PUn] of
the parameters that are present in the page URL, and the
list Pbody=[PB1,PB2,…PUm] of the parameters that are
present in links or forms contained in the page body](https://image.slidesharecdn.com/hpp-bheu2011-110318085935-phpapp02/85/HTTP-Parameter-Pollution-Vulnerabilities-in-Web-Applications-Black-Hat-EU-2011-27-320.jpg)
HTTP Parameter Pollution Vulnerabilities in Web Applications (Black Hat EU 2011)
- 1. HTTP Parameter Pollution Vulnerabilities in Web Applications Marco `embyte’ Balduzzi (C. Torrano, D.Balzarotti, E. Kirda) Do you have the last version of this presentation? http://www.iseclab.org/people/embyte/slides/BHEU2011/hpp-bhEU2011.pdf
- 2. Overview • Introduction • HTTP Parameter Pollution • Detection Approach • Tool • Experiments • Results • Demo • Conclusions
- 3. Who am I? • From Bergamo (IT) to the French Riviera • MSc in Computer Engineering • PhD student at EURECOM • 8+ years experience in IT Security • Engineer and consultant for different international firms • Co-founder of BGLug, Applied Uni Lab, (ex) SPINE Group, Nast, etc… • http://www.iseclab.org/people/embyte
- 4. The Web as We Know It • Has evolved from being a collection of simple and static pages to fully dynamic applications • Applications are more complex than they used to be • Multi-tier architecture is the normal • Many complex systems have web interfaces
- 6. Now
- 7. Increased Importance of Web Security • As a consequence: – Web security has increased in importance – OWASP, the Top Ten Project – Attack against web apps constitute 60% of attacks on the Internet (SANS’s The Top Cyber Security Risks) – Application being targeted for hosting drive-by- download content or C&C servers – Malware targeting browsers (e.g. key and network loggers)
- 8. Increased Importance of Web Security • A lot of work done to detect injection type flaws: – SQL Injection – Cross Site Scripting – Command Injection • Injection vulnerabilities have been well-studied, and tools exist – Sanitization routines in languages (e.g., PHP) – Static code analysis (e.g., Pixy, OWASP Orizon) – Dynamic techniques (e.g., Huang et al.) – Web Application Firewalls (WAF)
- 9. HTTP Parameter Pollution • A new class of Injection Vulnerability called HTTP Parameter Pollution (HPP) is less known – Has not received much attention – First presented by S. di Paola and L. Carettoni at OWASP 2009 • Attack consists of injecting encoded query string delimiters into existing HTTP parameters (e.g. GET/POST/Cookie) – If application does not sanitize its inputs, HPP can be used to launch client-side or server-side attacks – Attacker may be able to override existing parameter values, inject a new parameter or exploit variables out of a direct reach
- 10. Research Objectives • To create the first automated approach for detecting HPP flaws – Blackbox approach, consists of a set of tests and heuristics • To find out how prevalent HPP problems were on the web – Is the problem being exaggerated? – Is this problem known by developers? – Does this problem occur more in smaller sites than larger sites? – What is the significance of the problem?
- 11. HTTP Parameter Handling • During interaction with web application, client provides parameters via GET/POST/Cookie – http://www.site.com/login?login=alice • HTTP allows the same parameter to be provided twice – E.g., in a form checkbox http://www.w3schools.com/html/tryit.asp?filename=tryhtml_form_checkbox • What happens when the same parameter is provided twice? – http://www.site.com/login?login=alice&login=bob
- 12. Google example
- 13. Yahoo example
- 14. HTTP Parameter Handling • We manually tested common methods of 5 different languages Technology/Server Tested Method Parameter Precedence ASP/IIS Request.QueryString(“par”) All (comma-delimited string) PHP/Apache $_GET(“par”) Last JSP/Tomcat Request.getParameter(“par”) First Perl(CGI)/Apache Param(“par”) First Python/Apache getvalue(“par”) All (List) • There is nothing bad with it, if the developer is aware of this behavior • Languages provide secure functions (python’s getfirst())
- 15. HTTP Parameter Pollution • An HTTP Parameter Pollution (HPP) attack occurs – When a malicious parameter Pinj, preceded by an encoded query string delimiter (e.g. %26), is injected into an existing parameter Phost • Typical scenario (client-side) – Web application for election for two candidates
- 16. HTTP Parameter Pollution Url : http://host/election.jsp?poll_id=4568 Link1: <a href="vote.jsp?poll_id=4568&candidate=white”> Vote for Mr.White </a> Link2: <a href="vote.jsp?poll_id=4568&candidate=green”> Vote for Mrs.Green </a> • The two links are built from the URL ID = Request.getParameter(“pool_id”) href_link = “vote.jsp?poll_id=” + ID + ”&candidate=xyz” • No sanitization
- 17. HTTP Parameter Pollution • poll_id is vulnerable and Attacker creates URL: http://host/election.jsp?poll_id=4568%26candidate%3Dgreen • The resulting page now contains injected links: <a href=vote.jsp?pool_id=4568&candidate=green&candidate=white> Vote for Mr. White </a> <a href=vote.jsp?pool_id=4568&candidate=green&candidate=green> Vote for Mrs. Green </a> • If the developer expects to receive a single value – Jsp’s Request.getParameter(“candidate”)returns the 1st value – The parameter precedence is consistent… • Candidate Mrs. Green is always voted!
- 18. Consequence • Override existing (hardcoded) values • Inject a new parameter • Exploit a parameter out of a direct reach • Client-side (user) or server-side (web- application) attack
- 19. Parameter Pollution – More uses • Cross-channel pollution – HPP attacks can also be used to override parameters between different input channels (GET/POST/Cookie) – Good security practice: accept parameters only from where they are supposed to be supplied • HPP to bypass CSRF tokens – E.g. Yahoo Mail client-side attack (di Paola & Carrettoni)
- 20. Bonus • By concatenating the same parameter multiple time • Bypass WAFs input validation checks – Exploit ASP concatenation behavior and inline comments – Concatenate the attack payload after the WAF filtering Standard: show_user.aspx?id=5;select+1,2,3+from+users+where+id=1– Over HPP: show_user.aspx?id=5;select+1&id=2&id=3+from+users+where+id=1— Standard: show_user.aspx?id=5+union+select+*+from+users— Over HPP: show_user.aspx?id=5/*&id=*/union/*&id=*/select+*/*&id=*/from+users--
- 21. System for HPP Detection • Four main components: browser, crawler, two scanners
- 22. Main Components Instrumented browser fetches the web pages and renders their content – Full support for client-side scripts (e.g. Javascript) and external resources (e.g. <embed>) – Extracts all links and forms Crawler communicates with browser, determines URLs to visit and forms to submit. Passes the information to two scanners P-Scan: Determines page behavior when two parameters with the same name are injected V-Scan: Tests and attempts to verify that site is vulnerable to HPP
- 23. P-Scan: Analysis of the Parameter Precedence – Analyzes a page to determine the precedence of parameters, when multiple occurrences of the same parameter are submitted – Take parameter par1=val1, generate a similar value par1=new_val • Page0 (original): app.php?par1=val1 • Page1 (test 1) : app.php?par1=new_val • Page2 (test 2) : app.php?par1=val1&par1=new_val – How do we determine precedence? Naïve approach: • Page0==Page2 -> precedence on first parameter • Page1==Page2 -> precedence on second parameter
- 24. P-Scan: Problem with the naïve approach • In practice, naïve technique does not work well – Applications are complex, much dynamic content (publicity banners, RSS feeds, ads, etc.) – Hence, we perform pre-filtering to eliminate dynamic components (embedded content, applets, iframes, stylesheets, etc.) – Remove all self-referencing URLs (as these change when parameters are inserted) – We then perform different tests to determine similarity
- 25. P-Scan: Tests • Error test – The application crashes, or return an “internal” error, when an identical parameter is injected multiple times – Regexps from the sqlmap project • Identity test – Is the tested parameter considered by the application • Page0=Page1=Page2 • Base test – Test assumes that the pre-filtering works perfectly (seldom the case)
- 26. P-Scan: Tests • Join test – Are the two values are somehow combined together (e.g. ASP)? • Fuzzy test – It is designed to cope with pages whose dynamic components have not been perfectly sanitized – Based on the Gestalt Pattern Matching algorithm – Compute the similarity among the pages
- 27. V-Scan: Testing for HPP vulnerabilities • For every page, an innocuous URL-encoded parameter (nonce) is injected – E.g., “%26foo%3Dbar” – Then check if the “&foo=bar” string is included inside the URLs of links or forms in the answer page • V-Scan starts by extracting the list PURL=[PU1,PU2,…PUn] of the parameters that are present in the page URL, and the list Pbody=[PB1,PB2,…PUm] of the parameters that are present in links or forms contained in the page body
- 28. Where to inject the nonce • PA = PURL ∩ PBody : set of parameters that appear unmodified in the URL and in the page content (links, forms) • PB = p | p PURL p / PBody : URL parameters that do not appear in the page. Some of these parameters may appear in the page under a different name • PC = p | p / PURL p PBody : set of parameters that appear somewhere in the page, but that are not present in the URL
- 29. V-Scan: Special Cases • E.g., one of the URL parameters (or part of it) is used as the entire target of a link • Self-referencing links • Similar issues with printing, sharing functionalities • To reduce false positives, we use heuristics – E.g., the injected parameter does not start with http:// – Injection without URL-encoding
- 30. Implementation – The PAPAS tool • PAPAS: Parameter Pollution Analysis System • The components communicate via TCP/IP sockets – Crawler and Scanner are in Python – The browser component has been implemented as a Firefox extension – Advantage: We can see exactly how pages are rendered (cope with client-side scripts) – Support for multiple sessions (parallelization)
- 31. Implementation – The PAPAS tool • PAPAS is fully customizable – E.g., scanning depth, number of performed injections, page loading timeouts, etc. • Three modes are supported – Fast mode, extensive mode, assisted mode – In assisted mode, authenticated areas of a site can be scanned as well
- 32. Possible improvements • PAPAS does not support the crawling of links embedded in active content – E.g., flash • Support additional encoding schemas (UTF-8, Double URL) • PAPAS currently only focuses on client-side exploits where user needs to click on a link – HPP is also possible on the server side – but this is more difficult to detect – Analogous to detecting stored XSS
- 33. Ethical Considerations • Only client-side attacks. The server-side have the potential to cause harm • We provided the applications with innocuous parameters (&foo=bar). No malicious code. • Limited scan time (15min) and activity • We immediately informed, when possible, the security engineers of the affected applications – Thankful feedbacks
- 34. Two set of experiments We used PAPAS to scan a set of popular websites – About 5,000 sites collected by the first 500 of Alexa’s main categories – The aim: To quickly scan as many websites as possible and to see how common HPP flaws are We then analyzed some of the sites we identified to be HPP-vulnerable in more detail
- 35. The 5,016 tested sites Categories # of Tested Categories # of Tested Applications Applications Financial 110 Shopping 460 Games 300 Social Networking 117 Government 132 Sports 256 Health 235 Travel 175 Internet 698 University 91 News 599 Video 114 Organization 106 Others 1,401 Science 222
- 36. Efficient assessment • In 13 days, we tested 5,016 sites and more than 149,000 unique pages • To maximize the speed, the scanner – Crawled pages up to a distance of 3 from the homepage – Considered links with at least one parameter (except for the homepage) – Considered at max 5 instances for page (same page, different query string) – We disabled pop-ups, images, plug-ins for active content technologies
- 37. Evaluation – Parameter Precedence • Database Errors – Web developers does not seem conscious of the possibility to duplicate GET/POST parameters
- 38. Evaluation – Parameter Precedence • Parameter Inconsistency – Sites developed using a combination of heterogeneous technologies (e.g. PHP and Perl) – This is perfectly safe if the developer is aware of the HPP threat… this is not always the case
- 39. Evaluation – HPP Vulnerabilities • PAPAS discovered that about 1,500 (30%) websites contained at least one page vulnerable to HTTP Parameter Injection – The tool was able to inject (and verify) an encoded parameter • Vulnerable != Exploitable – Is the parameter precedence consistent? – Can a possible attacker override existing parameter values?
- 40. Vulnerable or exploitable? • Injection on link: – Parameter in the middle -> always overriding – Parameter at the begin/end -> automated check via P-Scan • Injection on form: – The injected value is automatically encoded by the browser – Still, someone may be able to run a two-step attack (client-side) or a server-side attack
- 41. Vulnerable or exploitable? • 702 applications are exploitable – About 14% – The injected parameter either overrides the value of an existing one or is accepted as “new parameter” • E.g. A new action is injected Url: pool.pl?par1=val1%26action%3Dreset Link: target.pl?x=y&w=z&par1=val1&action=reset
- 42. Evaluation • More sensitive sites are equally (or even more) affected by the problem
- 43. False Positives • 10 applications (1.12%) use the injected parameter as entire target for one link • Variation of the special case we saw in slide 18 (V-Scan: special cases) – The application applied a transformation to the parameter before using it as a link’s URL
- 44. Some Case Studies • We investigated some of the websites in more detail – Among our “victims”: Facebook, Google, Symantec, Microsoft, PayPal, Flickr, FOX Video, VMWare, … – We notified security officers and some of the problems were fixed – Facebook: share component – Several shopping cart applications could be manipulated to change the price of an item – Some banks were vulnerable and we could play around with parameters – Google: search engine results could be manipulated
- 48. Nasa.gov: coldfusion SQL Error
- 50. Your (secured) home banking
- 51. And Google
- 52. PAPAS Online Service • 5K websites tested – 30% sites are vulnerable: injectable parameters – 14% exploitable: possible to override or introduce arbitrary parameters/values • What about mine? • PAPAS @ http://papas.iseclab.org • Free-to-use service • Ownership token verification • Configurable
- 53. PAPAS Online Service DEMO
- 54. HPP Prevention • Input validation – Encoded query string delimiters • Use safe methods – Parameter precedence (ref. slide 14) – Channel (GET/POST/Cookie) validation (ref. slide 19) • Raise awareness – The client can provide the same parameter twice (or more)
- 55. Acknowledgments, References • Co-joint work: – M. Balduzzi, C. Torrano Gimenez, D. Balzarotti, and E. Kirda. Automated discovery of parameter pollution vulnerabilities in web applications. In NDSS’11, San Diego, CA. • http://papas.iseclab.org/cgi-bin/resources.py • Black Hat’s White Paper
- 56. Conclusion Presented the first technique and system to detect HPP vulnerabilities in web applications. • We call it PAPAS, http://papas.iseclab.org Conducted a large-scale study of the Internet • About 5,000 web sites Our results suggest that Parameter Pollution is a largely unknown, and wide-spread problem We hope our work will help raise awareness about HPP!