h2. Ruby On Rails Security Guide
This manual describes common security problems in web applications and how to avoid them with Rails. If you have any questions or suggestions, please
mail me, Heiko Webers, at 42 {_et_} rorsecurity.info. After reading it, you should be familiar with:
* All countermeasures _(highlight)that are highlighted_
* The concept of sessions in Rails, what to put in there and popular attack methods
* How just visiting a site can be a security problem (with CSRF)
* What you have to pay attention to when working with files or providing an administration interface
* The Rails-specific mass assignment problem
* How to manage users: Logging in and out and attack methods on all layers
* And the most popular injection attack methods
endprologue.
h3. Introduction
Web application frameworks are made to help developers building web applications. Some of them also help you with securing the web application. In fact one framework is not more secure than another: If you use it correctly, you will be able to build secure apps with many frameworks. Ruby on Rails has some clever helper methods, for example against SQL injection, so that this is hardly a problem. It‘s nice to see that all of the Rails applications I audited had a good level of security.
In general there is no such thing as plug-n-play security. Security depends on the people using the framework, and sometimes on the development method. And it depends on all layers of a web application environment: The back-end storage, the web server and the web application itself (and possibly other layers or applications).
The Gartner Group however estimates that 75% of attacks are at the web application layer, and found out "that out of 300 audited sites, 97% are vulnerable to attack". This is because web applications are relatively easy to attack, as they are simple to understand and manipulate, even by the lay person.
The threats against web applications include user account hijacking, bypass of access control, reading or modifying sensitive data, or presenting fraudulent content. Or an attacker might be able to install a Trojan horse program or unsolicited e-mail sending software, aim at financial enrichment or cause brand name damage by modifying company resources. In order to prevent attacks, minimize their impact and remove points of attack, first of all, you have to fully understand the attack methods in order to find the correct countermeasures. That is what this guide aims at.
In order to develop secure web applications you have to keep up to date on all layers and know your enemies. To keep up to date subscribe to security mailing lists, read security blogs and make updating and security checks a habit (check the Additional Resources chapter). I do it manually because that‘s how you find the nasty logical security problems.
h3. Sessions
A good place to start looking at security is with sessions, which can be vulnerable to particular attacks.
h4. What are sessions?
-- _HTTP is a stateless protocol. Sessions make it stateful._
Most applications need to keep track of certain state of a particular user. This could be the contents of a shopping basket or the user id of the currently logged in user. Without the idea of sessions, the user would have to identify, and probably authenticate, on every request.
Rails will create a new session automatically if a new user accesses the application. It will load an existing session if the user has already used the application.
A session usually consists of a hash of values and a session id, usually a 32-character string, to identify the hash. Every cookie sent to the client's browser includes the session id. And the other way round: the browser will send it to the server on every request from the client. In Rails you can save and retrieve values using the session method:
session[:user_id] = @current_user.id
User.find(session[:user_id])
h4. Session id
-- _The session id is a 32 byte long MD5 hash value._
A session id consists of the hash value of a random string. The random string is the current time, a random number between 0 and 1, the process id number of the Ruby interpreter (also basically a random number) and a constant string. Currently it is not feasible to brute-force Rails' session ids. To date MD5 is uncompromised, but there have been collisions, so it is theoretically possible to create another input text with the same hash value. But this has had no security impact to date.
h4. Session hijacking
-- _Stealing a user's session id lets an attacker use the web application in the victim's name._
Many web applications have an authentication system: a user provides a user name and password, the web application checks them and stores the corresponding user id in the session hash. From now on, the session is valid. On every request the application will load the user, identified by the user id in the session, without the need for new authentication. The session id in the cookie identifies the session.
Hence, the cookie serves as temporary authentication for the web application. Everyone who seizes a cookie from someone else, may use the web application as this user – with possibly severe consequences. Here are some ways to hijack a session, and their countermeasures:
* Sniff the cookie in an insecure network. A wireless LAN can be an example of such a network. In an unencrypted wireless LAN it is especially easy to listen to the traffic of all connected clients. This is one more reason not to work from a coffee shop. For the web application builder this means to _(highlight)provide a secure connection over SSL_.
* Most people don't clear out the cookies after working at a public terminal. So if the last user didn't log out of a web application, you would be able to use it as this user. Provide the user with a _(highlight)log-out button_ in the web application, and _(highlight)make it prominent_.
* Many cross-site scripting (XSS) exploits aim at obtaining the user's cookie. You'll read more about XSS later.
* Instead of stealing a cookie unknown to the attacker, he fixes a user's session identifier (in the cookie) known to him. Read more about this so-called session fixation later.
The main objective of most attackers is to make money. The underground prices for stolen bank login accounts range from $10–$1000 (depending on the available amount of funds), $0.40–$20 for credit card numbers, $1–$8 for online auction site accounts and $4–$30 for email passwords, according to the "Symantec Global Internet Security Threat Report":http://eval.symantec.com/mktginfo/enterprise/white_papers/b-whitepaper_internet_security_threat_report_xiii_04-2008.en-us.pdf.
h4. Session guidelines
-- _Here are some general guidelines on sessions._
* _(highlight)Do not store large objects in a session_. Instead you should store them in the database and save their id in the session. This will eliminate synchronization headaches and it won't fill up your session storage space (depending on what session storage you chose, see below).
This will also be a good idea, if you modify the structure of an object and old versions of it are still in some user's cookies. With server-side session storages you can clear out the sessions, but with client-side storages, this is hard to mitigate.
* _(highlight)Critical data should not be stored in session_. If the user clears his cookies or closes the browser, they will be lost. And with a client-side session storage, the user can read the data.
h4. Session storage
-- _Rails provides several storage mechanisms for the session hashes. The most important are ActiveRecordStore and CookieStore._
There are a number of session storages, i.e. where Rails saves the session hash and session id. Most real-live applications choose ActiveRecordStore (or one of its derivatives) over file storage due to performance and maintenance reasons. ActiveRecordStore keeps the session id and hash in a database table and saves and retrieves the hash on every request.
Rails 2 introduced a new default session storage, CookieStore. CookieStore saves the session hash directly in a cookie on the client-side. The server retrieves the session hash from the cookie and eliminates the need for a session id. That will greatly increase the speed of the application, but it is a controversial storage option and you have to think about the security implications of it:
* Cookies imply a strict size limit of 4kB. This is fine as you should not store large amounts of data in a session anyway, as described before. _(highlight)Storing the current user's database id in a session is usually ok_.
* The client can see everything you store in a session, because it is stored in clear-text (actually Base64-encoded, so not encrypted). So, of course, _(highlight)you don't want to store any secrets here_. To prevent session hash tampering, a digest is calculated from the session with a server-side secret and inserted into the end of the cookie.
That means the security of this storage depends on this secret (and on the digest algorithm, which defaults to SHA512, which has not been compromised, yet). So _(highlight)don't use a trivial secret, i.e. a word from a dictionary, or one which is shorter than 30 characters_. Put the secret in your environment.rb:
config.action_controller.session = {
:key => '_app_session',
:secret => '0x0dkfj3927dkc7djdh36rkckdfzsg...'
}
There are, however, derivatives of CookieStore which encrypt the session hash, so the client cannot see it.
h4. Replay attacks for CookieStore sessions
-- _Another sort of attack you have to be aware of when using CookieStore is the replay attack._
It works like this:
* A user receives credits, the amount is stored in a session (which is a bad idea anyway, but we'll do this for demonstration purposes).
* The user buys something.
* His new, lower credit will be stored in the session.
* The dark side of the user forces him to take the cookie from the first step (which he copied) and replace the current cookie in the browser.
* The user has his credit back.
Including a nonce (a random value) in the session solves replay attacks. A nonce is valid only once, and the server has to keep track of all the valid nonces. It gets even more complicated if you have several application servers (mongrels). Storing nonces in a database table would defeat the entire purpose of CookieStore (avoiding accessing the database).
The best _(highlight)solution against it is not to store this kind of data in a session, but in the database_. In this case store the credit in the database and the logged_in_user_id in the session.
h4. Session fixation
-- _Apart from stealing a user's session id, the attacker may fix a session id known to him. This is called session fixation._
!images/session_fixation.png(Session fixation)!
This attack focuses on fixing a user's session id known to the attacker, and forcing the user's browser into using this id. It is therefore not necessary for the attacker to steal the session id afterwards. Here is how this attack works:
# The attacker creates a valid session id: He loads the login page of the web application where he wants to fix the session, and takes the session id in the cookie from the response (see number 1 and 2 in the image).
# He possibly maintains the session. Expiring sessions, for example every 20 minutes, greatly reduces the time-frame for attack. Therefore he accesses the web application from time to time in order to keep the session alive.
# Now the attacker will force the user's browser into using this session id (see number 3 in the image). As you may not change a cookie of another domain (because of the same origin policy), the attacker has to run a JavaScript from the domain of the target web application. Injecting the JavaScript code into the application by XSS accomplishes this attack. Here is an example: +<script>
document.cookie="_session_id=16d5b78abb28e3d6206b60f22a03c8d9";
</script>+. Read more about XSS and injection later on.
# The attacker lures the victim to the infected page with the JavaScript code. By viewing the page, the victim's browser will change the session id to the trap session id.
# As the new trap session is unused, the web application will require the user to authenticate.
# From now on, the victim and the attacker will co-use the web application with the same session: The session became valid and the victim didn't notice the attack.
h4. Session fixation – Countermeasures
-- _One line of code will protect you from session fixation._
The most effective countermeasure is to _(highlight)issue a new session identifier_ and declare the old one invalid after a successful login. That way, an attacker cannot use the fixed session identifier. This is a good countermeasure against session hijacking, as well. Here is how to create a new session in Rails:
reset_session
If you use the popular RestfulAuthentication plugin for user management, add reset_session to the SessionsController#create action. Note that this removes any value from the session, _(highlight)you have to transfer them to the new session_.
Another countermeasure is to _(highlight)save user-specific properties in the session_, verify them every time a request comes in, and deny access, if the information does not match. Such properties could be the remote IP address or the user agent (the web browser name), though the latter is less user-specific. When saving the IP address, you have to bear in mind that there are Internet service providers or large organizations that put their users behind proxies. _(highlight)These might change over the course of a session_, so these users will not be able to use your application, or only in a limited way.
h4. Session expiry
-- _Sessions that never expire extend the time-frame for attacks such as cross-site reference forgery (CSRF), session hijacking and session fixation._
One possibility is to set the expiry time-stamp of the cookie with the session id. However the client can edit cookies that are stored in the web browser so expiring sessions on the server is safer. Here is an example of how to _(highlight)expire sessions in a database table_. Call +Session.sweep("20m")+ to expire sessions that were used longer than 20 minutes ago.
class Session < ActiveRecord::Base
def self.sweep(time_ago = nil)
time = case time_ago
when /^(\d+)m$/ then Time.now - $1.to_i.minute
when /^(\d+)h$/ then Time.now - $1.to_i.hour
when /^(\d+)d$/ then Time.now - $1.to_i.day
else Time.now - 1.hour
end
self.delete_all "updated_at < '#{time.to_s(:db)}'"
end
end
The section about session fixation introduced the problem of maintained sessions. An attacker maintaining a session every five minutes can keep the session alive forever, although you are expiring sessions. A simple solution for this would be to add a created_at column to the sessions table. Now you can delete sessions that were created a long time ago. Use this line in the sweep method above:
self.delete_all "updated_at < '#{time.to_s(:db)}' OR
created_at < '#{2.days.ago.to_s(:db)}'"
h3. Cross-Site Reference Forgery (CSRF)
-- _This attack method works by including malicious code or a link in a page that accesses a web application that the user is believed to have authenticated. If the session for that web application has not timed out, an attacker may execute unauthorized commands._
!images/csrf.png!
In the session chapter you have learned that most Rails applications use cookie-based sessions. Either they store the session id in the cookie and have a server-side session hash, or the entire session hash is on the client-side. In either case the browser will automatically send along the cookie on every request to a domain, if it can find a cookie for that domain. The controversial point is, that it will also send the cookie, if the request comes from a site of a different domain. Let's start with an example:
* Bob browses a message board and views a post from a hacker where there is a crafted HTML image element. The element references a command in Bob's project management application, rather than an image file.
* +<img src="http://www.webapp.com/project/1/destroy">+
* Bob's session at www.webapp.com is still alive, because he didn't log out a few minutes ago.
* By viewing the post, the browser finds an image tag. It tries to load the suspected image from www.webapp.com. As explained before, it will also send along the cookie with the valid session id.
* The web application at www.webapp.com verifies the user information in the corresponding session hash and destroys the project with the ID 1. It then returns a result page which is an unexpected result for the browser, so it will not display the image.
* Bob doesn't notice the attack -- but a few days later he finds out that project number one is gone.
It is important to notice that the actual crafted image or link doesn't necessarily have to be situated in the web application's domain, it can be anywhere – in a forum, blog post or email.
CSRF appears very rarely in CVE (Common Vulnerabilities and Exposures) -- less than 0.1% in 2006 -- but it really is a 'sleeping giant' [Grossman]. This is in stark contrast to the results in my (and others) security contract work – _(highlight)CSRF is an important security issue_.
h4. CSRF Countermeasures
-- _First, as is required by the W3C, use GET and POST appropriately. Secondly, a security token in non-GET requests will protect your application from CSRF._
The HTTP protocol basically provides two main types of requests - GET and POST (and more, but they are not supported by most browsers). The World Wide Web Consortium (W3C) provides a checklist for choosing HTTP GET or POST:
*Use GET if:*
* The interaction is more _(highlight)like a question_ (i.e., it is a safe operation such as a query, read operation, or lookup).
*Use POST if:*
* The interaction is more _(highlight)like an order_, or
* The interaction _(highlight)changes the state_ of the resource in a way that the user would perceive (e.g., a subscription to a service), or
* The user is _(highlight)held accountable for the results_ of the interaction.
If your web application is RESTful, you might be used to additional HTTP verbs, such as PUT or DELETE. Most of today‘s web browsers, however do not support them - only GET and POST. Rails uses a hidden +_method+ field to handle this barrier.
_(highlight)The verify method in a controller can make sure that specific actions may not be used over GET_. Here is an example to verify the use of the transfer action over POST. If the action comes in using any other verb, it redirects to the list action.
verify :method => :post, :only => [:transfer], :redirect_to => {:action => :list}
With this precaution, the attack from above will not work, because the browser sends a GET request for images, which will not be accepted by the web application.
But this was only the first step, because _(highlight)POST requests can be sent automatically, too_. Here is an example for a link which displays www.harmless.com as destination in the browser's status bar. In fact it dynamically creates a new form that sends a POST request.
To the harmless survey
Or the attacker places the code into the onmouseover event handler of an image:
There are many other possibilities, including Ajax to attack the victim in the background.
The _(highlight)solution to this is including a security token in non-GET requests_ which check on the server-side. In Rails 2 or higher, this is a one-liner in the application controller:
protect_from_forgery :secret => "123456789012345678901234567890..."
This will automatically include a security token, calculated from the current session and the server-side secret, in all forms and Ajax requests generated by Rails. You won't need the secret, if you use CookieStorage as session storage. It will raise an ActionController::InvalidAuthenticityToken error, if the security token doesn't match what was expected.
Note that _(highlight)cross-site scripting (XSS) vulnerabilities bypass all CSRF protections_. XSS gives the attacker access to all elements on a page, so he can read the CSRF security token from a form or directly submit the form. Read more about XSS later.
h3. Redirection and Files
Another class of security vulnerabilities surrounds the use of redirection and files in web applications.
h4. Redirection
-- _Redirection in a web application is an underestimated cracker tool: Not only can the attacker forward the user to a trap web site, he may also create a self-contained attack._
Whenever the user is allowed to pass (parts of) the URL for redirection, it is possibly vulnerable. The most obvious attack would be to redirect users to a fake web application which looks and feels exactly as the original one. This so-called phishing attack works by sending an unsuspicious link in an email to the users, injecting the link by XSS in the web application or putting the link into an external site. It is unsuspicious, because the link starts with the URL to the web application and the URL to the malicious site is hidden in the redirection parameter: http://www.example.com/site/redirect?to= www.attacker.com. Here is an example of a legacy action:
def legacy
redirect_to(params.update(:action=>'main'))
end
This will redirect the user to the main action if he tried to access a legacy action. The intention was to preserve the URL parameters to the legacy action and pass them to the main action. However, it can exploited by an attacker if he includes a host key in the URL:
http://www.example.com/site/legacy?param1=xy¶m2=23&host=www.attacker.com
If it is at the end of the URL it will hardly be noticed and redirects the user to the attacker.com host. A simple countermeasure would be to _(highlight)include only the expected parameters in a legacy action_ (again a whitelist approach, as opposed to removing unexpected parameters). _(highlight)And if you redirect to an URL, check it with a whitelist or a regular expression_.
h5. Self-contained XSS
Another redirection and self-contained XSS attack works in Firefox and Opera by the use of the data protocol. This protocol displays its contents directly in the browser and can be anything from HTML or JavaScript to entire images:
+data:text/html;base64,PHNjcmlwdD5hbGVydCgnWFNTJyk8L3NjcmlwdD4K+
This example is a Base64 encoded JavaScript which displays a simple message box. In a redirection URL, an attacker could redirect to this URL with the malicious code in it. As a countermeasure, _(highlight)do not allow the user to supply (parts of) the URL to be redirected to_.
h4. File uploads
-- _Make sure file uploads don't overwrite important files, and process media files asynchronously._
Many web applications allow users to upload files. _(highlight)File names, which the user may choose (partly), should always be filtered_ as an attacker could use a malicious file name to overwrite any file on the server. If you store file uploads at /var/www/uploads, and the user enters a file name like “../../../etc/passwd”, it may overwrite an important file. Of course, the Ruby interpreter would need the appropriate permissions to do so – one more reason to run web servers, database servers and other programs as a less privileged Unix user.
When filtering user input file names, _(highlight)don't try to remove malicious parts_. Think of a situation where the web application removes all “../” in a file name and an attacker uses a string such as “....//” - the result will be “../”. It is best to use a whitelist approach, which _(highlight)checks for the validity of a file name with a set of accepted characters_. This is opposed to a blacklist approach which attempts to remove not allowed characters. In case it isn't a valid file name, reject it (or replace not accepted characters), but don't remove them. Here is the file name sanitizer from the "attachment_fu plugin":http://github.com/technoweenie/attachment_fu/tree/master:
def sanitize_filename(filename)
returning filename.strip do |name|
# NOTE: File.basename doesn't work right with Windows paths on Unix
# get only the filename, not the whole path
name.gsub! /^.*(\\|\/)/, ''
# Finally, replace all non alphanumeric, underscore
# or periods with underscore
name.gsub! /[^\w\.\-]/, '_'
end
end
A significant disadvantage of synchronous processing of file uploads (as the attachment_fu plugin may do with images), is its _(highlight)vulnerability to denial-of-service attacks_. An attacker can synchronously start image file uploads from many computers which increases the server load and may eventually crash or stall the server.
The solution to this is best to _(highlight)process media files asynchronously_: Save the media file and schedule a processing request in the database. A second process will handle the processing of the file in the background.
h4. Executable code in file uploads
-- _Source code in uploaded files may be executed when placed in specific directories. Do not place file uploads in Rails' /public directory if it is Apache's home directory._
The popular Apache web server has an option called DocumentRoot. This is the home directory of the web site, everything in this directory tree will be served by the web server. If there are files with a certain file name extension, the code in it will be executed when requested (might require some options to be set). Examples for this are PHP and CGI files. Now think of a situation where an attacker uploads a file “file.cgi” with code in it, which will be executed when someone downloads the file.
_(highlight)If your Apache DocumentRoot points to Rails' /public directory, do not put file uploads in it_, store files at least one level downwards.
h4. File downloads
-- _Make sure users cannot download arbitrary files._
Just as you have to filter file names for uploads, you have to do so for downloads. The send_file() method sends files from the server to the client. If you use a file name, that the user entered, without filtering, any file can be downloaded:
send_file('/var/www/uploads/' + params[:filename])
Simply pass a file name like “../../../etc/passwd” to download the server's login information. A simple solution against this, is to _(highlight)check that the requested file is in the expected directory_:
basename = File.expand_path(File.join(File.dirname(__FILE__), '../../files'))
filename = File.expand_path(File.join(basename, @file.public_filename))
raise if basename =!
File.expand_path(File.join(File.dirname(filename), '../../../'))
send_file filename, :disposition => 'inline'
Another (additional) approach is to store the file names in the database and name the files on the disk after the ids in the database. This is also a good approach to avoid possible code in an uploaded file to be executed. The attachment_fu plugin does this in a similar way.
h3. Intranet and Admin security
-- _Intranet and administration interfaces are popular attack targets, because they allow privileged access. Although this would require several extra-security measures, the opposite is the case in the real world._
In 2007 there was the first tailor-made "Trojan":http://www.symantec.com/enterprise/security_response/weblog/2007/08/a_monster_trojan.html which stole information from an Intranet, namely the "Monster for employers" web site of Monster.com, an online recruitment web application. Tailor-made Trojans are very rare, so far, and the risk is quite low, but it is certainly a possibility and an example of how the security of the client host is important, too. However, the highest threat to Intranet and Admin applications are XSS and CSRF.
*XSS* If your application re-displays malicious user input from the extranet, the application will be vulnerable to XSS. User names, comments, spam reports, order addresses are just a few uncommon examples, where there can be XSS.
Having one single place in the admin interface or Intranet, where the input has not been sanitized, makes the entire application vulnerable. Possible exploits include stealing the privileged administrator's cookie, injecting an iframe to steal the administrator's password or installing malicious software through browser security holes to take over the administrator's computer.
Refer to the Injection section for countermeasures against XSS. It is _(highlight)recommended to use the SafeErb plugin_ also in an Intranet or administration interface.
*CSRF* Cross-Site Reference Forgery (CSRF) is a gigantic attack method, it allows the attacker to do everything the administrator or Intranet user may do. As you have already seen above how CSRF works, here are a few examples of what attackers can do in the Intranet or admin interface.
A real-world example is a "router reconfiguration by CSRF":http://www.symantec.com/enterprise/security_response/weblog/2008/01/driveby_pharming_in_the_
wild.html. The attackers sent a malicious e-mail, with CSRF in it, to Mexican users. The e-mail claimed there was an e-card waiting for them, but it also contained an image tag that resulted in a HTTP-GET request to reconfigure the user's router (which is a popular model in Mexico). The request changed the DNS-settings so that requests to a Mexico-based banking site would be mapped to the attacker's site. Everyone who accessed the banking site through that router saw the attacker's fake web site and had his credentials stolen.
Another example changed Google Adsense's e-mail address and password by "CSRF":http://www.0x000000.com/index.php?i=213&bin=11010101. If the victim was logged into Google Adsense, the administration interface for Google advertisements campaigns, an attacker could change his credentials.
Another popular attack is to spam your web application, your blog or forum to propagate malicious XSS. Of course, the attacker has to know the URL structure, but most Rails URLs are quite straightforward or they will be easy to find out, if it is an open-source application's admin interface. The attacker may even do 1,000 lucky guesses by just including malicious IMG-tags which try every possible combination.
For _(highlight)countermeasures against CSRF in administration interfaces and Intranet applications, refer to the countermeasures in the CSRF section_.
h4. Additional precautions
The common admin interface works like this: it's located at www.example.com/admin, may be accessed only if the admin flag is set in the User model, re-displays user input and allows the admin to delete/add/edit whatever data desired. Here are some thoughts about this:
* It is very important to _(highlight)think about the worst case_: What if someone really got hold of my cookie or user credentials. You could _(highlight)introduce roles_ for the admin interface to limit the possibilities of the attacker. Or how about _(highlight)special login credentials_ for the admin interface, other than the ones used for the public part of the application. Or a _(highlight)special password for very serious actions_?
* Does the admin really have to access the interface from everywhere in the world? Think about _(highlight)limiting the login to a bunch of source IP addresses_. Examine request.remote_ip to find out about the user's IP address. This is not bullet-proof, but a great barrier. Remember that there might be a proxy in use, though.
* _(highlight)Put the admin interface to a special sub-domain_ such as admin.application.com and make it a separate application with its own user management. This makes stealing an admin cookie from the usual domain, www.application.com, impossible. This is because of the same origin policy in your browser: An injected (XSS) script on www.application.com may not read the cookie for admin.application.com and vice-versa.
h3. Mass assignment
-- _Without any precautions Model.new(params[:model]) allows attackers to set any database column's value._
The mass-assignment feature may become a problem, as it allows an attacker to set any model's attributes by manipulating the hash passed to a model's +new()+ method:
def signup
params[:user] #=> {:name => “ow3ned”, :admin => true}
@user = User.new(params[:user])
end
Mass-assignment saves you much work, because you don't have to set each value individually. Simply pass a hash to the new() method, or assign attributes=(attributes) a hash value, to set the model's attributes to the values in the hash. The problem is that it is often used in conjunction with the parameters (params) hash available in the controller, which may be manipulated by an attacker. He may do so by changing the URL like this:
"name":http://www.example.com/user/signup?user=ow3ned&user[admin]=1
This will set the following parameters in the controller:
params[:user] #=> {:name => “ow3ned”, :admin => true}
So if you create a new user using mass-assignment, it may be too easy to become an administrator.
h4. Countermeasures
To avoid this, Rails provides two class methods in your ActiveRecord class to control access to your attributes. The attr_protected method takes a list of attributes that will not be accessible for mass-assignment. For example:
attr_protected :admin
A much better way, because it follows the whitelist-principle, is the +attr_accessible+ method. It is the exact opposite of attr_protected, because _(highlight)it takes a list of attributes that will be accessible_. All other attributes will be protected. This way you won't forget to protect attributes when adding new ones in the course of development. Here is an example:
attr_accessible :name
If you want to set a protected attribute, you will to have to assign it individually:
params[:user] #=> {:name => "ow3ned", :admin => true}
@user = User.new(params[:user])
@user.admin #=> false # not mass-assigned
@user.admin = true
@user.admin #=> true
h3. User management
-- _Almost every web application has to deal with authorization and authentication. Instead of rolling your own, it is advisable to use common plug-ins. But keep them up-to-date, too. A few additional precautions can make your application even more secure._
There are some authorization and authentication plug-ins for Rails available. A good one saves only encrypted passwords, not plain-text passwords. The most popular plug-in is +restful_authentication+ which protects from session fixation, too. However, earlier versions allowed you to login without user name and password in certain circumstances.
Every new user gets an activation code to activate his account when he gets an e-mail with a link in it. After activating the account, the activation_code columns will be set to NULL in the database. If someone requested an URL like these, he would be logged in as the first activated user found in the database (and chances are that this is the administrator):
http://localhost:3006/user/activate
http://localhost:3006/user/activate?id=
This is possible because on some servers, this way the parameter id, as in params[:id], would be nil. However, here is the finder from the activation action:
User.find_by_activation_code(params[:id])
If the parameter was nil, the resulting SQL query will be
SELECT * FROM users WHERE (users.activation_code IS NULL) LIMIT 1
And thus it found the first user in the database, returned it and logged him in. You can find out more about it in "my blog post":http://www.rorsecurity.info/2007/10/28/restful_authentication-login-security/. _(highlight)It is advisable to update your plug-ins from time to time_. Moreover, you can review your application to find more flaws like this.
h4. Brute-forcing accounts
-- _Brute-force attacks on accounts are trial and error attacks on the login credentials. Fend them off with more generic error messages and possibly require to enter a CAPTCHA._
A list of user names for your web application may be misused to brute-force the corresponding passwords, because most people don't use sophisticated passwords. Most passwords are a combination of dictionary words and possibly numbers. So armed with a list of user name's and a dictionary, an automatic program may find the correct password in a matter of minutes.
Because of this, most web applications will display a generic error message “user name or password not correct”, if one of these are not correct. If it said “the user name you entered has not been found”, an attacker could automatically compile a list of user names.
However, what most web application designers neglect, are the forgot-password pages. These pages often admit that the entered user name or e-mail address has (not) been found. This allows an attacker to compile a list of user names and brute-force the accounts.
In order to mitigate such attacks, _(highlight)display a generic error message on forgot-password pages, too_. Moreover, you can _(highlight)require to enter a CAPTCHA after a number of failed logins from a certain IP address_. Note, however, that this is not a bullet-proof solution against automatic programs, because these programs may change their IP address exactly as often. However, it raises the barrier of an attack.
h4. Account hijacking
-- _Many web applications make it easy to hijack user accounts. Why not be different and make it more difficult?_
h5. Passwords
Think of a situation where an attacker has stolen a user's session cookie and thus may co-use the application. If it is easy to change the password, the attacker will hijack the account with a few clicks. Or if the change-password form is vulnerable to CSRF, the attacker will be able to change the victim's password by luring him to a web page where there is a crafted IMG-tag which does the CSRF. As a countermeasure, _(highlight)make change-password forms safe against CSRF_, of course. And _(highlight)require the user to enter the old password when changing it_.
h5. E-Mail
However, the attacker may also take over the account by changing the e-mail address. After he changed it, he will go to the forgotten-password page and the (possibly new) password will be mailed to the attacker's e-mail address. As a countermeasure _(highlight)require the user to enter the password when changing the e-mail address, too_.
h5. Other
Depending on your web application, there may be more ways to hijack the user's account. In many cases CSRF and XSS will help to do so. For example, as in a CSRF vulnerability in "Google Mail":http://www.gnucitizen.org/blog/google-gmail-e-mail-hijack-technique/. In this proof-of-concept attack, the victim would have been lured to a web site controlled by the attacker. On that site is a crafted IMG-tag which results in a HTTP GET request that changes the filter settings of Google Mail. If the victim was logged in to Google Mail, the attacker would change the filters to forward all e-mails to his e-mail address. This is nearly as harmful as hijacking the entire account. As a countermeasure, _(highlight)review your application logic and eliminate all XSS and CSRF vulnerabilities_.
h4. CAPTCHAs
-- _A CAPTCHA is a challenge-response test to determine that the response is not generated by a computer. It is often used to protect comment forms from automatic spam bots by asking the user to type the letters of a distorted image. The idea of a negative CAPTCHA is not to ask a user to proof that he is human, but reveal that a robot is a robot._
But not only spam robots (bots) are a problem, but also automatic login bots. A popular CAPTCHA API is "reCAPTCHA":http://recaptcha.net/ which displays two distorted images of words from old books. It also adds an angled line, rather than a distorted background and high levels of warping on the text as earlier CAPTCHAs did, because the latter were broken. As a bonus, using reCAPTCHA helps to digitize old books. "ReCAPTCHA":http://ambethia.com/recaptcha/ is also a Rails plug-in with the same name as the API.
You will get two keys from the API, a public and a private key, which you have to put into your Rails environment. After that you can use the recaptcha_tags method in the view, and the verify_recaptcha method in the controller. Verify_recaptcha will return false if the validation fails.
The problem with CAPTCHAs is, they are annoying. Additionally, some visually impaired users have found certain kinds of distorted CAPTCHAs difficult to read. The idea of negative CAPTCHAs is not to ask a user to proof that he is human, but reveal that a spam robot is a bot.
Most bots are really dumb, they crawl the web and put their spam into every form's field they can find. Negative CAPTCHAs take advantage of that and include a "honeypot" field in the form which will be hidden from the human user by CSS or JavaScript.
Here are some ideas how to hide honeypot fields by JavaScript and/or CSS:
* position the fields off of the visible area of the page
* make the elements very small or colour them the same as the background of the page
* leave the fields displayed, but tell humans to leave them blank
The most simple negative CAPTCHA is one hidden honeypot field. On the server side, you will check the value of the field: If it contains any text, it must be a bot. Then, you can either ignore the post or return a positive result, but not saving the post to the database. This way the bot will be satisfied and moves on. You can do this with annoying users, too.
You can find more sophisticated negative CAPTCHAs in Ned Batchelder's "blog post":http://nedbatchelder.com/text/stopbots.html:
* Include a field with the current UTC time-stamp in it and check it on the server. If it is too far in the past, or if it is in the future, the form is invalid.
* Randomize the field names
* Include more than one honeypot field of all types, including submission buttons
Note that this protects you only from automatic bots, targeted tailor-made bots cannot be stopped by this. So _(highlight)negative CAPTCHAs might not be good to protect login forms_.
h4. Logging
-- _Tell Rails not to put passwords in the log files._
By default, Rails logs all requests being made to the web application. But log files can be a huge security issue, as they may contain login credentials, credit card numbers et cetera. When designing a web application security concept, you should also think about what will happen if an attacker got (full) access to the web server. Encrypting secrets and passwords in the database will be quite useless, if the log files list them in clear text. You can _(highlight)filter certain request parameters from your log files_ by the filter_parameter_logging method in a controller. These parameters will be marked [FILTERED] in the log.
filter_parameter_logging :password
h4. Good passwords
-- _Do you find it hard to remember all your passwords? Don't write them down, but use the initial letters of each word in an easy to remember sentence._
Bruce Schneier, a security technologist, "has analysed":http://www.schneier.com/blog/archives/2006/12/realworld_passw.html 34,000 real-world user names and passwords from the MySpace phishing attack mentioned earlier. It turns out that most of the passwords are quite easy to crack. The 20 most common passwords are:
password1, abc123, myspace1, password, blink182, qwerty1, ****you, 123abc, baseball1, football1, 123456, soccer, monkey1, liverpool1, princess1, jordan23, slipknot1, superman1, iloveyou1, and monkey.
It is interesting that only 4% of these passwords were dictionary words and the great majority is actually alphanumeric. However, password cracker dictionaries contain a large number of today's passwords, and they try out all kinds of (alphanumerical) combinations. If an attacker knows your user name and you use a weak password, your account will be easily cracked.
A good password is a long alphanumeric combination of mixed cases. As this is quite hard to remember, it is advisable to enter only the _(highlight)first letters of a sentence that you can easily remember_. For example "The quick brown fox jumps over the lazy dog" will be "Tqbfjotld". Note that this is just an example, you should not use well known phrases like these, as they might appear in cracker dictionaries, too.
h4. Regular expressions
-- _A common pitfall in Ruby's regular expressions is to match the string's beginning and end by ^ and $, instead of \A and \z._
Ruby uses a slightly different approach than many other languages to match the end and the beginning of a string. That is why even many Ruby and Rails books make this wrong. So how is this a security threat? Imagine you have a File model and you validate the file name by a regular expression like this:
class File < ActiveRecord::Base
validates_format_of :name, :with => /^[\w\.\-\+]+$/
end
This means, upon saving, the model will validate the file name to consist only of alphanumeric characters, dots, + and -. And the programmer added \^ and $ so that file name will contain these characters from the beginning to the end of the string. However, _(highlight)in Ruby ^ and $ matches the *line* beginning and line end_. And thus a file name like this passes the filter without problems:
file.txt%0A
Whereas %0A is a line feed in URL encoding, so Rails automatically converts it to "file.txt\n<script>alert('hello')</script>". This file name passes the filter because the regular expression matches – up to the line end, the rest does not matter. The correct expression should read:
/\A[\w\.\-\+]+\z/
h4. Privilege escalation
-- _Changing a single parameter may give the user unauthorized access. Remember that every parameter may be changed, no matter how much you hide or obfuscate it._
The most common parameter that a user might tamper with, is the id parameter, as in +":id":http://www.domain.com/project/1+, whereas 1 is the id. It will be available in params in the controller. There, you will most likely do something like this:
@project = Project.find(params[:id])
This is alright for some web applications, but certainly not if the user is not authorized to view all projects. If the user changes the id to 42, and he is not allowed to see that information, he will have access to it anyway. Instead, _(highlight)query the user's access rights, too_:
@project = @current_user.projects.find(params[:id])
Depending on your web application, there will be many more parameters the user can tamper with. As a rule of thumb, _(highlight)no user input data is secure, until proven otherwise, and every parameter from the user is potentially manipulated_.
Don‘t be fooled by security by obfuscation and JavaScript security. The Web Developer Toolbar for Mozilla Firefox lets you review and change every form's hidden fields. _(highlight)JavaScript can be used to validate user input data, but certainly not to prevent attackers from sending malicious requests with unexpected values_. The Live Http Headers plugin for Mozilla Firefox logs every request and may repeat and change them. That is an easy way to bypass any JavaScript validations. And there are even client-side proxies that allow you to intercept any request and response from and to the Internet.
h3. Injection
-- _Injection is a class of attacks that introduce malicious code or parameters into a web application in order to run it within its security context. Prominent examples of injection are cross-site scripting (XSS) and SQL injection._
Injection is very tricky, because the same code or parameter can be malicious in one context, but totally harmless in another. A context can be a scripting, query or programming language, the shell or a Ruby/Rails method. The following sections will cover all important contexts where injection attacks may happen. The first section, however, covers an architectural decision in connection with Injection.
h4. Whitelists versus Blacklists
-- _When sanitizing, protecting or verifying something, whitelists over blacklists._
A blacklist can be a list of bad e-mail addresses, non-public actions or bad HTML tags. This is opposed to a whitelist which lists the good e-mail addresses, public actions, good HTML tags and so on. Although, sometimes it is not possible to create a whitelist (in a SPAM filter, for example), _(highlight)prefer to use whitelist approaches_:
* Use before_filter :only => [...] instead of :except => [...]. This way you don't forget to turn it off for newly added actions.
* Use attr_accessible instead of attr_protected. See the mass-assignment section for details
* Allow <strong> instead of removing <script> against Cross-Site Scripting (XSS). See below for details.
* Don't try to correct user input by blacklists:
** This will make the attack work: "<sc<script>ript>".gsub("<script>", "")
** But reject malformed input
Whitelists are also a good approach against the human factor of forgetting something in the blacklist.
h4. SQL Injection
-- _Thanks to clever methods, this is hardly a problem in most Rails applications. However, this is a very devastating and common attack in web applications, so it is important to understand the problem._
h5. Introduction
SQL injection attacks aim at influencing database queries by manipulating web application parameters. A popular goal of SQL injection attacks is to bypass authorization. Another goal is to carry out data manipulation or reading arbitrary data. Here is an example of how not to use user input data in a query:
Project.find(:all, :conditions => "name = '#{params[:name]}'")
This could be in a search action and the user may enter a project's name that he wants to find. If a malicious user enters ' OR 1=1', the resulting SQL query will be:
SELECT * FROM projects WHERE name = '' OR 1 --'
The two dashes start a comment ignoring everything after it. So the query returns all records from the projects table including those blind to the user. This is because the condition is true for all records.
h5. Bypassing authorization
Usually a web application includes access control. The user enters his login credentials, the web applications tries to find the matching record in the users table. The application grants access when it finds a record. However, an attacker may possibly bypass this check with SQL injection. The following shows a typical database query in Rails to find the first record in the users table which matches the login credentials parameters supplied by the user.
User.find(:first, "login = '#{params[:name]}' AND password = '#{params[:password]}'")
If an attacker enters ' OR '1'='1 as the name, and ' OR '2'>'1 as the password, the resulting SQL query will be:
SELECT * FROM users WHERE login = '' OR '1'='1' AND password = '' OR '2'>'1' LIMIT 1
This will simply find the first record in the database, and grants access to this user.
h5. Unauthorized reading
The UNION statement connects two SQL queries and returns the data in one set. An attacker can use it to read arbitrary data from the database. Let's take the example from above:
Project.find(:all, :conditions => "name = '#{params[:name]}'")
And now let's inject another query using the UNION statement:
') UNION SELECT id,login AS name,password AS description,1,1,1 FROM users --
This will result in the following SQL query:
SELECT * FROM projects WHERE (name = '') UNION
SELECT id,login AS name,password AS description,1,1,1 FROM users --')
The result won't be a list of projects (because there is no project with an empty name), but a list of user names and their password. So hopefully you encrypted the passwords in the database! The only problem for the attacker is, that the number of columns has to be the same in both queries. That's why the second query includes a list of ones (1), which will be always the value 1, in order to match the number of columns in the first query.
Also, the second query renames some columns with the AS statement so that the web application displays the values from the user table. Be sure to update your Rails "to at least 2.1.1":http://www.rorsecurity.info/2008/09/08/sql-injection-issue-in-limit-and-offset-parameter/.
h5. Countermeasures
Ruby on Rails has a built in filter for special SQL characters, which will escape ' , " , NULL character and line breaks. Using +Model.find(id)+ or +Model.find_by_some thing(something)+ automatically applies this countermeasure. But in SQL fragments, especially in conditions fragments (+:conditions => "..."+), the +connection.execute()+ or +Model.find_by_sql()+ methods, it has to be applied manually.
Instead of passing a string to the conditions option, you can pass an array to sanitize tainted strings like this:
Model.find(:first, :conditions => ["login = ? AND password = ?", entered_user_name, entered_password])
As you can see, the first part of the array is an SQL fragment with question marks. The sanitized versions of the variables in the second part of the array replace the question marks. Or you can pass a hash for the same result:
Model.find(:first, :conditions => {:login => entered_user_name, :password => entered_password})
The array or hash form is only available in model instances. You can try +sanitize_sql()+ elsewhere. _(highlight)Make it a habit to think about the security consequences when using an external string in SQL_.
h4. Cross-Site Scripting (XSS)
-- _The most widespread, and one of the most devastating security vulnerabilities in web applications is XSS. This malicious attack injects client-side executable code. Rails provides helper methods to fend these attacks off._
h5. Entry points
An entry point is a vulnerable URL and its parameters where an attacker can start an attack.
The most common entry points are message posts, user comments, and guest books, but project titles, document names and search result pages have also been vulnerable - just about everywhere where the user can input data. But the input does not necessarily have to come from input boxes on web sites, it can be in any URL parameter – obvious, hidden or internal. Remember that the user may intercept any traffic. Applications, such as the "Live HTTP Headers Firefox plugin":http://livehttpheaders.mozdev.org/, or client-site proxies make it easy to change requests.
XSS attacks work like this: An attacker injects some code, the web application saves it and displays it on a page, later presented to a victim. Most XSS examples simply display an alert box, but it is more powerful than that. XSS can steal the cookie, hijack the session, redirect the victim to a fake website, display advertisements for the benefit of the attacker, change elements on the web site to get confidential information or install malicious software through security holes in the web browser.
During the second half of 2007, there were 88 vulnerabilities reported in Mozilla browsers, 22 in Safari, 18 in IE, and 12 in Opera. The "Symantec Global Internet Security threat report":http://eval.symantec.com/mktginfo/enterprise/white_papers/b-whitepaper_internet_security_threat_report_xiii_04-2008.en-us.pdf also documented 239 browser plug-in vulnerabilities in the last six months of 2007. "Mpack":http://pandalabs.pandasecurity.com/archive/MPack-uncovered_2100_.aspx is a very active and up-to-date attack framework which exploits these vulnerabilities. For criminal hackers, it is very attractive to exploit an SQL-Injection vulnerability in a web application framework and insert malicious code in every textual table column. In April 2008 more than 510,000 sites "were hacked":http://www.0x000000.com/?i=556 like this, among them the British government, United Nations, and many more high targets.
A relatively new, and unusual, form of entry points are banner advertisements. In earlier 2008, malicious code appeared in banner ads on popular sites, such as MySpace and Excite, according to "Trend Micro":http://blog.trendmicro.com/myspace-excite-and-blick-serve-up-malicious-banner-ads/.
h5. HTML/JavaScript Injection
The most common XSS language is of course the most popular client-side scripting language JavaScript, often in combination with HTML. _(highlight)Escaping user input is essential_.
Here is the most straightforward test to check for XSS:
This JavaScript code will simply display an alert box. The next examples do exactly the same, only in very uncommon places:
)
h6. Cookie theft
These examples don't do any harm so far, so let's see how an attacker can steal the user's cookie (and thus hijack the user's session). In JavaScript you can use the document.cookie property to read and write the document's cookie. JavaScript enforces the same origin policy, that means a script from one domain cannot access cookies of another domain. The document.cookie property holds the cookie of the originating web server. However, you can read and write this property, if you embed the code directly in the HTML document (as it happens with XSS). Inject this anywhere in your web application to see your own cookie on the result page:
For an attacker, of course, this is not useful, as the victim will see his own cookie. The next example will try to load an image from the URL http://www.attacker.com/ plus the cookie. Of course this URL does not exist, so the browser displays nothing. But the attacker can review his web server's access log files to see the victims cookie.
The log files on www.attacker.com will read like this:
GET http://www.attacker.com/_app_session=836c1c25278e5b321d6bea4f19cb57e2
You can mitigate these attacks (in the obvious way) by adding the "httpOnly":http://dev.rubyonrails.org/ticket/8895 flag to cookies, so that document.cookie may not be read by JavaScript. Http only cookies can be used from IE v6.SP1, Firefox v2.0.0.5 and Opera 9.5. Safari is still considering, it ignores the option. But other, older browsers (such as WebTV and IE 5.5 on Mac) can actually cause the page to fail to load. Be warned that cookies "will still be visible using Ajax":http://ha.ckers.org/blog/20070719/firefox-implements-httponly-and-is-vulnerable-to-xmlhttprequest/, though.
h6. Defacement
With web page defacement an attacker can do a lot of things, for example, present false information or lure the victim on the attackers web site to steal the cookie, login credentials or other sensitive data. The most popular way is to include code from external sources by iframes:
This loads arbitrary HTML and/or JavaScript from an external source and embeds it as part of the site. This iFrame is taken from an "actual attack":http://www.symantec.com/enterprise/security_response/weblog/2007/06/italy_under_attack_mpack_gang.html on legitimate Italian sites using the "Mpack attack framework":http://isc.sans.org/diary.html?storyid=3015. Mpack tries to install malicious software through security holes in the web browser – very successfully, 50% of the attacks succeed.
A more specialized attack could overlap the entire web site or display a login form, which looks the same as the site's original, but transmits the user name and password to the attackers site. Or it could use CSS and/or JavaScript to hide a legitimate link in the web application, and display another one at its place which redirects to a fake web site.
Reflected injection attacks are those where the payload is not stored to present it to the victim later on, but included in the URL. Especially search forms fail to escape the search string. The following link presented a page which stated that "George Bush appointed a 9 year old boy to be the chairperson...":
http://www.cbsnews.com/stories/2002/02/15/weather_local/main501644.shtml?zipcode=1-->