This site is the archived OWASP Foundation Wiki and is no longer accepting Account Requests.
To view the new OWASP Foundation website, please visit https://owasp.org

Codereview-Authorization

From OWASP
Revision as of 14:39, 11 October 2008 by Rahimjina (talk | contribs)

Jump to: navigation, search
OWASP Code Review Guide Table of Contents

Introduction

Authorization issues cover a wide array of layers in a web application; from the functional authorization of a user to gain access to a particular function of the application is at the application layer to the Database access authorization and least privilege issues at the persistence layer. So what to look for when performing a code review? From an attack perspective the most common issues are a result of curiosity and also exploitation of vulnerabilities such as SQL injection. Example: A Database account used by an application with system/admin access upon which the application was vulnerable to SQL injection would result in a higher degree of impact rather than the same vulnerable application with a least privilege database account.

Authorisation is key in multiuser environments where user data should be segregated. Different clients/users should not see other clients' data (Horizontal authorisation). Authorisation can also be used to restrict functionality to a subset of users. "Super users" would have extra admin functionality that a "regular user" would not have access to (Vertical authorisation).

Authorisation is a very bespoke area in application development. It can be implemented via a lookup table in a users' session which is loaded upon successful authentication. It could be a real-time interrogation of a backend LDAP or database system upon each request.

How to locate the potentially vulnerable code

Business logic errors are key areas in which to look for authorization errors. Areas wherein authorization-checks are performed are worth looking at. Logical conditional cases are areas for examination, such as malformed logic:

if user.equals("NormalUser"){
   grantUser(Normal_Permissions);
}else{ //user must be admin/super
  grantUser("Super_Persmissions);
}

For classic ASP pages, authorization is usually performed using include files that contain the access control validation and restrictions. So you usually will look for something like

<!--#include file="ValidateUser.inc"-->

We have an additional issue in this sentence a Information disclosure as the inc file might be called directly and disclose application functionality as ASP code will not be executed given that Inc extension is not recognized.

Vulnerable Patterns for Authorization issues

One area of examination is to see if the authorization model simply relies on not displaying certain functions which the user has not authorization to use, security by obscurity in effect. If a crawl can be performed on the application, links may be discovered which are not on the users GUI. Simple HTTP Get requests can uncover "Hidden" links. Obviously a map on the server-side should be used to see if one is authorized to perform a task and we should not rely on the GUI "hiding" buttons and links.

So disabling buttons on the client due to the authorization level of user shall not prevent the user from executing the action relating to the button.

document.form.adminfunction.disabled=true;
<form action="./doAdminFunction.asp">

By simply saving the page locally and editing the disabled=true to disabled=false and adding the absolute form action one can proceed to activate the disabled button.

HotSpots

The Database: The account used by the application to access the database. Ensure least privilege is in effect.

ASP.NET: (web.config)

The <authorization> element controls ASP.NET URL authorization and the accessibility to gain access to specific folders, pages and resources by users/web clients. Make sure that only authenticated users are authorized to see/visit certain pages.

<system.web>
 <authorization>
   <deny users="?"/>   <-- Anonymous users are denied access. Users must be authenticated.
 </authorization>
</system.web>


The roleManager Element in ASP.NET 2.0 is used to assist in managing roles within the framework. It assists the developer as not as much bespoke code needs to be developed. In web.config, to see if it is enabled check:

<system.web>
..........
<roleManager enabled="true|false" <providers>...</providers> </roleManager>
..........
</system.web>

Apache 1.3

In Apache 1.3 there is a file called httpd. Access control can be implemented from here in the form of the Allow and Deny directives. allow from address is the usage where address is the IP address or domain name to apply access to. Note this granularity is host level granularity.

deny from 124.20.0.249 denies access to that IP.

Order ensures that the 'order'of access is observed.

Order Deny,Allow Deny from all Allow from owasp.org

Above, all is denied apart from owasp.org

To move the authorization to the user level in apache we can use the Satisfy directive.

Good Example

Check authorisation upon every user request.


String action = request.getParameter("action")
if (action == "doStuff"){
  boolean permit = session.authTable.isAuthorised(action); // check table if authoirsed to do action
}
if (permit){
 doStuff();
}else{
 throw new (InvalidRequestException("Unauthorised request"); // inform user of no authorisation
 session.invalidate(); // Kill session
}

Authorisation being performed upon all requests from external entities

Authorisation.jpg

Bad Example

Building the GUI based on the users authorisation. "If he cant see the control we wont be able to use it" 

- Common enough error. If a user has the URL the functionality can still be called. This is due to no authorisation check being performed upon every HTTP request.

Related Vulnerabilities

Operating System injection can be used to totally ignore authorisation constraints. Access to the underlying host is a key objective of system breach. The application is simply a conduit for access to data.
SQL injection can be used to circumvent authorisation. Again, systems are breached to obtain underlying data, they are not breached for the applications themselves. SQL injection is in essence accessing the data via an "out of band" channel not intended by the application.
The root of all evil - Need we say more :)

Insecure class files, folders in deployment may be used to attack an application outside the actual application itself.
Impersonation can obviously be used to gain unauthorised privilege.

In a multi user, multi-threaded environment thread safety is important as one may obtain another individuals session in error.