Crawling code is the practice of scanning a code base of the review target in question. It is, in effect, looking for key pointers wherein a possible security vulnerability might reside. Certain APIs are related to interfacing to the external world or file IO or user management, which are key areas for an attacker to focus on. In crawling code we look for APIs relating to these areas. We also need to look for business logic areas which may cause security issues, but generally these are bespoke methods which have bespoke names and can not be detected directly, even though we may touch on certain methods due to their relationship with a certain key API.
We also need to look for common issues relating to a specific language; issues that may not be *security* related but which may affect the stability/availability of the application in the case of extraordinary circumstances. Other issues when performing a code review are areas such a simple copyright notice in order to protect one’s intellectual property. Generally these issues should be part of your companies Coding Guidelines, and should be enforceable during a code review (i.e. a reviewer can fail code review because the code violates something in the Coding Guidelines, regardless of whether or not the code would work in its current state, and regardless on whether the original developer agrees or not).
Crawling code can be done manually or in an automated fashion using automated tools. Crawling code manually is probably not effective, as (as can be seen below) there are plenty of indicators which can apply to a language. Tools as simple as grep or wingrep can be used. Other tools are available which would search for key words relating to a specific programming language. If you are using a particular review tool which allows you to specify strings to be highlighted in a review (e.g. Python based review tools using pygments syntax highlighter, or an in-house tool for which you can change the source code) then you could add the relevant string indicators from the lists below and have them highlighted to reviewers automatically.
The following sections shall cover the function of crawing code for Java/J2EE, .NET and Classic ASP. This section is best used in conjunction with the transactional analysis section also detailed in this guide.
Searching for Key Indicators
The basis of the code review is to locate and analyse areas of code which may have application security implications. Assuming the code reviewer has a thorough understanding of the code, what it is intended to do, and the context in which it is to be used, firstly one needs to sweep the code base for areas of interest.
This can be done by performing a text search on the code base looking for keywords relating to APIs and functions. Below is a guide for .NET framework 1.1 & 2.0
Searching for Code in .NET
Firstly one needs to be familiar with the tools one can use in order to perform text searching, following this one needs to know what to look for.
In this section we will assume you have a copy of Visual Studio (VS) .NET at hand. VS has two types of search "Find in Files" and a cmd line tool called Findstr.
To start off, one could scan thorough the code looking for common patterns or keywords such as "User", "Password", "Pswd", "Key", "Http", etc... This can be done using the "Find in Files" tool in VS or using findstring as follows:
findstr /s /m /i /d:c:\projects\codebase\sec "http" *.*
HTTP Request Strings
Requests from external sources are obviously a key area of a security code review. We need to ensure that all HTTP requests received are data validated for composition, max and min length, and if the data falls with the realms of the parameter white-list. Bottom-line is this is a key area to look at and ensure security is enabled.
Here we are looking for responses to the client. Responses which go unvalidated or which echo external input without data validation are key areas to examine. Many client side attacks result from poor response validation. XSS relies on this somewhat.
SQL & Database
Locating where a database may be involved in the code is an important aspect of the code review. Looking at the database code will help determine if the application is vulnerable to SQL injection. One aspect of this is to verify that the code uses either SqlParameter, OleDbParameter, or OdbcParameter(System.Data.SqlClient). These are typed and treat parameters as the literal value and not executable code in the database.
<link to SQL injection section>
|exec sp_||select from||insert||update||delete from where||delete|
|execute sp_||exec xp_||exec @||execute @||executestatement||executeSQL|
Cookie manipulation can be key to various application security exploits, such as session hijacking/fixation and parameter manipulation. One should examine any code relating to cookie functionality, as this would have a bearing on session security.
Many of the HTML tags below can be used for client side attacks such as cross site scripting. It is important to examine the context in which these tags are used and to examine any relevant data validation associated with the display and use of such tags within a web application.
|<frame security||<iframe security||<body>|
The input controls below are server classes used to produce and display web application form fields. Looking for such references helps locate entry points into the application.
The .NET Framework relies on .config files to define configuration settings. The .config files are text-based XML files. Many .config files can, and typically do, exist on a single system. Web applications refer to a web.config file located in the application’s root directory. For ASP.NET applications, web.config contains information about most aspects of the application’s operation.
|forms protection||appSettings||ConfigurationSettings||appSettings||connectionStrings||authentication mode|
Each application has its own Global.asax if one is required. Global.asax sets the event code and values for an application using scripts. One must ensure that application variables do not contain sensitive information, as they are accessible to the whole application and to all users within it.
Logging can be a source of information leakage. It is important to examine all calls to the logging subsystem and to determine if any sensitive information is being logged. Common mistakes are logging userID in conjunction with passwords within the authentication functionality or logging database requests which may contains sensitive data.
Its important that many variables in machine.config can be overridden in the web.config file for a particular application.
Threads and Concurrency
Locating code that contains multithreaded functions. Concurrency issues can result in race conditions which may result in security vulnerabilities. The Thread keyword is where new threads objects are created. Code that uses static global variables which hold sensitive security information may cause session issues. Code that uses static constructors may also cause issues between threads. Not synchronizing the Dispose method may cause issues if a number of threads call Dispose at the same time, this may cause resource release issues.
Public and Sealed relate to the design at class level. Classes which are not intended to be derived from should be sealed. Make sure all class fields are Public for a reason. Don't expose anything you don't need to.
Code may be generated dynamically at runtime. Code that is generated dynamically as a function of external input may give rise to issues. If your code contains sensitive data, does it need to be serialized?
Exceptions & Errors
Ensure that the catch blocks do not leak information to the user in the case of an exception. Ensure when dealing with resources that the finally block is used. Having trace enabled is not great from an information leakage perspective. Ensure customised errors are properly implemented.
|catch||finally||trace enabled||customErrors mode|
If cryptography is used then is a strong enough cipher used, i.e. AES or 3DES? What size key is used? The larger the better. Where is hashing performed? Are passwords that are being persisted hashed? They should be. How are random numbers generated? Is the PRNG "random enough"?
If storing sensitive data in memory, I recommend one uses the following.
Authorization, Assert & Revert
Bypassing the code access security permission? Not a good idea. Also below is a list of potentially dangerous permissions such as calling unmanaged code, outside the CLR.
Some standard functions that should be checked in any context include the following.
Searching for Code in Java
Input and Output Streams
These are used to read data into one’s application. They may be potential entry points into an application. The entry points may be from an external source and must be investigated. These may also be used in path traversal attacks or DoS attacks.
<are some of these a bit wide ranging? java.io?>
These API calls may be avenues for parameter, header, URL, and cookie tampering, HTTP Response Splitting and information leakage. They should be examined closely as many of such APIs obtain the parameters directly from HTTP requests.
Cross Site Scripting
These API calls should be checked in code review as they could be a source of Cross Site Scripting vulnerabilities.
Response splitting allows an attacker to take control of the response body by adding extra CRLFs into headers. In HTTP the headers and bodies are separated by 2 CRLF characters, and thus if an attackers input is used in a response header, and that input contained 2 CRLFs, then anything after the CRLFs would be interpreted as the response body. In code review ensure you are sanitizing any information being put into headers.
Any time your application is sending a redirect response, ensure that the logic involved cannot be manipulated by an attackers input. Especially when input is used to determine where the redirect goes to.
SQL & Database
Searching for Java Database related code this list should help you pinpoint classes/methods which are involved in the persistence layer of the application being reviewed.
Looking for code which utilises SSL as a medium for point to point encryption. The following fragments should indicate where SSL functionality has been developed.
The following APIs should be checked in code review when they control session management.
Here we may be vulnerable to command injection attacks or OS injection attacks. Java linking to the native OS can cause serious issues and potentially give rise to total server compromise.
We may come across some information leakage by examining code below contained in one’s application.
If we can identify major architectural components within that application (right away) it can help narrow our search, and we can then look for known vulnerabilities in those components and frameworks:
### Java Server Faces (JSF)
Searching for Code in Classic ASP
These API in ASP are commonly used to retrieve the input from the request. Therefore code review should ensure these requests (and dependent logic) cannot be manipulated by an attacker.
These APIs are used by ASP to write the response body, that will be sent to the end user. Code review should check these requests are used in a proper manner and no sensitive information can be returned.
Cookies can be a source of information leakage.
<link to error handling section> Ensure errors in your application are handled properly, otherwise an attacker could use error conditions to manipulate you application.
|err.||Server.GetLastError||On Error Resume Next|
|On Error GoTo 0|
Information in URL
These APIs are used to extract information from the URL object in the request. Code review should check that the information extracted from the URL is sanitized.
These APIs can be used to interact with a database, which can lead to SQL attacks. Code review can check these API calls use sanitized input.
|insert into||delete from where||exec|
These API calls can control session within ASP applications.
The following ASP APIs can help prevent DOS attacks against your application.
Leaking information to a log can be of use to an attacker, hence the following API call can be checked in code review to ensure no sensitive information is being written to logs.
Do not allow attacker input to control when and where rejection occurs.
Searching for Code in C++ and Apache
Commonly when a C++ developer is building a web service they will build a CGI program to be invoked by a web server (though this is not efficient) or they will use the Apache httpd framework and write a handler or filter to process HTTP requests/responses. To aid these developers, this section deals with generic C/C++ functions used when processing HTTP input and output, along with some of the common Apache APIs that are used in handlers.
Legacy C/C++ Methods
For any C/C++ code interacting with web requests, code that handles strings and outputs should be checked to ensure the logic does not have any flaws.
When coding within Apache, the following APIs can be used to obtain data from the HTTP request object.
Depending on the type of response you wish to send to the client, the following Apache APIs can be used.
Cookie can be obtained from the list of request headers, or from specialized Apache functions.
Log messages can be implemented using custom loggers included in your module (e.g. log4cxx, boost::log, etc), by using the Apache provided logging API, or by simply writing to standard out or standard error.
When you have got a handle for the HTML input or output in the C/C++ handler, the following methods can be used to ensure/check HTML encoding.