|
Blog-N-Play.com
|
Top Three Links You Must Click On
Security Variadic Functions: How They Contribute To Security Vulnerabilities and How To Fix Them
Variadic functions are implemented using either the ANSI C stdarg approach or, historically, the UNIX System V vararg approach
By: Robert Seacord
Dec. 6, 2005 03:15 PM
C/C++ language variadic functions are functions that accept a variable number of arguments. Variadic functions are implemented using either the ANSI C stdarg approach or, historically, the UNIX System V vararg approach. Both approaches require that the contract between the developer and user of the variadic function not be violated by the user.
These functions accept a fixed format string argument that specifies, among other things, the number and type of arguments that are expected. If the contents of the format string are incorrect (by error or by malicious intent), the resulting behavior of the function is undefined. Incautious use of formatted I/O functions have led to numerous, exploitable vulnerabilities. The majority of these vulnerabilities occur when a potentially malicious user is able to control all or some portion of the format specification string as shown in the following program:
1. #include <stdio.h> These vulnerabilities are often referred to as "format string" vulnerabilities. Exploits take a variety of forms, the most dangerous of which involves using the %n conversion specifier to overwrite memory and transfer control to arbitrary code of the attacker's choosing. The easiest way to prevent format string vulnerabilities is to ensure that the format string does not include characters from untrusted sources. Because of internationalization, however, format strings and message text are often moved into external catalogs or files that the program opens at runtime. An attacker can alter the values of the formats and strings in the program by modifying the contents of these files. The entire topic of formatted output is covered in detail in my book on Secure Coding in C/C++. Format string vulnerabilities have been discovered in a variety of deployed C language programs, including:
1. int average(int first, ...) { Variadic functions are declared using a partial parameter list followed by the ellipsis notation. The variadic average() function accepts a single, fixed integer argument followed by a variable argument list. Like other functions, the arguments to the variadic function are pushed on the calling stack. Variadic functions are problematic for a number of reasons. The first and foremost is that the implementation has no real way of knowing how many arguments were passed (even though this information is available at compile time). The termination condition for the argument list is a contract between the programmers who implement the library function and the programmers who use the function in an application. In this implementation of the average() function, termination of the variable argument list is indicated by an argument whose value is -1. This means, for example, that average(5, -1, 2, -1) is 5, not 2, as the programmer might expect. Also, if the programmer calling the function neglects to provide this argument, the average() function will continue to process the next argument indefinitely until a -1 value is encountered or an exception occurs. A second problem with variadic functions is a complete lack of type checking. In the case of formatted output functions, the type of the arguments is determined by the corresponding conversion specifier in the format string. For example, if a %d conversion specifier is encountered, the formatted output function assumes that the corresponding argument is an integer. If a %s is found, the corresponding argument is interpreted as a pointer to a string. This could result in a program fault, for example, if the corresponding argument was actually a small integer value. Every time a variadic function consumes an argument, an internal argument pointer is incremented to reference the next argument on the stack. If there is some type confusion, it is possible that the argument pointer is incorrectly incremented. This happens less than you might imagine on a 32-bit architecture such as the 32-bit Intel Architecture (IA-32) because almost all arguments (including addresses, char, short, int, and long int) use four bytes. However, conversion specifiers such as a, A, e, E, f, F, g, or G are used to output a 64-bit floating-point number, thereby incrementing the argument pointer by 8. The standard C formatted output functions need modifications to print 64-bit integer and pointer values in hexadecimal. The %x modifier will only print out the first 32 bits of the value that is passed to it and increment the internal argument pointer by 4 bytes. To print out a 64-bit pointer, the ANSI C %p directive needs to be used rather than %x or %u. To print 64-bit integers, you need to use the one size specifier.
Solutions Reader Feedback: Page 1 of 1
Subscribe to our RSS feeds now and receive the next article instantly!
Subscribe to the World's Most Powerful Newsletters
Linux Links You Must Click On !
|
Lo Ultimo
|
||||||||||||||||||||||||||||||||||||||||||