Eliminating Bad Characters
Some characters can interefere with the working of the protocol or technology and render the exploit useless. In order to be able to carry out a successful execution of code, these bad characters have to be identified and not used in any form during the exploit. Some example of bad characters that should be avoided,
0x00 - Null character signifies the end of a string copy. So this is a common bad character to be avoided.
0x0D - This is a return character that signifies the end of a http field. So if the attack is over http, then it should be avoided.
Now in order to identify the set of bad characters for an exploit, the entire range of characters have to be submitted to the application and the eliminated one by one.
This set can be included in the fuzzer script as the payload and then the stack can be observed through the immunity debugger to check where the strings terminate. Then that character can be removed and the exploit run again to check which next character causes the same problem. The process can be repeated until the end of the set.
The set of bad characters can then be used as input at the time of payload creation to avoid including them into the payload.
Last updated
Was this helpful?