Obfuscation is not a panaceaPublished 16 April 2012 10:52 am
So, you want to obfuscate your .NET application. My question to you is:
What are your aims when your obfuscate your application? To protect your IP & algorithms? Prevent crackers from breaking your licensing? Your boss says you need to? To give you a warm fuzzy feeling inside? Obfuscating code correctly can be tricky, it can break your app if applied incorrectly, it can cause problems down the line. Let me be clear – there are some very good reasons why you would want to obfuscate your .NET application. However, you shouldn’t be obfuscating for the sake of obfuscating.
Security through Obfuscation?
Once your application has been installed on a user’s computer, you no longer control it. If they do not want to pay for your application, then nothing can stop them from cracking it, even if the time cost to them is much greater than the cost of actually paying for it. Some people will not pay for software, even if it takes them a month to crack a $30 app. And once it is cracked, there is nothing stopping them from putting the result up on the internet.
There should be nothing suprising about this; there is no software protection available for general-purpose computers that cannot be cracked by a sufficiently determined attacker. Only by completely controlling the entire stack – software, hardware, and the internet connection, can you have even a chance to be uncrackable. And even then, someone somewhere will still have a go, and probably succeed. Even high-end cryptoprocessors have known vulnerabilities that can be exploited by someone with a scanning electron microscope and lots of free time.
So, then, why use obfuscation? Well, the primary reason is to protect your IP. What obfuscation is very good at is hiding the overall structure of your program, so that it’s very hard to figure out what exactly the code is doing at any one time, what context it is running in, and how it fits in with the rest of the application; all of which you need to do to understand how the application operates. This is completely different to cracking an application, where you simply have to find a single toggle that determines whether the application is licensed or not, and flip it without the rest of the application noticing.
However, again, there are limitations. An obfuscated application still has to run in the same way, and do the same thing, as the original unobfuscated application. This means that some of the protections applied to the obfuscated assembly have to be undone at runtime, else it would not run on the CLR and do the same thing. And, again, since we don’t control the environment the application is run on, there is nothing stopping a user from undoing those protections manually, and reversing some of the obfuscation.
It’s a perpetual arms race, and it always will be. We have plenty of ideas lined about new protections, and the new protections added in SA 6.6 (method parent obfuscation and a new control flow obfuscation level) are specifically designed to be harder to reverse and reconstruct the original structure.
So then, by all means, obfuscate your application if you want to protect the algorithms and what the application does. That’s what SmartAssembly is designed to do. But make sure you are clear what a .NET obfuscator can and cannot protect you against, and don’t expect your obfuscated application to be uncrackable. Someone, somewhere, will crack your application if they want to and they don’t have anything better to do with their time. The best we can do is dissuade the casual crackers and make it much more difficult for the serious ones.