Re: Simple .NET licensing mechanism - how to crack this (assuming exe is obfuscated?) - C# Discussion Boards

There's a serious side to my answer. Suppose I tell him how to crack the application - then I have posted an answer in a forum that is crawled by various web crawlers; at this stage I have told just about any script kiddy out there how to do it as well. Now, this is a simplistic application, but the technique that applies will be similar.

I'm not a stalker, I just know things. Oh by the way, you're out of milk.

Forgive your enemies - it messes with their heads

My blog | My articles | MoXAML PowerToys | Onyx

Pete O'Hanlon wrote:
<layer>Suppose I tell him how to crack the application - then I have posted an answer in a forum that is crawled by various web crawlers;

..right

A script kid would merely download a crack, not create one - by your reasoning, we'd be helping hackers simply by reporting leaks in Windows. After all, they are holes in existing systems, and Google would index them Suspicious | :suss:

I'm a firm believer that any hole should be pointed out, so that it can be closed.

I are Troll Suspicious | :suss:

That's right, so the only way to be secure and safe is never ever discuss security implications.

I'm considering buying third party + on top build one myself so two layers.

dev

Yes but how's that going to solve your problem? If their implementation is on .NET layer... then we might as well code it yourself right?

If on the other hand they patch the binary, I've reviewed one vendor's solution, you'd still end up need to write .NET level permissioning statements such as:

<br />
bool IsAuthorized = ThirdPartyLicensControl("ModuleName");<br />
if(IsAuthorized)<br />
{<br />
    Module.IsEnabled = true;<br />
}<br />

dev

For native executables, placing NOP's in the appropriate locations does the trick. I did this to mine when investigating how hard/easy it would be for them to bypass my licensing mechanism. Now granted, I knew where to look in the executable but it ended up working so I'd say, your wasting your time. I gave up with licensing schemes but still choose to obfuscate as much as is practical.

[modified]
Geez, whoever one voted my comment has obviously never applied the machine instruction 90h (as many times as needed) to an native binary exectuable to manipulate branch statement execution. You might want to look it up in the intel handbooks under "instruction sets" and maybe try manipulating the execution paths in the executable. I think you'll learn a thing or two. Also you might want to read a security book or two about why you shouldn't put any faith in hiding secrets in your code.

Granted the OP was talking about managed code but the concepts are basically the same.

modified on Friday, March 11, 2011 12:20 PM

yes I hear you (though I dont know assembly programming)

Binary patch is the bottom line - if hacker knows where the code is which makes the comparison

<br />
 IsAuthorized = LicenseController.CheckPermission("ModuleName");<br />
if(IsAuthorized)<br />
{<br />
    ... load the module ..<br />
}<br />

The question is, how hard/easy to spot the line? If code is obfuscated (In machine code it'd be same cmp command but there're still thousands of them hacker would need to try patching them one by one, right? This leads me to thinking dynamically generate millions of comparison operators [which is never on actual runtime execution path of course] in .NET code to further throw the sucker off the trail)

Another question is where to store decryption key?

dev

modified on Saturday, March 12, 2011 7:58 PM

For native executables, you can attach to the process with a debugger and if the application throws up a message box saying "you don't have a license", you've basically brought them very close to the code path they are looking for. Also, there are various techniques to creating breakpoints that will make the porcess of tracing through much easier since they will be in the vicinity of the code, even if you don't give many hints.

I had a licensing scheme that obfuscated all sensitive string literals, comparison keys, etc... using a blowfish encryption scheme and I dynamically generated the encryption key using a function. All it took to bypass my code was to open the executable in binary mode (using Visual Studio) going to the byte sequence I had narrowed it down to using a simple debug trace of the process, and I replaced the appropriate bytes in the executable with the opcode for "NOP" (0x90) to eliminate the branch statement path that would have exited the application if the license was not correct and saved my settings. I was able to run the executable with it bypassing all my licensing code. All my hard work trying to protect my app was basically useless.

This was with a native app. Someone trying to manipulate your managed code will probably have an easier time backward engineering your efforts (obfuscated or not) and will easily bypass any protective licensing mechanism. Every book I have on security basically states, "don't rely on hiding secrets in your code". I would have to agree with that statement but I would still say it's not a complete waste of time trying to discourage reverse enginerring to some degree. Just don't get too convinced that you'll be able to completely prevent it as it is an exercise in futility, unfortunately.

I'd just hate to see someone else spend too much time on this endeavor since no matter how complicated you make it, it will probably boil down to a simple if/else statement in the end and those are easily manipulated.

Some may scold me for talking about "how to" here as they will say it is giving ammunition to those who want to bypass our code protections but I would argue that every security book considers this "false sense of security 101" and usually talks about this within the first 100 pages anyway. I think we are stuck with this reality, as unfortunate as it is.

Anyway, I would still encourage you to obfuscate to some degree to help protect your intellectual property.
Good luck with your project.

Many thanks.

I think I'm included to implement *many* licensing checks at different places and dummy if-else (generate a dummy cs file/class with a simple console program) to throw them off the track (security by obscurity yes, but I'm not worried about political correctness)

Also, that "You do not have license" message box should comes from a thread, with variable delay.

dev

Hello...

i have found some issue during, design my project.

my client request is they want improve system performance to almost max limit.

so i suggest Parallel programming solution to my client.

but how can i run my each multi thread in each core ?

i need each thread to run on each core. ( that base on c# language. )

current system model is running multi thread on multi core, but they can't control one to one structure.

For example, on quad core system model that have three business logic thread and have one monitor thread And they need running one to one.

do you have any solution in this case?

thank for your read.

SungBae.Han

This[^] Stack Overflow article may help.

------------------------------------

I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave

CCC Link[^]
Trolls[^]

oops. i understand.

thanks for your reply.

Thread pool is more nice solution about multi thread.

but current status is max in cpu.

even always each cpu is full.

client will be add more server. but, current status will be continues.

because they are running every day big size request process.

so i want more detail control cpu core.

can i do it?

You could have a look at the Thread class, paying attention to "processor affinity".

I have strong doubts it makes much sense on any Windows system though, in my experience it isn't worth anything, i.e. the Windows scheduler does a good job distributing the load over the cores, and not switching threads around unduly. The one situation I would use it is when I had N groups of threads where each group needs to share a limited amount of cached data, so having all its threads on the same core really is relevant.

When I have a number of similar/identical jobs, I tend to create up to 2*N threads where N equals Environment.ProcessorCount; then I let Windows make the best of it, which it does. And I don't use the ThreadPool in such situations.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.

My database has a few tables with 250,000+ rows. We need to grab a few of these tables at app startup, so it takes a minute or so. Is there any "little known flag" to do that more effeciently?

I guess I could cache a local copy and do a timestamp type thing... I did try using Microsoft Sync Services before, but it was much slower then just hitting the database.

BUT... that was when our DB was a lot smaller. Now it would probably be a lot faster past the initial startup time.

That number of rows (not forgetting the number of fields per row) is going to take time to load regardless.

Can you not thread it and do the load whilst the user is doing his log on?

------------------------------------

I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave

CCC Link[^]
Trolls[^]

The user enters his username and password and then it starts grabbing data (in a background thread), but the user can't really use the app til its loaded and that loading takes a while.

Rethink.

And why would your app need to load 250K rows, is the user going to read all that instantaneously?
Use pagination, and load small chunks.

BTW: that is also why this forum shows a limited number of messages at any time.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.

Damn good answer.

I took him at his word that he needed to load them all at the front end, didn't occur to ask him if it was necessary.

------------------------------------

I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave

CCC Link[^]
Trolls[^]

Dalek Dave wrote:
he needed to load them all at the front end

I don't know, I'm just shaking the tree here.

I sure wouldn't read such a lot it the user's expense, even if that means a complete redesign. Besides, data loaded in memory could become stale pretty soon in a multi-user system.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.

Luc Pattyn wrote:
data loaded in memory could become stale pretty soon in a multi-user
system.

This is true.

Unless there is some exceptional reason, I would think it better to poll the db on request, rather than suck it all in.

There is also the memory resources problem.

------------------------------------

I will never again mention that I was the poster of the One Millionth Lounge Post, nor that it was complete drivel. Dalek Dave

CCC Link[^]
Trolls[^]

Yeah, the original concern was preventing UI lag which is why we suck down tables at once. But yeah, this is a multi-user system, so if someone leaves the UI open for a long period of time because they don't like the startup cost, the data could be stale.

Some of the other large tables, I load "on demand" because I don't really need that data most of the time. I definitely need Table1, but that only has like 5 rows, so no biggie. Table2 has the Table1 -> Table3 (1:M) relationships.

If I say open Table1.Row1Entity then I only really need a subset of Table2 and a subset of Table3, but we are still talking ~350 rows at that level.

Now there is a further level of objects in the tree (this is the 250,000 rows)... I guess I don't need all 250,000 at once, but its going to be around 350 x 108, so 37000 rows when I open Table1.Row1Entity. The 37000 rows are obviously not all displayed at once, so I guess technically I could only grab 108 at a time.

Hmm... I guess there are some optimizations I can make in regards to loading only subsets of data, but I certainly don't want to introduce "pauses" in the UI. I'd rather have 1 pause at startup and have it fast from then on...

Not sure if loading ~300 rows at a time would introduce UI lag?

loading 1000 rows should be fine, unless your data structure really isn't appropriate and requires a very complex and expensive query returning few results. Such was not suggested by your message.

I suggest you just give it a try with a mock query.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, improve readability, and make me actually look at the code.