Creating Authentication Systems

[MyndFyre]

I usually keep 1 algorithm for each series of programs, and I only change it if I know it's been cracked.  I think I should start changing my aglorithims each release now -.-

It wouldn't make a difference if the hacker figured out how to bypass the algo altogether.

[UserLoser.]

I usually keep 1 algorithm for each series of programs, and I only change it if I know it's been cracked.  I think I should start changing my aglorithims each release now -.-

Should look into your own encryption/scrambling system if you want it to be more secure.  By secure, I mean take more effort to crack; rather than modifying about 3 bytes of your code in the executable

[iago]

Is it possible to do encryptions like that for VB programs?  I'm assuming that RaBBiT is using VB, and I know VB is really limited (much like Java) for low level stuff like that.

[MyndFyre]

Is it possible to do encryptions like that for VB programs?  I'm assuming that RaBBiT is using VB, and I know VB is really limited (much like Java) for low level stuff like that.

You could do a simple encryption with a stub program, I imagine: encrypt the entire program by doing something simple like XORing each byte with an arbitrary byte such as 0x6f or something.  Then prepend a stub that loads the data, decrypts it, and jumps to where the actual entry point *should* be.

[Eibro]

Is it possible to do encryptions like that for VB programs?  I'm assuming that RaBBiT is using VB, and I know VB is really limited (much like Java) for low level stuff like that.

You could do a simple encryption with a stub program, I imagine: encrypt the entire program by doing something simple like XORing each byte with an arbitrary byte such as 0x6f or something.  Then prepend a stub that loads the data, decrypts it, and jumps to where the actual entry point *should* be.

A system like that is equally as weak. I could just convert the descrambling code into an IDC script and run it on scrambled sections.

[MyndFyre]

A system like that is equally as weak. I could just convert the descrambling code into an IDC script and run it on scrambled sections.

Yes I know.  That's why I said "You could do a simple encryption."  The point was that it would be possible, not the strength of the scheme of protection.

[Banana fanna fo fanna]

Should encrypt the exe on disk, and use code modification in-memory to randomly scramble around program flow.

[R.a.B.B.i.T]

I usually keep 1 algorithm for each series of programs, and I only change it if I know it's been cracked.  I think I should start changing my aglorithims each release now -.-

It wouldn't make a difference if the hacker figured out how to bypass the algo altogether.

It would change the place where the jump occured, however!

[UserLoser.]

I usually keep 1 algorithm for each series of programs, and I only change it if I know it's been cracked.  I think I should start changing my aglorithims each release now -.-

It wouldn't make a difference if the hacker figured out how to bypass the algo altogether.

It would change the place where the jump occured, however!

So?  Then all someone has to do is disassemble it or do what they did the first time to find it

[R.a.B.B.i.T]

True, but it does get annoying to decompile and change the program every time.  Cracking the serial would work until the algorithm would change.

[Banana fanna fo fanna]

it would take prolly 5 min to crack that protection.

[R.a.B.B.i.T]

I never said it was the best, or even GOOD.  It was just an example.

[TheMinistered]

Yes, it is possible-- although quite ugly!  The key to doing it is a v-table hack & signatures.  We use the v-table hack so that we can write our scramble/descramble encrypt/decrypt in assembly.  Then the assembly needs signatures, since we can't do inline assembly we give the scramble/descramble encrypt/decrypt a dummy parameter and we pass something obvious like 0x12345678 or 0xBADBEEF, something our routines can scan for to know where to stop at.  It also requires a bit more for signatures using this method because the signatures will be pushed onto the stack before the call, etc.

Code Select Expand


Dim m_objProtection as new clsProtection

Private Sub Form_Load()
    Set m_objProtection = new clsProtection
End Sub

Private Sub DoMath()
  Dim lngX as long, lngY as long   

    Call m_objProtection.Decramble(&H12345678)
        lngX = 1
        lngY = 1
        lngX = lngX + lngY
    Call m_objProtection.Scramble(&H87654321)
End Sub

Private Sub Form_Unload(ByRef Cancel as Integer)
    Set m_objProtection = Nothing
End Sub


[Arta]

Nice to see you got that working in the end

[Newby]

Nice to see I didn't have to write the scrambling part after all.