cd ../../Random Posts

'Injection' Without Injection

2 minutes read

.shared sections are used as a way to share data between multiple instances(e.g. A' and A'') of an application. If you modify the shared data in A' process, you will get the modified data in A'' process, even if A'' is created after the data modification.

How can we use this to ‘inject’ a code?

If we have an application 'X', which contains encrypted/packed code and we want to decrypt the code and execute it from a remote process, after decrypting/unpacking the code, we need to write the code into a remote process (e.g. via WriteProcessMemory, etc) and execute it (e.g. via CreateRemoteThread, QueueUserAPC, etc).

Execution of child process or chain of child processes is a good way to bypass/hinder many security products, but WriteProcessMemory, CreateRemoteThread, etc. are famous functions used for process injections, so they are suspicious.

'inject' code using a .shared section

We can use .shared section to modify data in a remote process by just modifying the data inside the current process.

Create a .shared section

  • Unpack/decrypt code
  • Copy to the shared section
  • Execute the current application (the second instance)

It’s the second instance of the sample application but with modified .shared section (there is decrypted/unpacked code)

Conclusion

The first instance of the application does nothing but decrypting/unpacking a code. The second instance of the application does nothing but executing the code decrypted/unpacked by the first instance. This way of separating unpacking and unpacked code execution into two processes without writing the unpacked code to the second instance (e.g. via WriteProcessMemory, etc.) maybe can be used to bypass generic unpackers and/or security products.

PoC:

Any feedback would be greatly appreciated. @_qaz_qaz