v0.7.0

[0vercl0k]

This PR:

• Makes the MemoryReader_t interface safer. You can't access its raw buffer anymore, you have to go through ReadMemory which will enforce memory safety. On top of that, the MemBlock_t has been updated to not expose a raw Data pointer anymore but instead an offset. This allows us to kinda make sure somebody won't keep a pointer around after the dump instance has been destructed & run into UaFs. As long as the buffer is kept alive by the MemoryReader_t, this interface should be hard to misuse.
• Also the MemoryReader_t interface has been designed such that, if somebody wanted, they could also read a user-mode dump directly from a FILE* and not necessarily use a memory-mapped file. I haven't implemented it, but it should be doable with this new interface.

[0vercl0k]

@mrexodia I think I'm kinda done tweaking things around; I'd love if you could give it a shot & let me know if anything breaks / have any feedback :)

Cheers

[mrexodia]

Also the MemoryReader_t interface has been designed such that, if somebody wanted, they could also read a user-mode dump directly from a FILE* and not necessarily use a memory-mapped file.

I don't see how this would work? The MemoryReader_t still exposes the raw pointers via ViewBase and ViewEnd and Read is not a virtual function, so a derived class wouldn't be able to override it.

Otherwise the new interface works fine for me!

[0vercl0k]

Oh yes you're right, but making Read virtual is one line change; none of the algorithm need changing, which was the annoying part :)

And yes you're also right about ViewBase / ViewEnd but this will become an implementation details of the memory readers, not the future reader interface (whenever somebody asks for the file implementation); but right now again the algorithm don't use the base / end (nor the py bindings) so it'll be an easy fix as well.

Hope it makes sense, and really appreciate you taking a look, thank you!

Will try to merge this tomorrow or something.

Cheers