Fuzzing

Hi,

I just stumbled upon those slides, where they test file system code in the linux kernel (ext4, btrfs, xfs, ...) using a fuzzing method (they do find bugs):

http://events.linuxfoundation.org/sites/events/files/slides/AFL%20filesystem%20fuzzing%2C%20Vault%202016.pdf

I have not studied it in detail, but it looks like they used the open source "American Fuzzy Lop" (afl, written for userland code) and got some ideas how to employ it for kernel code.

Could this perhaps be used to test some parts of cryptsetup?

Best wishes,
Lars

I don't really think so. Fuzzing is good to find non-obscure bugs and authorization problems. There is no authorization in cryptsetup because the user executing cryptsetup has full controll over the code and the machine anyways.

And while cryptsetup surely has some errors left, I think most of them are either obscure or end up here anyways and get fixed. If nobody runs into bugs, then they do not matter unless they are in some software that protects some kind of perimeter. Cryptsetup does not protect any perimeter.

Just my take on this, others may disagree.

Note that filesystems and more generally datastructures are very good targets for fuzzing or randomized testing. I have made excellent experiences with that. But they are different. If cryptsetup works wrongly, you do not get an unlocked container. That is blatantly obvious. If a datastructure or filesystem works wrongly, you get subtle errors and corruption that you may not notice for a long time.

Regards,
Arno

Arno Wagner,     Dr. sc. techn., Dipl. Inform.,    Email: [email protected]
GnuPG: ID: CB5D9718  FP: 12D6 C03B 1B30 33BB 13CF  B774 E35C 5FA1 CB5D 9718
----
A good decision is based on knowledge and not on numbers. -- Plato

If it's in the news, don't worry about it.  The very definition of 
"news" is "something that hardly ever happens." -- Bruce Schneier