# CVE-2023-4911 - Looney Tunables
This is a (atm very rough) proof of concept for CVE-2023-4911. So far I've only verified it works on
Ubuntu 22.10 kinetic. Current version of the exploit contains a fair amount of "magic" offsets. If you have suggestions
on how to improve the heap shaping, feel free to send a PR my way :).
This exploit is basically an implementation of [Qualys' excellent write up][qualys]. It deviates in
some places. That's not necessarily because I think I know better, this is just what worked on
my VM. There is probably room for improvement.
## Usage
1. Compile the exploit and suid library using `make`.
2. Run `./exploit` from a writable directory, containing both `suid_lib.so` and `exploit`. It'll create a folder called `U` in the current
directory and populate it with the necessary libraries.
In my limited testing, the exploit needs somewhere between 4000-8000 attempts. The stack spray is not quite as good as Qualys' implementation, so it takes a bit longer.
Here's an example of the exploit in action (sped up): https://youtu.be/uw0EJ5zGEKE
## Improvement ideas
If you feel like hacking on this, here are some ideas to get you started:
- Make it self-contained. See https://github.com/ly4k/PwnKit for inspiration.
- Make the heap shaping more robust. Right now it's very fragile and depends on a lot of magic
offsets.
- Make it work on more distros.
[qualys]: https://www.qualys.com/2023/10/03/cve-2023-4911/looney-tunables-local-privilege-escalation-glibc-ld-so.txt
[4.0K] /data/pocs/1ee3c2d66beac40f15a2b00300b787bfdc73501c
├── [4.0K] main.c
├── [ 282] Makefile
├── [1.5K] README.md
└── [1.1K] suid_lib.c
0 directories, 4 files