Note the emphasis on "default" above; you can use the Linux sandboxing features such as seccomp-bpf to build a sandbox which is truly memory-safe, closing this hole. The OS is in charge of the features it exposes, and Rust can't do much about that.
Note also that the existence of totally_safe_transmute doesn't mean that Rust's memory safety features are pointless. Empirically, memory-safe programming languages result in far fewer memory safety vulnerabilities, because they make exploitation way harder.
The LLM hype has really jumped the shark.
And even then, it doesn't help with whole-system security when every program in every other language on your system, including safe ones like Java and Python, have the same capability. (Although if I'm wrong that there's no precedence for languages attempting to block this, I'd love to see the prior art.)
ETXTBSY.
How about ptracing yourself?
How about undervolting your CPU such that it malfunctions?
How about running your program off a FUSE filesystem that changes the contents out from under it?
How about editing the raw block device you’re running from?
How about modifying /dev/mem on older systems that allow that?
This particular rabbit hole is extremely deep, and I don’t think it’s practical to address it for real short of making a real security model for what running code may and may not to do the rest of the world. (I think doing that is an extremely worthwhile endeavor, but disallowing opening specific files won’t be part of it.)
That's the easy part, you'd provide some OS-specific API for getting this file handle, call it `std::os::fs::proc()` or thereabouts, and make it an `unsafe` function. The "hard to do" is the bigger problem, because if you're providing an unsafe alternative then you'd like to plug all the holes in the safe interface, which AFAIK is non-trivial.