this post was submitted on 29 Aug 2024
518 points (98.7% liked)

Linux

48684 readers
339 users here now

From Wikipedia, the free encyclopedia

Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).

Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.

Rules

Related Communities

Community icon by Alpár-Etele Méder, licensed under CC BY 3.0

founded 5 years ago
MODERATORS
 

Wedson Almeida Filho is a Microsoft engineer who has been prolific in his contributions to the Rust for the Linux kernel code over the past several years. Wedson has worked on many Rust Linux kernel features and even did a experimental EXT2 file-system driver port to Rust. But he's had enough and is now stepping away from the Rust for Linux efforts.

From Wedon's post on the kernel mailing list:

I am retiring from the project. After almost 4 years, I find myself lacking the energy and enthusiasm I once had to respond to some of the nontechnical nonsense, so it's best to leave it up to those who still have it in them.

...

I truly believe the future of kernels is with memory-safe languages. I am no visionary but if Linux doesn't internalize this, I'm afraid some other kernel will do to it what it did to Unix.

Lastly, I'll leave a small, 3min 30s, sample for context here: https://youtu.be/WiPp9YEBV0Q?t=1529 -- and to reiterate, no one is trying force anyone else to learn Rust nor prevent refactorings of C code."

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 13 points 3 months ago (1 children)

The first directory block is a hole. But type == DIRENT, so no error is reported. After that, we get a directory block without '.' and '..' but with a valid dentry. This may cause some code that relies on dot or dotdot (such as make_indexed_dir()) to crash

The problem isn't that the block is a hole. It's that the downstream function expects the directory block to contain . and .., and it gets given one without because of incorrect error handling.

You can encode the invariant of "has dot and dot dot" using a refinement type and smart constructor. The refined type would be a directory block with a guarantee it meets that invariant, and an instance of it could only be created through a function that validates the invariant. If the invariant is met, you get the refined type. If it isn't, you only get an error.

This doesn't work in C, but in languages with stricter type systems, refinement types are a huge advantage.

[–] [email protected] 1 points 3 months ago (1 children)

Wouldn't it still crash when the smart constructor was called?

[–] [email protected] 3 points 3 months ago* (last edited 3 months ago) (1 children)

If it were poorly designed and used exceptions, yes. The correct way to design smart constructors is to not actually use a constructor directly but instead use a static method that forces the caller to handle both cases (or explicitly ignore the failure case). The static method would have a return type that either indicates "success and here's the refined type" or "error and this is why."

In Rust terminology, that would be a Result<T, Error>.

For Go, it would be (*RefinedType, error) (where dereferencing the first value without checking it would be at your own peril).

C++ would look similar to Rust, but it doesn't come as part of the standard library last I checked.

C doesn't have the language-level features to be able to do this. You can't make a refined type that's accessible as a type while also making it impossible to construct arbitrarily.

[–] [email protected] 1 points 3 months ago (1 children)
[–] [email protected] 1 points 3 months ago* (last edited 3 months ago) (1 children)

You're going to need to cite that.

I'm not familiar with C23 or many of the compiler-specific extensions, but in all the previous versions I worked with, there is no type visibility other than "fully exposed" or opaque and dangerous (void*).

You could try wrapping your Foo in

typedef struct {
    Foo validated
} ValidFoo;

But nothing stops someone from being an idiot about it and constructing it by hand:

ValidFoo trustMeBro;
trustMeBro.validated = someFoo;
otherFunction(trustMeBro);

Or even just casting it.

Foo* someFoo;
otherFunction((ValidFoo*) someFoo);
[–] [email protected] 1 points 3 months ago (1 children)

Yes, this is like not checking an error code.

[–] [email protected] 2 points 3 months ago

That's not the point, though. The point is to use a nominal type that asserts an invariant and make it impossible to create an instance of said type which violates the invariant.

Both validation functions and refinement types put the onus on the caller to ensure they're not passing invalid data around, but only refinement types can guarantee it. Humans are fallible, and it's easy to accidentally forget to put a check_if_valid() function somewhere or assume that some function earlier in the call stack did it for you.

With smart constructors and refinement types, the developer literally can't pass an unvalidated type downstream by accident.