What does HackerNews think of MIRAI?

Rust mid-level IR Abstract Interpreter

Language: Rust

Here's a 2020 overview of Rust verification tools https://alastairreid.github.io/rust-verification-tools/ - it says

> Auto-active verification tools

> While automatic tools focus on things not going wrong, auto-active verification tools help you verify some key properties of your code: data structure invariants, the results of functions, etc. The price that you pay for this extra power is that you may have to assist the tool by adding function contracts (pre/post-conditions for functions), loop invariants, type invariants, etc. to your code.

> The only auto-active verification tool that I am aware of is Prusti. Prusti is a really interesting tool because it exploits Rust’s unusual type system to help it verify code. Also Prusti has the slickest user interface: a VSCode extension that checks your code as you type it!

> https://marketplace.visualstudio.com/items?itemName=viper-ad...

Now, on that list, there is also https://github.com/facebookexperimental/MIRAI that, alongside the crate https://crates.io/crates/contracts (with the mirai_assertion feature enabled) enables writing code like this

    #[ensures(person_name.is_some() -> ret.contains(person_name.unwrap()))]
    fn geeting(person_name: Option<&str>) -> String {
        let mut s = String::from("Hello");
        if let Some(name) = person_name {
            s.push(' ');
            s.push_str(name);
        }
        s.push('!');
        s
    }
And have it checked at compile time that the assertion holds! Which is a bit like Liquid Haskell in capability: https://ucsd-progsys.github.io/liquidhaskell/

... and now I just noticed that prusti has a crate prusti_contracts that can do the same thing!! https://github.com/viperproject/prusti-dev/blob/master/prust...

Now I'm wondering which tool is more capable (as I understand, they leverage a SMT solver like Z3 to discharge the proof obligations, right?)

Traditional design by contract checks the contracts at runtime. They can be understood as a form of dynamic typing with quite complicated types, which may be equivalent to refinement types

But you can check contracts at compile time too. It's quite the same thing as static typing with something like refinement types. That's because, while with contracts we can add preconditions like "the size of this array passed as parameter must be a prime number", with refinement types we can define the type of arrays whose size is a prime number, and then have this type as the function argument. (likewise, postconditions can be modeled by the return type of the function)

See for example this Rust library: https://docs.rs/contracts/latest/contracts/

It will by default check the contracts at runtime, but has an option to check them at compile time with https://github.com/facebookexperimental/MIRAI

Now, this Rust library isn't generally understood as creating another type system on top of Rust, but we could do the legwork to develop a type theory that models how it works, and show the equivalence.

Or, another example, Liquid Haskell: https://ucsd-progsys.github.io/liquidhaskell/ it implements a variant of refinement types called liquid types, which is essentially design by contract checked at compile type. In this case, the type theory is already developed. I expect Liquid Haskell to be roughly comparable to Rust's contracts checked by MIRAI.

Now, what we could perhaps say is that refinement types are so powerful that they don't feel like regular types! And, while that's true, there are type systems even more powerful: dependent types used in languages like Coq, Lean and F* to prove mathematical theorems (your type is a theorem, and your code, if it typechecks, is a proof of that theorem).

Dependent types were leveraged to create a verified TLS implementation that mathematically proves the absence of large class of bugs, miTLS https://www.mitls.org/ (they discovered a number of vulnerabilities in TLS implementations and proved that their implementation isn't vulnerable), and HACL* https://github.com/hacl-star/hacl-star a verified crypto implementation used by Firefox and Wireguard. They are part of Project Everest https://project-everest.github.io/ which aims to develop provably secure communications software.

Because it's convenient and familiar to most programmers. Not providing bounds-checked indexing makes some kinds of code very hard to write.

But note his problem also happens with integer division.

In Rust, a[x] on an array or vec is really a roughly a shortand for a.get(x).unwrap() (with a different error message)

Likewise, a / b on integers is a kind of a shortand for a.checked_div(b).unwrap()

The thing is, if the index ever is out of bounds, or if the denominator is zero, the program has a bug, 100% of time. And if you catch a bug using an assertion there is seldom anything better than interrupting the execution (the only thing I can think of is restarting the program or the subsystem). If you continue execution past a programming error, you may sometimes corrupt data structures or introduce bizarre, hard to debug situations.

Doing a pattern match on a.get(x) doesn't help because if it's ever None (and your program logic expects that x is in bounds) then you are kind of forced to bail.

The downside here is that we aren't catching this bug at compile time. And it's true that sometimes we can rewrite the program to not have an indexing operation, usually using iterators (eliding the bounds check will make the program run faster, too). But in general this is not possible, at least not without bringing formal methods. But that's what tests are for, to ensure the correctness of stuff type errors can't catch.

Now, there are some crates like https://github.com/dtolnay/no-panic or https://github.com/facebookexperimental/MIRAI that will check that your code is panic free. The first one is based on the fact that llvm optimizations can often remove dead code and thus remove the panic from a[x] or a / b - if it doesn't, then compilation fails. The second one employs formal methods to mathematically prove that there is no panic. I guess those techniques will eventually be ported to the kernel even if panics happen differently there (by hooking on the BUG mechanism or whatever)

Nice, I just would have liked to get all these different verification tools combined under the same interface, just being different backends as drafted by the rust verification tools work of project oak: have "cargo verify" as common command and use common test annotations, allowing the same test to be verified with different backends or just fuzzed/proptested (see https://project-oak.github.io/rust-verification-tools/using-... and https://project-oak.github.io/rust-verification-tools/using-...).

The model checking approach seems to be a bit limited regarding loops. There are also abstract interpreters, such as https://github.com/facebookexperimental/MIRAI, and symbolic executers, such as https://github.com/dwrensha/seer or https://github.com/GaloisInc/crucible.

Overall I believe this space would benefit from more coordination and focus on developing something that has the theoretical foundations to cover as many needs as possible and then make a user-friendly tool out of it that is endorsed by the Rust project similar to how Rust analyzer is the one language server to come.

also what https://github.com/facebookexperimental/MIRAI does for rust: you add contracts (preconditions, postconditions, and invariants) using https://docs.rs/contracts/0.6.0/contracts/ and it checks whether they are true

here is an example using mirai and contracts:

https://github.com/facebookexperimental/MIRAI/blob/master/ex...

(note: the contracts crate can be used for verifying at runtime as well)