3K Lines Of New Rust Infrastructure Code Head Into Linux 6.13

[phoronix]

Phoronix: 3K Lines Of New Rust Infrastructure Code Head Into Linux 6.13

Overnight the Rust for Linux lead developer Miguel Ojeda submitted the big set of Rust infrastructure/toolchain updates for the Linux 6.13 holiday kernel...

https://www.phoronix.com/news/Linux-6.13-Rust

[MillionToOne]

This is a pointless article IMO. Just made to advertise Rust. But at least we got another monthly article where we can argue about which language is better

[Vorpal]

Could we just have a civilised discussion? For once? Please?

[Volta]

This is a pointless article IMO. Just made to advertise Rust. But at least we got another monthly article where we can argue about which language is better

Not only this. LWN became rust propagandistic tube. I bet rust woke foundation is paying them.

[Dirlridus]

This is a pointless article IMO. Just made to advertise Rust. But at least we got another monthly article where we can argue about which language is better

Development news about the kernel this site is centered around? Exciting progress for developers? Wait, my bad, Rust is mentioned so it must be propaganda and advertisement.

[uid313]

Which part or subsystem of the Linux kernel would like the most to see adopt Rust?

Is it graphics device drivers? or the network stack? or is it file systems? ext4 or Btrfs or a brand-new one? Perhaps the CPU scheduler or the I/O scheduler?

[Gusar]

This is a pointless article IMO. Just made to advertise Rust. But at least we got another monthly article where we can argue about which language is better

Phoronix reports on many kernel subsystem additions every cycle - input, graphics, filesystems, etc, etc. Are all those articles pointless too?

[edxposed]

Hopefully there won't be any really valuable rust-only features until rust is actually fully stabilized. The current rust is a toy, you can't even build it with the latest or slightly older rust toolchain, and it even restricts C toolchain versions (LTO/CFI).

[intelfx]

you can't even build it with the latest or slightly older rust toolchain

Let's check the facts, shall we?

$ /usr/bin/rustc --version
rustc 1.82.0 (f6e511eec 2024-10-15) (Arch Linux rust 1:1.82.0-3)

$ CC=/usr/bin/clang BINDGEN=/usr/bin/bindgen RUSTC=/usr/bin/rustc scripts/rust_is_available.sh && echo OK
OK

$ make LLVM=1 rustavailable
Rust is available!​

https://github.com/torvalds/linux/bl...ion.sh#L35-L37

and it even restricts C toolchain versions (LTO/CFI)

Yeah, big surprise, for IR-level transformations in a multi-language project both language toolchains must be IR-compatible. Shocking, I tell you!

[Old Grouch]

One of the challenges with Rust, as I see it, is that there is currently no formal specification of the language.

Now, one can say that 'the documentation is the code', and ultimately, that is the truth, but it is a little difficult to work with. Formal specifications matter if you are trying to validate multiple compilers, or determine if behaviour is correctly implemented.

It also increases the challenge of attempting proofs of formal correctness of code.

Linus obviously thinks that it is worthwhile to make it possible to do kernel programming in anything other than C or assembly language, despite the problems of experienced C programmers not wanting to take up the integration/maintenance burden of code written in Rust. I think he has earned the right for his opinions to be taken seriously, and he knows more about kernel programming and maintenance than the average commentator.

All the problems of integrating Rust-sourced code into the kernel will happen with any new language, so if nothing else, this is necessary preparatory work for other languages, which might or might not be invented yet, whether Rust succeeds or not. If nothing else, it will test the ability of people programming comfortably in one language to communicate and cooperate well with people working in another. I suspect this will become more necessary in the future of kernel development, even if not in the near future, depending on the outcome of this Rust 'experiment'.