Rust Null Block Driver Published To Begin Experimenting With Rust For Linux Storage

[chocolate]

By the way, there is an interesting presentation on it here:

https://ntrs.nasa.gov/api/citations/...S_Tutorial.pdf

Oh my, thank you so much for this. I'm in heaven for the slide about multiple return statements being better for static analyzers vs. mutable variable & one return. I was sure of it, but I hadn't read anything about it until now. I've lost count of the times I had to battle with software people wearing MISRA blinders over this matter. I'll be smiling for days just for this.

[marios]

Yes because its a stupid idea, if someone wants to disable bound checking they should be explicit about it. The last thing we need is distro maintainers/programmers thinking they are "smart" by disabling bounds checking using some compiler flag to then proceed creating security issues for upstream.



Understand what, the fact that you can write unsafe C code?

From now on, I will never again answer to you, unless you quote me. Just ignore me and I will also ignore your pitiful existence. I think this will better for both of us.

[oleid]

My replies here had been mostly because I disliked the misinformation spread by some Rust enthusiasts that rewriting code in Rust automatically produces a better result.

Why to you keep putting things into other people's mouth?

[oleid]

Interestingly, the Rust support in Linux might as well not be there since you can compile Linux 6.3.1 on a system that lacks a Rust compiler and there is no real difference. The rust work happening right now is intended to enable people to write new drivers in it, but it really is not suitable for that yet since it lacks support for all architectures and it does not even have a minimum compiler version.

People already write graphics drivers for Apple M1 (and later) in rust. It is here to stay, I'm positive.

https://twitter.com/LinaAsahi/status/1577667445719912450

[ryao]

Why to you keep putting things into other people's mouth?

mdedetrich is a huge proponent of rewriting all C/C++ code into Rust or any other language. He has made comments to this effect many times in the past. If he were not such a zealot, I would not have had a reason to reply.

Then you have others making remarks like “Our goal should be to get unsafe code so small such that considerable effort can be spent on verifying its safety” with regard to Linux rust support, which is very different than what Linus actually agreed to do. He did not agree to a rewrite in Rust, but Rust proponents here will happily insinuate he did when he is not around to say otherwise.

[ryao]

People already write graphics drivers for Apple M1 (and later) in rust. It is here to stay, I'm positive.

https://twitter.com/LinaAsahi/status/1577667445719912450

None of that is in Linus’ tree at the moment. It also appears that she wrote kernel driver code for the M1 in both C and Rust in her private tree:

https://github.com/AsahiLinux/linux/tree/asahi/drivers/gpu/drm/Apple

https://github.com/AsahiLinux/linux/tree/asahi/drivers/gpu/drm/asahi

I am not sure how this is relevant. Whether Linux’s Rust experiment succeeds is far too early to call.

[oleid]

None of that is in Linus’ tree at the moment. It also appears that she wrote kernel driver code for the M1 in both C and Rust in her private tree:

AFAIR it is in the review process. As for the C code best ask Lina. To me it looks like some glue code and support for display backlight.

I am not sure how this is relevant. Whether Linux’s Rust experiment succeeds is far too early to call.

It's relevant as it proves it works fine for something as complicated as an drm kernel driver and rust support will be kept around for anything that wants to support recent Apple hardware.

[oleid]

mdedetrich is a huge proponent of rewriting all C/C++ code into Rust or any other language. He has made comments to this effect many times in the past.

Maybe so, but I did not read such a comment in this thread. Or you interpret things differently than I do.

[wswartzendruber]

Then you have others making remarks like “Our goal should be to get unsafe code so small such that considerable effort can be spent on verifying its safety” with regard to Linux rust support, which is very different than what Linus actually agreed to do. He did not agree to a rewrite in Rust, but Rust proponents here will happily insinuate he did when he is not around to say otherwise.

Well, that's a fair response. I was half thinking of Windows when I typed that out.

I do believe that Linux will ultimately follow suit, though. If not with Rust, than with something else.

[mdedetrich]

Maybe so, but I did not read such a comment in this thread. Or you interpret things differently than I do.

No you are correct, he is imagining things/putting words into other peoples mouths as usual.