Linux 6.14 With Rust: "We Are Almost At The 'Write A Real Driver In Rust' Stage Now"

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • phoronix
    Administrator
    • Jan 2007
    • 67377

    Linux 6.14 With Rust: "We Are Almost At The 'Write A Real Driver In Rust' Stage Now"

    Phoronix: Linux 6.14 With Rust: "We Are Almost At The 'Write A Real Driver In Rust' Stage Now"

    Greg Kroah-Hartman today sent out the pull request of all the driver core updates for Linux 6.14, which ends up being a big deal for those interested in the prospects of Rust drivers for the Linux kernel...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite
  • bosslog
    Junior Member
    • May 2020
    • 18

    #2
    Memory safety is important, but it's very unfortunate that the way many have chosen to go about it is through the use of Rust.

    Comment

    • Quackdoc
      Senior Member
      • Oct 2020
      • 5089

      #3
      bcachefs in rust when

      Comment

      • ayumu
        Senior Member
        • Oct 2008
        • 668

        #4
        It'd be nice if these rust devs worked on redox instead.

        They'd certainly get a lot more productivity per invested time.

        Comment

        • ZenoArrow
          Junior Member
          • Nov 2018
          • 35

          #5
          Originally posted by bosslog View Post
          Memory safety is important, but it's very unfortunate that the way many have chosen to go about it is through the use of Rust.
          Why is it unfortunate to use Rust for this? You expect to do the same in C?

          Comment

          • mdedetrich
            Senior Member
            • Nov 2019
            • 2540

            #6
            Originally posted by bosslog View Post
            Memory safety is important, but it's very unfortunate that the way many have chosen to go about it is through the use of Rust.
            How did you expect it to be done, by magic?

            Comment

            • fotomar
              Phoronix Member
              • Jun 2024
              • 96

              #7
              Originally posted by ayumu View Post
              It'd be nice if these rust devs worked on redox instead.

              They'd certainly get a lot more productivity per invested time.
              absolutely

              Comment

              • TheMightyBuzzard
                Senior Member
                • Sep 2021
                • 416

                #8
                Originally posted by mdedetrich View Post

                How did you expect it to be done, by magic?
                Maybe the old fashioned way? You know, by learning how to code at least passably before you go trying to get your awesome O(n^n^n) code into the kernel? Memory safety isn't the only reason to not want noobs wanking all up in the kernel.

                Comment

                • ZenoArrow
                  Junior Member
                  • Nov 2018
                  • 35

                  #9
                  Originally posted by TheMightyBuzzard View Post

                  Maybe the old fashioned way? You know, by learning how to code at least passably before you go trying to get your awesome O(n^n^n) code into the kernel? Memory safety isn't the only reason to not want noobs wanking all up in the kernel.
                  So your main complaint is algorithmic complexity / code speed? Oh and regarding "learning how to code", the reason that memory safety is harder in C is because it permits you to do things in the wrong way, if you ended up with "safe" C code it would increase in complexity too. Are you going to argue that the kernel devs using C don't know how to write safe code, or will you accept some languages are better designed for safety?

                  Comment

                  • CommunityMember
                    Senior Member
                    • Oct 2019
                    • 1379

                    #10
                    Originally posted by TheMightyBuzzard View Post
                    Maybe the old fashioned way?
                    While the few remaining graybeards might agree, it has been shown to not work if we also want progress and features rather than stagnation and stability. We have a long set of examples (including in the Linux kernel) that mistakes will be made, and not be caught until the code makes it into the wild.

                    Comment

                    Working...
                    X