Announcement

Collapse
No announcement yet.

Linux 5.15-rc2 Released With Many Fixes, Addressing Issues Raised By "-Werror"

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Linux 5.15-rc2 Released With Many Fixes, Addressing Issues Raised By "-Werror"

    Phoronix: Linux 5.15-rc2 Released With Many Fixes, Addressing Issues Raised By "-Werror"

    Linux 5.15-rc2 is now available as the latest weekly release candidate for this next version of the Linux kernel. Linux 5.15 in turn should be out as stable around the start of November...

    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

  • #2
    Code:
    $ uname -a
    Linux ubuntu 5.15-rc2-051500rc2+customidle-generic #051500rc2+202107311230 SMP PREEMPT Sun Sep 19 19:21:58 PDT 2021 x86_64 x86_64 x86_64 GNU/Linux
    Up and running if anyone needs. See here if you're interested. Ubuntu/Debian mostly, but something for everyone. Cheers.

    --

    edit: Also, if anyone is interested, I just ran some quick benchmarks of pts/osbench comparing it to 5.4.147, across different GCC versions. (but only 9.3.0 for 5.15-rc2 as it's the fastest for my Haswell desktop. See results here: https://openbenchmarking.org/result/...TJ-OSBENCHSE60

    too lazy; didn't click

    Code:
    Geometric Mean Of All Test Results
    Result Composite - osbench-sept-2021
    Geometric Mean > Higher Is Better
    
    5.15-rc2 gcc-9.3.0-17ubuntu120.04 . 4.574 |=================================================  ========================
    5.4.147 gcc-7.5.0-6ubuntu2 ........ 4.615 |=================================================  =========================
    5.4.147 gcc-11.1.0-1ubuntu120.04 .. 4.632 |=================================================  ==========================
    5.4.147 gcc-8.4.0-3ubuntu2 ........ 4.653 |=================================================  ==========================
    5.4.147 gcc-10.3.0-1ubuntu120.04 .. 4.676 |=================================================  ===========================
    5.4.147 gcc-9.3.0-17ubuntu120.04 .. 4.715 |=================================================  ============================
    Disclaimer: I also want to say that this isn't nearly an "apples to apples" comparison because I have different patches for both (but very similar across kernel versions). Please take with extreme grain of salt and lean on Michael's benchmarks. These are just my personal tests and benchmarks. Nothing more, nothing less. But still a general idea of where things are I think.

    Actually, compiling the latest 5.15-rc2 with the latest GCC would be a better idea than compiling it with 9.3.0. Might do that at some point. But don't care enough right now.

    This one is interesting. 5.15-rc2 much slower at creating processes. Creating threads seems to be fine.

    Last edited by perpetually high; 19 September 2021, 11:50 PM.

    Comment


    • #3
      We have time tonight. I take back what I said above and I'm compiling 5.15-rc2 with exact same settings but latest gcc-11 instead of gcc-9. Will be interesting to see the results. Will report back once it's done.

      edit: Might as well make this post useful since I know you people are lazy and go straight to the comments and don't read a) Michael's great posts, b) Linus's announcement c) my posts on every thread. Do way better, folks. /endedit

      edit 2: Here are the results. gcc-11 performed same/slightly worse than gcc-9 for 5.15-rc2 on my machine. See here: https://openbenchmarking.org/result/...TJ-OSBENCHSE27

      too lazy; didn't click

      Code:
      Geometric Mean Of All Test Results
      Result Composite - osbench-sept-2021
      Geometric Mean > Higher Is Better
      
      5.15-rc2 gcc-11.1.0-1ubuntu120.04 . 4.557 |================================================= ========================
      5.15-rc2 gcc-9.3.0-17ubuntu120.04 . 4.574 |================================================= ========================
      5.4.147 gcc-7.5.0-6ubuntu2 ........ 4.615 |================================================= =========================
      5.4.147 gcc-11.1.0-1ubuntu120.04 .. 4.632 |================================================= ==========================
      5.4.147 gcc-8.4.0-3ubuntu2 ........ 4.653 |================================================= ==========================
      5.4.147 gcc-10.3.0-1ubuntu120.04 .. 4.676 |================================================= ===========================
      5.4.147 gcc-9.3.0-17ubuntu120.04 .. 4.715 |================================================= ============================
      Linux 5.15-rc2
      2021-09-20 0:40 Linus Torvalds
      From: Linus Torvalds @ 2021-09-20 0:40 UTC (permalink / raw)
      To: Linux Kernel Mailing List

      So I've spent a fair amount of this week trying to sort out all the
      odd warnings, and I want to particularly thank Guenter Roeck for his
      work on tracking where the build failures due to -Werror come from.

      Is it done? No. But on the whole I'm feeling fairly good about this
      all, even if it has meant that I've been looking at some really odd
      and grotty code. Who knew I'd still worry about some odd EISA driver
      on alpha, after all these years? A slight change of pace

      The most annoying thing is probably the "fix one odd corner case,
      three others rear their ugly heads". But I remain convinced that it's
      all for a good cause, and that we really do want to have a clean build
      even for the crazy odd cases.

      We'll get there.

      Anyway, I hope this release will turn more normal soon - but the rc2
      week tends to be fairly quiet for me, so the fact that I then ended up
      looking at reports of odd warnings-turned-errors this week wasn't too
      bad.

      There's obviously other fixes in here too, only a small subset of the
      shortlog below is due to the warning fixes, even if that's what I've
      personally been most involved with.

      Go test, and keep the reports coming,

      Linus
      Last edited by perpetually high; 20 September 2021, 12:21 AM.

      Comment


      • #4
        I'm one of the ones cheering this on. I'm not a direct NetBSD user. I use FreeBSD for my file & local DNS server, so there's plenty of BSD code swapping going on. However, I remember NetBSD did something similar to what Linus is talking about some years ago. It apparently helped clean up a lot of bugs and odd behavior in the base system.

        Comment


        • #5
          Nice to see so many warnings melt away. In my opinion, warnings should be regarded as code errors the compiler had sufficient information to patch itself. It is an error in your code/design, you just don't see it because you had a helpful gnome fixing the problem for you.

          There are exceptions to every rule, of course, but warnings should be treated as errors in almost every case.

          Comment


          • #6
            Originally posted by wertigon View Post
            Nice to see so many warnings melt away. In my opinion, warnings should be regarded as code errors the compiler had sufficient information to patch itself. It is an error in your code/design, you just don't see it because you had a helpful gnome fixing the problem for you.

            There are exceptions to every rule, of course, but warnings should be treated as errors in almost every case.
            That's bollocks.

            Practically no major software project compiles without warnings.

            Firefox, Node, Rust, Libreoffice, Chromium, LLVM, FFMPEG, Gnome, KDE, Wayland etc etc all generate a shit ton of warnings when built. And even the much hated Xorg + Xserver.

            Go right ahead and watch the Linux desktop crumble if -Werror is enforced on such projects. I know I will love to see that.

            Comment


            • #7
              Originally posted by Sonadow View Post
              Practically no major software project compiles without warnings.
              We enforce a strict no-new-warning policy at work in the shared code base.
              According to cloc, we currently have > 1000000 lines of C and C++ code.
              Merge requests cannot get merged if the warning count is increased.
              On the supported toolchains, no warnings are generated.
              Still, we don't enforce -Werror typically.

              Comment


              • #8
                Originally posted by Sonadow View Post
                That's bollocks.
                No, it is not. Nonsense is only the practise of turning on warnings to then ignore them. You do not want warnings? Then do not turn them on. This is really just a case of a bad habit having been sustained for far too long.
                Last edited by sdack; 20 September 2021, 05:53 AM.

                Comment


                • #9
                  Been using 5.15-rc2, and wow is my system flying. It's really a combination of a ton of things, but absolutely *nobody* can tell me it's not the year of the Linux Desktop. I'm witnessing it. If not, you just don't know what you're doing. Sorry, harsh truth. Cheers.

                  edit: Seriously, look at this. Don't tell me Linux Desktop isn't here. It's been here.



                  Click it, it's worth it.
                  Last edited by perpetually high; 20 September 2021, 06:17 AM.

                  Comment


                  • #10
                    Hardlocks my Nvidia card when starting up KMS. Also happened with 5.14 RC2 and RC3.

                    As always, the whole goddamned kernel has no guarantee or assurance that it will work with every new release or update. Useless garbage.

                    Linux desktop? Garbage since 2003, and will forever be garbage. All that talk about Linux supporting more hardware than Windows is yet another steaming pile of lies spread by linuxfags.

                    In Windows, literally everything just works perfectly. Heck, my Nvidia card can do native resolution in Windows even without the Nvidia drivers installed. Linux can't even boot that same card to a damned TTY with KMS on the latest rc kernel.
                    Last edited by Sonadow; 20 September 2021, 06:49 AM.

                    Comment

                    Working...
                    X