Announcement

Collapse
No announcement yet.

Benchmarking The Performance Overhead To LKRG 0.8 For Better Security

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

  • Benchmarking The Performance Overhead To LKRG 0.8 For Better Security

    Phoronix: Benchmarking The Performance Overhead To LKRG 0.8 For Better Security

    Back in March I benchmarked the Linux Kernel Runtime Guard (LKRG) as a means of achieving additional security safeguards for a ~5% performance hit. With LKRG 0.8 having been released a few days ago, here is a fresh look at the LKRG performance compared to the stock kernel on Ubuntu 20.04 LTS.

    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
    How can LKRG run faster for gaming in Xonotic?! o-o

    It appears that syscall-heavy apps are the most affected...

    Comment


    • #3
      Originally posted by tildearrow View Post
      How can LKRG run faster for gaming in Xonotic?! o-o
      maybe it is outside the sensitivity of the test?

      But this seems like they have done some awesome work, I wonder if we will start seeing distro's pick it up for their desktops

      Comment


      • #4
        Wow! A 52% performance reduction on launching programs? That seems like it'd really hit performance of shell scripts and similar workloads. The hit on file creation also looks pretty painful.

        This actually seems like pretty bad news for a lot of server workloads.

        Comment


        • #5
          If you run Arch see arch-sign-modules - I added support today to sign the p_lkrg kernel module (in linux-hardened / linux-zen / linux-lts)

          Comment


          • #6
            Originally posted by coder View Post
            Wow! A 52% performance reduction on launching programs? That seems like it'd really hit performance of shell scripts and similar workloads. The hit on file creation also looks pretty painful.

            This actually seems like pretty bad news for a lot of server workloads.
            In the test, the launched program simply exits and does nothing else. If the program did actual work, the cumulative performance hit might be significantly lower. YMMV, of course.

            Details here: https://github.com/mbitsnbites/osbench

            Comment

            Working...
            X