Thorsten Leemhuis (acct. 1/4)

""We recently ran across a strange higgs-bugson that manifested itself in a critical system […] A higgs-bugson is a bug that is reported in practice but difficult to reproduce, […]

The Linux kernel does not actually implement this cache as suggested by the RFC, so I wrote a kernel patch to add this functionality and mailed it off upstream. I also learned that the FreeBSD kernel actually already implements this, so this is new-to-Linux but not new-to-NFS.

[…]

Both of these patches are now upstream and will be available in Linux 6.16.""

https://blog.janestreet.com/a-higgs-bugson-in-the-linux-kernel/

#Linux #LinuxKernel #kernel #nfs

A Higgs-bugson in the Linux Kernel

We recently ran across a strange higgs-bugson that manifested itself in a critical system that stores and distributes the firm’s trading activity data, calle...

Jane Street Blog
Jul 3 at 3:36amWeb
Show threadJohannes BechbergerJul 3
@kernellogger Jake Hillion also found such a sporadic bug recently using a chaotic scheduler: https://blog.hillion.co.uk/posts/kernel-driver-scheduling-bug-with-chaos/
Fixing a Kernel Driver Scheduling Bug with scx_chaos

TL:DR; after deploying a sched_ext scheduler to many machines I received reports that a piece of software which measures the turbo frequency of machines was occasionally failing. Reviewing the kernel code we found the issues, but needed to reproduce for tesing. scx_chaos, a sched_ext scheduler designed for testing processes under weird scheduling behaviour, facilitated straightforward reproduction and the patches have been sent upstream. It’s been a while since I last posted on this blog, but I plan to start sharing some publications and interesting problems I come across in my work that can be shared publicly. Here we have an example debugging a kernel problem using my new tool, scx_chaos - enjoy!

Jake Hillion
Show threadJohannes BechbergerJul 3

@kernellogger The chaos scheduler has first been implemented in Java: https://mostlynerdless.de/blog/2025/02/25/helle-ebpf-concurrency-testing-using-custom-linux-schedulers-19/

I hope scx_chaos can help many other people find hard-to-detect bugs

Hello eBPF: Concurrency Testing using Custom Linux Schedulers (19) - Mostly nerdless

Learn how to write a custom Linux scheduler for concurrency fuzzing with sched-ext and hello-ebpf in this week's blog post.

Mostly nerdless
Show threadJames Just JamesJul 3
@kernellogger Why are they still using NFSv3 and I hope this is also fixed in NFSv4!