After collecting some experience in kernel development with my own hobby OS, I figured I could look into writing a patch for Linux. My only experience for writing code for other more „professional“ operating systems is a minuscule patch to OpenBSD and it didn't even got accepted. OpenBSD or NetBSD would probably be preferable for writing little patches, since there are still plenty of not-so-difficult things left to be done, but I run Linux on my PC and I don't want to write code for an OS that I use in a virtual machine only. I do have respect for Linux, because that is a whole other world. Have any of you ever written a patch for Linux? I'm aware of the basic advice of looking first with fuzzers, but I would prefer to just read code. Is there any subsystem that is particularly newbie friendly? USB sounded interesting for me, so I read all the recent posts in that mailing list, yet it does seem as difficult as every other subsystem, very different from what I did with my hobby OS. I want to hear all your experiences and what helped you find things in the kernel to write patches for (except from reading code, the mailing list, documentation, and fuzzer results). I'm not looking for an „easy trick to avoid the hard work“, more just experienced from kernel developers who knows their shit, and can give some advice.
>>110185214Can you find an annoying bug and fix itIdeally one that annoys you Worst case, get an LLM to help you find something that might be low-hanging fruitWhether you use it to help you fix the bug too is another question entirely
>>110185567a lot of the LKML is random dudes using LLMs on kernel code to submit typo corrections
>>110186594Is there any OS where it's worth putting effort into learning the internals to contribute to it, then? I want to believe that there are still people in the LKML who care about contributing to the kernel and like putting effort into what they do.
what feature is the kernel even missing, what would you work on?
>>110189084>Is there any OS where it's worth putting effort into learning the internals to contribute to it, then?If you use linux and you find something that fixes or improves things that help you out / you care about I dont see why not> I want to believe that there are still people in the LKML who care about contributing to the kernel and like putting effort into what they do.there are plenty of those too
>>110189424you'd be surprised how much of the kernel code, especially drivers are ancient
>>110192951Well, 99% of the kernel is drivers. Freetards start screeching when legacy archaic shit gets dropped support for their shitboxes in new kernel releases. Maybe monolithic was a mistake after, kernel updates weight lile 300 megs these days.
