1. 15

  2. 4

    This is a good write-up of the different modular kernel projects being developed. I’ll definitely be taking a look a few of them. I suppose I should start with looking at Fuschia. It seems like the most likely candidate to have a more mainstream appearance in the “near” future.

    I was hoping for more of a what the era of a modular OS kernel would look like and how things are changing to push OS development in that direction, though.

    1. 4

      Cool to see Genode mentioned in the article!

      Related to eBPF, I believe NetBSD has a full Lua VM available in kernel space since 2013 or so? Though I can’t seem to be able to google up any articles exploring using it to actually write something :(

      Also, IIUC, the “rump kernel” feature added to NetBSD at some point in the past, makes it already somewhat more modular than most other monolithic kernels; to the extent of apparently making it easy enough to reuse some parts of the codebase in other OS projects (e.g. Haiku IIRC …oops, it seems Haiku reuses FreeBSD drivers, actually… however, it seems Genode does reuse NetBSD filesystem drivers).

      1. 2

        NetBSD is popular for this. QNX re-used their code if I recall correctly. Also cost them at least one vulnerability because of mismatch between their model and the model NetBSD’s code was designed for. It’s those subtle integration bugs that get ya every time you try to cheat by bringing in foreign code.