Mastodawn

LisPi Nov 3, 2023

@lanodan @cwebber Did you know that #OpenGenera had a #C and an #Ada compiler (among others)?

You don't need to rewrite the drivers, you can just implement a shim (with error handling) so they can run on top of the system in their little isolated bubble like the #rumpkernel idea was about.

You do need those bits in some specific areas, but you can tightly constrain access to them to very specific modules & threads.

Erlang's message-passing approach isn't required for condition handling.

blake shaw 🇵🇸Sep 22, 2023

The Rise and Fall of the Operating System

"We ended up building a #unikernel based on the drivers offered by rump kernels via rumpkernel.org: Rumprun. We were not trying to build an #OS-like layer but one day simply realized that we could build one which would just work, with minimal effort. The noteworthiness of the Rumprun unikernel does not come from the fact that existing software such as Nginx, PHP, and mpg123 can be cross-compiled in the normal fashion and then run directly on the cloud or on bare metal. The noteworthiness comes from the fact that the implementation is a few thousand lines of code ... plus drivers. The ratio of drivers to “operating system” is on the order of 100:1, so there is very little operating system in there. The #Rumprun implementation is that of an orchestrating system, which conducts the drivers."

#bsd #rumpkernel
http://www.fixup.fi/misc/usenix-login-2015/login_oct15_02_kantee.pdf

blake shaw 🇵🇸May 19, 2023

A practical solution for #Hurd's lack of drivers: #NetBSD's #rumpkernel framework at #fosdem:
https://archive.fosdem.org/2022/schedule/event/dzammit/

found this via @janneke's recent patch series, working to bring it to #guix:
https://lists.gnu.org/archive/html/guix-patches/2023-05/msg00633.html

FOSDEM 2022 - A practical solution for GNU/Hurd's lack of drivers: NetBSD's rumpkernel framework

Show thread

lbll Dec 5, 2022

@picasoft

Avez-vous entendu parler du framework Genode OS qui, par une approche sécurité (on spécialise l'OS en n'utilisant que le code utile à ce que l'on veut faire) réduit drastiquement la surface de code exécutée au runtime ce qui laisse supposer de meilleures performances et un code optimisé (à l'inverse du bloatware où on empile les stacks)
#microkernel #rumpkernel