Remove?

Do you support this proposal?

  • Total voters
    5
  • This poll will close: .
D

Dinghy0878

Motivation:
The BSDs have the hard life trying to get drivers for current, past and future devices, meanwhile Linux get a relatively easy life, being the first class citizens of drivers in certain domains like embedded computing, and having a tonnage of investment to get those drivers, for devices where it is not getting first class citizen treatments.

But not only the BSDs have the hard life, but every new project of developing an operating system, with the driver problem being the number one cause of the impossibility of the innovation of new operating systems.

The Plan:
The proposed solution is creating an VM for driver writing at the style portable eBPF, and then get it implemented in every BSD, and Linux.

What do the FreeBSD project think of this project?
 
I obviously voted no, since the idea runs into all manner of brick walls. To begin with, you have to rewrite ALL kernels involved. Which would cause licenses and heads to explode.

Here is a much simpler starting point: Using whatever VM technology is readily accessible, run multiple OSes on the machine. Put all the hardware interfacing (be it GUI = X-windows = Wayland, be it other hardware drivers) into one Linux VM, and run the OS of your desire (that is, FreeBSD) into another VM. We can argue which VM should be the host, if any, let's not argue. Then on the FreeBSD machine, redirect IO (using the very fast internal bridge network) to the Linux VM. For X-windows this is trivial, just "setenv DISPLAY=myhost_linux.example.com"; for other protocols, one would have to develop thin shims that run on both sides.
 
You must log in or register to reply here.
Back
Top